Subscribe now

Systems of Record Won the SaaS Era - Clearinghouses Will Win the Agents Era

Back in December I wrote about the fight to become the front door to the systems of record. In that post I wrote about distribution, who sits between the user and the data (and why sitting there is strategic). This post is an expansion of that post (and the 1 or 2 I wrote after about similar topics). What really should AI companies be racing towards? If systems of record won in the SaaS era (ie they had the durable moats), what’s the equivalent in the AI era? Of course the answer is still partially “the system of record”, where maybe you swap out “record” with something like “agents” or “work".” But let’s come up with something new :)

Let’s start by looking at the SaaS era, and what qualities created durable successful companies. In SaaS, one main goal just about every company aspired towards was to become a system of record. Store and govern critical data. Own the place where workflows kicked off (or where workflows touched). If you got there (Salesforce for customer data, Workday for employee data, NetSuite for financial data) you ended up with an incredibly deep moat. Everything integrated with the system of record. And when you had hundreds of independent workflows and solutions built on top of the system of record, even thinking about ripping out that foundational layer was impossible. Ripping out a system of record was so painful that customers would really avoid it at all costs... The change management was almost never worth it. If you think upgrading a piece of software to a new version breaks things, imaging ripping out the layer all your workflows touched.. Price increases, mediocre product velocity, difficult upgrade cycles... didn’t matter. The data was in there, the workflows were built around it, and the switching costs got harder and harder with every year that went by.

So what’s the equivalent prize in AI? I think it’s becoming The Clearinghouse for agents.

In financial markets, the clearinghouse sits between different parties that aren’t able to fully trust each other. The clearinghouse verifies / authorizes / settles trades, and ultimately keeps the receipt. Nobody really loves the clearinghouse, but it’s clear it has to exist for the ecosystem to transact.

Now think about where enterprise software is heading. Agents from tons of different vendors, acting autonomously, touching your most critical data, and even in the future spending real money. Some company has to sit in the middle of all that and decide: which agent is cleared to act? On what data? With what limits? And can you prove what happened after the fact? Whoever holds that seat holds incredibly “strategic real estate.” (and every founder I’ve worked with has probably heard me discuss strategic real estate over and over). That’s the clearinghouse.

This may sound counterintuitive, but owning the clearinghouse for agents (given agent companies themselves will want to be the clearinghouse) may create a deeper moat than the one systems of record had. A system of record controlled your data. It kind of controlled your workflows (but not always, oftentimes someone else controlled the workflows, but the data in the system of record was a critical part of the path). The Clearinghouse controls four things: memory (what your agents know), context (what they see and how it’s served), execution (what they’re allowed to do), and governance (who’s allowed to do what, plus the audit trail behind all of it). If migrating off a system of record was painful, migrating off the thing that holds your policies, your permissions, and your entire audit history is probably harder (especially when the agents start to handle more and more of the work). AND - I think these agent companies that become The Clearinghouse will start to look more and more like systems of record in their own right. Data in systems of record were oftentimes transactional data. Data in agent systems of records (ie Clearinghouses) will be agent traces, agent evals, agent telemetry data, agent A/B data, etc

And governance is obviously a really important part of the story here. Governance used to be more of a compliance checkbox at the end of the sales cycle. It was the thing the security team made you sit through after the deal was basically done and agreed to by the business folks. Now it’s what CIOs will focus on at the forefront of a deal (or at a minimum are thinking about from meeting #1). Once agents act autonomously, the buying question changes. It’s no longer “is the model good?” Every model is good (or good enough). The question becomes “can I see what every agent did, set policy on what it can touch, and prove it to my auditors?” The biggest companies are already showing the way... Databricks leads with evaluation and governance. KPMG just announced they’re wrapping Microsoft’s Agent 365 in their “Trusted AI” framework. These companies are selling clearance. Governance is some very strategic real estate on how you earn the clearinghouse seat (and like systems of record, what makes it nearly impossible to lose once you have it).

This is why everyone is racing. Microsoft (Agent 365 + Copilot woven through Windows and Office), Salesforce, Snowflake, Databricks aren’t fighting over model quality anymore (not that they ever really were…). They all see the same prize: the clearinghouse is the toll booth every agent action passes through.

And lots of people going after this! The data players (Snowflake, Databricks) believe they win from below (data gravity becomes clearing gravity, because the agents have to come to where the data lives). The OS and productivity players (Microsoft) believe they win from above (own the surface where employees actually kick off agents, and the clearing happens wherever the user is). The new crop of Agent native companies believe an entirely new layer emerges! All arguments are right about one thing: whoever wins gets to define the knowledge graph, the governance frameworks, and which workflows get automated first.

Which brings me to the founder question. If you’re building in this market, ask yourself honestly: am I on a path to becoming a clearinghouse, or am I building a feature that clears through someone else’s?

The good news - there are real paths to the clearinghouse seat, even for startups. The first is vertical: earn clearinghouse status in an industry the horizontal players won’t go deep on. Things like owning proprietary data, simplifying regulatory complexity, or workflow depth they can’t absorb from the outside. The second is a bit harder, but more impactful. Become the clearinghouse across clearinghouses. No enterprise is going to run only Microsoft’s agents. Or only Anthropic models. Someone has to govern the multi-vendor mess (the parallel was multi-cloud). Set policy across all of it, clear actions across all of it, hold the audit trail across all of it, etc. Become the “single pane of glass.” The incumbents can’t credibly be neutral here (Microsoft governing Salesforce’s agents? Good luck). That neutral seat is a super strategic position in software right now.

I’ll end with this. Becoming a system of record got you the moat in SaaS. Becoming The Clearinghouse for agents gets you that same moat in AI. The lock in just moves from your data to your permissions. The source-of-truth era is transitioning into the source-of-permission era. Founders should pick their path to the clearinghouse now, because in 18 months there won’t be room to be undecided.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.1x

• Top 5 Median: 26.8x

• 10Y: 4.5%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 16.7x

• Mid Growth Median: 4.8x

• Low Growth Median: 2.6x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 16%

• Median Gross Margin: 76%

• Median Operating Margin 2%

• Median FCF Margin: 21%

• Median Net Retention: 110%

• Median CAC Payback: 44 months

• Median S&M % Revenue: 34%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 13%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Where Are the American Open Source Models?

I’ve been investing in open source companies for nearly my entire venture career. I love open source businesses, think they’re generally great for the ecosystem, and can also create a lot of commercial value (but this can be tricky!). We’ve seen all kinds of open source businesses become successful. Databases (Mongo, Clickhouse, etc), Data Infrastructure (Databricks, Confluent, etc) Developer Tools (Hashicorp, GitLab, etc). And many other categories.

The nuance lies in how you define “open source.” A lot of this comes down to what open source license the open source project uses. At a high level there are three main buckets. First is fully open and permissive (MIT, Apache 2.0, etc) - anyone can use, modify, and distribute the code however they want, including building commercial products on top of it. Second is copyleft (GPL, AGPL, etc) - the code is open, but if you distribute or build on it you have to open source your own code too. Third is source available - the code is publicly readable but the license restricts commercial use in specific ways. This is where you see licenses like BSL (Business Source License) and SSPL, which HashiCorp and MongoDB have both moved to at various points. The main point with source available licenses is that you can basically do everything you can in the permissive license, EXCEPT monetize the project (ie you can’t take the project, create a hosted version, and charge for it). The most “pure” opensource proponents would say only the fully open and permissive licenses are truly open source. Everything else isn’t.

The playbook for many of these open source companies was / is to first become the default in their field. This is the most important part. Spark became the default in large scale data processing, Kafka became the default for real time data streaming and event ingestion. Mongo for NoSQL databases. Etc. Each project became widespread and ubiquitous in their respective fields - they were / are the dominant platform for their use case. Then the commercial companies behind each project figured out a way to monetize. Sometimes this was through hosted offerings - where the company runs and manages the infrastructure for you, so you don't have to worry about deploying, scaling, or maintaining the project yourself. In other instances, proprietary features were gated behind a commercial offering (this could be better performance, enterprise features, etc). Regardless - there were playbooks to monetize, and successful examples of large commercial businesses being built on top of successful open source projects.

But this wasn’t always the case. For a long time, the conventional wisdom was that open source couldn't be a real business. If the code is free, who pays? And the skepticism was genuine - why would a company pay for something their engineers could just download and run themselves? Early open source companies had to fight hard to prove out the commercial model, and a lot of investors passed on these deals early on because the business logic wasn't obvious. "You're just going to give it away?" was a common reaction. And honestly, it wasn't an unreasonable question. The answer turned out to be: yes, you give the software away, but you charge for everything around it - the hosted version, the enterprise features, the support, the integrations, the proprietary engines, etc. The open source project served as distribution, and the commercial product was something different. And after a wave of successful open source commercial companies emerged, it became more obvious what was possible (commercially). And this was very important - investors (like myself) deeply believed successful open source projects could translate into successful commercial companies. This created an important incentive set - an incentive to fund open source companies (by investors) and an incentive to start commercial open source companies (by founders)

Let me tie this back to what’s happening in open source AI. Today, one of the biggest disappointments is a lack of a strong open source model ecosystem in the US. All of the frontier models are closed source from OpenAI / Anthropic. The leading opensource models are largely all from Chinese labs. DeepSeek being the most prominent example, but also Qwen from Alibaba, Kimi from Moonshot AI, and others. Jensen Huang has talked about the importance of having American leaders at every layer of the AI stack - chips, models, infrastructure, applications. The US is clearly winning at chips (Nvidia) and has strong incumbents at the application layer, but the open source model layer is a real gap. And that gap matters more than people realize. Open source models are how developers learn, experiment, and build. They're the distribution layer for the next generation of AI applications. If the dominant open source models all come from Chinese labs, that has long term implications for where developer communities form, where tooling gets built around, and ultimately where commercial ecosystems emerge.

So why hasn’t there been more of a blossoming US open source model ecosystem? I think it all comes down to incentives. In the same way investors had questions about the business model of open source companies many years ago, they have similar questions about open source labs today. How do you monetize an open source model? You could “host” it - which in the case of models probably means just means hosting an API endpoint people can hit for inference. BUT if you’re doing this you’re also going to directly compete with anyone else who can pick up your model and host an inference API endpoint. Baseten, Fireworks, all the hyperscalers, etc. And the question then becomes do you as the model provider have an advantage that lets you serve the model either more performantly or more cost efficiently (or both). The answer to this question is probably “maybe” but not by enough to win with high margins over the long term.

The next natural place is to move into fine tuning / eval feedback loops. Most folks want to take an open source model, fine tune it for their use case / data, observe / monitor / improve that fine tuned model (and rinse repeat). So the open source labs could move into this line of business and charge for it. But then you’re still going to compete with the Fireworks / Baseten / Hyperscalers who will also offer this.

And then there’s the question of what does it mean to be an open source model? There's actually a spectrum here too - and it maps pretty cleanly to the licensing buckets we talked about earlier. The MIT / Apache equivalent in the model world would be truly open weights with no commercial restrictions - download it, fine tune it, host it, build a business on top of it, do whatever you want. The source available equivalent would be something like Llama - the weights are publicly available, you can experiment and build, but there are commercial restrictions that kick in at scale. And then true open source, the equivalent of releasing the full source code, would mean releasing not just the weights but the training data, the training pipeline, the evals, everything needed to fully replicate the model from scratch. Almost nobody does this. So when we talk about the "open source model ecosystem," we're really mostly talking about open weights models with varying degrees of commercial restrictions. Which matters, because just like source available licenses changed the commercial dynamics for traditional open source companies, the restrictions baked into open weights models will shape where developer communities form, what gets built on top, and ultimately who captures the commercial value.

I think the main issue facing the open source model ecosystem is incentives. People (ie investors) just question the business feasibility of open source models. Today, most of the world is using the frontier models for everything. However, what’s really come into focus very recently is costs. It just doesn’t make sense to drive the Ferrari down the street to the grocery store…The Honda will do just fine. You don’t need the “expensive frontier” tokens for everything. I don’t know where the line is, but let’s just use the 80/20 rule. Maybe 80% of the use cases are generic enough where a smaller, less powerful, but cost efficient model will do fine. And for the 20% of use cases you’ll really need the frontier expensive tokens (i’m completely making up that ratio, just using an illustrative example).

So then - why shouldn’t open source dominate in this next part of the cycle?? When people shift usage from most expensive tokens to cheaper tokens, wouldn’t open source be the natural alternative? Yes - but where are the models?? The other baked in assumption is that the open source models are “good enough.” Said another way, the gap between the frontier closed source models and open source isn’t “that wide” (and more importantly is the trajectory widening or closing).

Right now all the best open source models Chinese models… And these models have largely gotten to where they are by distilling frontier models. But I think that was much easier when the models were more “basic” language models. Question / answer type models where you could create a data set by asking millions of questions to the frontier model. I think distillation is getting harder. The most frontier models are now being released with increasingly complex harnesses. Or the frontier models aren’t released via an API but are wrapped into a product themselves. Or look at Claude - playing around with Claude Code, I have workstreams that kick off 50+ agents in parallel. How do you distill a model with 50+ agents running in parallel?? It’s much harder than just getting millions of question / answer pairs… All of this makes it much harder to “distill.” And the cost to pre-train (without the head start of distilling) is massive. All of this to say - I find myself today probably on the side of the gap widening (not shrinking) between the frontier and open source.

So who will be able to raise massive amounts of money, to pre-train a model, that will maybe work, but might end up being too far behind the frontier to capture real market share, and even if you get all of the previous points right there are still questions about how you monetize over the long haul (the first half of this post). Everyone WANTS this to happen. But who’s going to fund it?

I go back to the early days of Neoclouds where Nvidia really propped up the ecosystem. They funded (massively) a class of companies who might not have gotten the funding elsewhere. Do we need King Jensen again?? Or do we need governments to fund it out of a sovereign interest?? I’m not sure. But I really hope the US open source ecosystem finds a way to thrive.

What I think it will take - some sort of research breakthrough that open source gets first. Let’s hope that happens! I also think (no inside info) at some point the closed source frontier labs will get back into open sourcing models. Maybe not the frontier. Maybe a slimmed down version of the frontier, or maybe open sourcing one model release prior to the frontier. This is the most hopeful scenario!

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.4x

• Top 5 Median: 28.9x

• 10Y: 4.5%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 16.3x

• Mid Growth Median: 5.0x

• Low Growth Median: 3.0x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 12%

• Median LTM growth rate: 16%

• Median Gross Margin: 76%

• Median Operating Margin 2%

• Median FCF Margin: 21%

• Median Net Retention: 110%

• Median CAC Payback: 42 months

• Median S&M % Revenue: 34%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 13%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

The Second Life of a GPU

Last week I wrote a post on the opportunity for Neoclouds. At the end I teased out an idea that these businesses could really surprise people if chips retained value after a 4-5 year useful life, and I wanted to unpack that a bit this week.

First - it’s important to go through some of the unit economics / business model of these Neoclouds to understand why the useful life of these chips matter. There’s largely three different types of “deals” different offtakers (ie labs, hyperscalers, AI natives, etc) make with these neoclouds. Bare metal, “managed kubernetes”, and “full cloud.” Bare metal is the most stripped-down offering. The neocloud delivers the physical GPUs, networking, and power, and the customer brings everything else (their own scheduler, orchestration, storage layer, software stack), essentially renting the raw iron. Managed Kubernetes is the middle ground. The neocloud handles the orchestration layer on top of the bare metal (so the customer doesn't have to babysit cluster management, node failures, networking config, etc), but the customer is still running their own workloads and software. Full cloud is the closest analog to what AWS / Azure / GCP offer. The neocloud bundles compute with a full suite of services (storage, databases, networking primitives, managed inference endpoints, observability, etc), and the customer is effectively buying a "cloud" experience rather than just chips. As you move higher up (from bare metal to full cloud) generally you can charge higher rates / hour because you’re offering more.

For this post, I’ll use more “bare metal” assumptions. From the Neoclouds perspective this generally means they’re signing a single offtake site - one customer takes the entire site. It’s not a “cloud” where many different customers (ie scaling AI native startups) all use / share the resources.

To understand the unit economics, let’s first dig into the costs. Neoclouds generally buy the hardware (vs lease, there’s tradeoffs to both). BUT - buying gives you access to the residual value of the chips (which is the point of this post), and generally a lower cost of capital. The other point to highlight is many of these businesses are structured as “sub projects” or “sub entities".” Think of these projects as an individual data center. There’s a parent co (the Neocloud), with a number of projects (ie data centers) sitting beneath the parent.

To finance each project, the sources generally include some combo of project level debt, project level equity (think co-investors who invest equity for return on their equity), and equity from the parent co.

The debt is generally tied to the single contract for the site. The debt provider is looking at counter party risk. What is the likelihood that customer is around for the life of the contract? If they stick around and pay, we (the debt provider get paid). If they go away, go out of business, break their contract, etc, then the Neocloud has to find a replacement customer or they risk not being able to pay back the debt. So it really matters who the end customer is. If that end customer (the offtake) has high credit worthiness, a good reputation, etc, the debt provider probably gives a better rate.

When I say the debt is tied to the contract, what I mean is the debt is amortized over the lifetime of the contact. If the single offtaker is signing a 4-5 year deal, the debt will have a 4-5 year term (ie paid off in 4-5 years).

Ok - so let’s bring this back to the original point of the article. What are the implications of chips having a longer useful life than 4-5 years. After the Neolabs have “completed” a contract with an initial offtaker (ie 4-5 year deal), the debt has been fully paid off. IF the chips have value after that period, the Neolab can “recontract” the site. It will certainly be at meaningfully lower price per hour per chip, BUT without an interest expense the profit margins can skyrocket. These recontracting deals will almost certainly be for inference vs training (folks will want to train on frontier chips), BUT if you believe inference will explode (which I do), there will be SOO much demand for inference compute in the future. And these recontracting deals could prove to be extremely profitable.

I’m writing this post much later in the evening than normal, and I think I’m starting to ramble / not write as clearly as I’d like…But the takeaway - I believe chips will have a longer useful life than the original contracts they’re being signed to support, and if they do we could see a meaningful lift to neolab profitability.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.2x

• Top 5 Median: 26.8x

• 10Y: 4.4%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 15.9x

• Mid Growth Median: 4.8x

• Low Growth Median: 2.6x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin 0%

• Median FCF Margin: 21%

• Median Net Retention: 109%

• Median CAC Payback: 33 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

The Neocloud Boom

** Caveat — This post is not a recommendation on current Neocloud stocks, or commentary on their valuation, but instead some personal observations on the market as a whole **

A Neocloud boom feels inevitable. Clicking out one layer, the data center infrastructure buildout feels like it could turn into one of the largest wealth creation moments ever in physical infrastructure. Now that I’ve spoken in absolutes like this, we can bookmark this post for later when we look back on “signs of the top” :)

Let me caveat this post with the fact that I’m very AGI pilled. Just about any estimate for “tokens consumed by X date” or model progress or data centers built or total demand I’m taking the over.

In all seriousness, the numbers are staggering. Rumors / reports peg Anthropic / OpenAI at ~3-3.5GW of capacity to end 2025. OpenAI has talked about getting to 30GW by 2030. Let’s assume Anthropic has similar plans. Just those two alone will bring on (or plan to bring on) ~55GW over the next ~4.5 years. Now let’s assume OpenAI / Anthropic represent 30-40% of the new builds over the coming 4.5 years (I asked Claude for its best guess on this figure, I have no more specific insights here other than Claude’s estimation). That’s >150GW of new capacity in the next 4.5 years alone.

Ballpark figures are ~$50b total costs to build 1GW. For a Blackwell dominated data center, that roughly breaks down into:

• ~70% for chips / compute.

  • 50% GPUs

  • 10% networking

  • 5% CPUs

  • 3% Storage

  • 2% assembly

• ~20% for data center shell + power infra (substations, transformers, switchgear, backup generators, hv interconnect, etc)

• ~5% cooling

• ~5% land, permitting, and other costs

So in total, ~$50b for a 1GW build, multiplied by 150GW = $7.5Tn total spend. Spread across 4.5 years that’s ~$1.7Tn / year, or ~5% of annual US GDP (~$32Tn / year). I asked Claude to come up with some historical precedents to put this in perspective:

• US Railroad building out in the 1880s was ~6% of GDP

• Telcom building in early 2000’s was ~1.2% of GDP

• The Manhattan Project was ~0.4% of GDP

So this “AI Buildout” is high, but not as high as the railroad buildout.

So that’s the costs - $7.5Tn over 4.5 years. Someone is going to make a ton of money financing all of that.

But how about the revenue / enterprise value creation serving all of it. This is where neoclouds come in. Here’s a quick scorecard on 3 public neoclouds today: Coreweave, Nebius and IREN

This is a totally made up multiple, but Coreweave / IREN are both at ~$90b of enterprise value per live GW. Nebius quite a bit higher (again, not a comment on valuation, just doing some fun math).

Let’s say the Coreweave / Nebius “math” holds for rest of industry and $90b of enterprise value is created for every deployed GW. If we’re going to bring ~150GW online in next 4.5 years, that equates to $13.5Tn of enterprise value creation…Again, this is all funny math…Lots of that build will be done by hyperscalers, a lot of it for their own internal consumption vs to sell to others. But the numbers are wild!

Even if you assume only ~20% is captured by Neoclouds, that’s over $2.5Tn of enterprise value created in the next 4.5 years. The numbers are so big, it can’t be served by only a few players. The only natural explanation is that we’ll see a boom in Neoclouds, hence the title of this post. In a few years I bet there will be a handful of very large Neoclouds, and then a long tail (including independently owned single site neoclouds, similar to how the IPP (independent power providers) market played out. Expect PE to play a big role!

Obviously you could say my initial assumption is BS - that we’re not going to bring online another 150GW by 2030. Either because you don’t believe AI will garner that demand (but I’ll point you to my initial caveat! I’m very AI pilled…). or you think the costs are just too high as a percent of GDP. But let’s dream the dream - this could be a MASSIVE opportunity…And as far as the venture communities go, there is still a lot of skepticism around neoclouds (I’d ask them, who else is going to build out the compute for AGI!). In many ways, Neoclouds are the perfect hedge on a venture portfolio :) If everything else goes to zero because AGI captured all the value, someone will have to build, deliver, and serve that compute :)

Ok last little bit here. I was done writing the post, but thought I couldn’t write a post about Neoclouds and not bring up SpaceX. SpaceX? How are they a Neocloud? I’m sure most of the readers of this blog have been plugged into the AI news cycle, but in case not, here’s a quick summary. xAI was a AI lab Elon founded. They created two very large data centers of their own. Colossus 1 (which is estimated at ~300MW), and Colossus 2 (which has total capacity of 2GW, but estimates peg it at ~500MW operational today). SpaceX then bought xAI. They then proceeded to “rent” capacity at Colossus 1 and Colossus 2 to Anthropic! In the SpaceX S-1, they gave some details on this commercial relationship.

According to the S-1, Anthropic is paying SpaceX $15b per year (!) for this capacity. They supposedly have all of the capacity in Colossus 1 (~300MW). Impossible to know how much of Colossus 2 they have, but let’s assume it’s ~40%? That would give them an incremental ~200MW. SpaceX/xAI has to keep some for itself…and Cursor has announced they have some of that capacity (which Elon as agreed to buy as well…are you keeping up with all the acquisitions??).

So that’s $15b / year for ~500MW. To put those numbers in perspective:

• SpaceX did $18.7b of revenue in the full year 2025. And they just “created” a new business line item of ~$15b in an instant! Insane. SpaceX was founded in 2002. It took them 23 years to get to an annual revenue figure of ~$19b. And with their new “neocloud” they generated almost that same amount. The opportunity for Neoclouds is massive!

• This math also assumes Anthropic is paying ~$30m / MW ($15,000M / 500MW). That would be really high…Most neocloud comps are closer to the $10-$12m / MW. Coreweave guided to ~$18.5b annualized run rate revenue to end 2026, and said they’re “on track to reach or exceed our target of more than 1.7GW by the end of 2026.” This math comes out to ~$11m / MW. If Anthropic was paying $11m / MW that would mean they have ~1.4GW of capacity from SpaceX (which isn’t possible).

The $30m / MW does feel high…BUT - the contract has a 90 day out. Anyone can cancel the deal with a 90 day warning. And Anthropic is VERY compute constrained. So might they stretch on a one time deal with an out? Maybe!

Either way - I’m bulled up on the opportunity for Neoclouds :) The revenue potential is clearly massive. Now they just need to prove they can make the business model work! On revenue multiples, CoreWeave trades at ~6x NTM rev, Nebius / IREN at ~10x. On NTM EBITDA, CoreWeave is trading ~10x, with IREN / Nebius ~20x.

I believe trillions of enterprise value will be created in the neocloud space in the next 5+ years. Go grab it! I may write a piece next week that discusses why these businesses could really explode if the chips retain value after years 4-5.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.2x

• Top 5 Median: 25.0x

• 10Y: 4.6%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 15.6x

• Mid Growth Median: 4.8x

• Low Growth Median: 2.6x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin 1%

• Median FCF Margin: 20%

• Median Net Retention: 110%

• Median CAC Payback: 36 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 14%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

The Real App Store Opportunity

Last October I wrote a piece saying OpenAI had their “App Store” moment after they released the Apps SDK. 7 months later that prediction doesn’t look great… I don’t think we’ve seen an explosion of custom ChatGPT apps. ChatGPT hasn’t turned into the “super app” yet. Hopefully one day they will!

I think there may be a separate “app store” moment happening with Anthropic. BUT - more of a B2B app store moment than B2C apps. Over the last few months I’ve seen massive adoption of “skills” in Claude. A skill is essentially an "onboarding doc" for an AI agent - a folder of instructions (often just a markdown file) that Claude pulls in only when the task calls for it. Anyone in a company can write one in an afternoon, which is why I think the distribution dynamic looks less like a consumer app store and more like internal tooling that spreads bottoms up. At our own firm we’ve seen a proliferation of skills being created and shared with the team.

What really changed thigs for me was the /skill-creator “skill.” It’s become so easy to start building skills (maybe skills are the new name for agents??) to automate real parts of my day to day. The biggest challenge I have with building skills is a creative one. I know there are 10x more ways I can be using / building skills, but I don’t always know where to start. So I discuss with my colleagues, with other friends in similar lines of work, to try and explore how other people are leveraging skills so I can do something similar.

The “app store” moment hasn’t happened yet, but I think we’re close. Today, we can share skills within our own organization. But we can’t yet share them externally. And there’s not great approval flows for making sure a built skill follows all security / compliance protocols. Anthropic has their own set of pre-built skills (including ones from partners like Notion, Figma, Atlassian, etc), but it’s managed by them (ie isn’t a marketplace). I met with a software vendor this week who has some of their own “skills” available out of the box (the interface is their own software UI, but they still called some of the automated workflows “skills.”)

All of this to say - I think the trend here is a set of skills people can build and then publish to some sort of marketplace. I think you can kind of do this today by exporting the markdown file and publishing on GitHub, but that feels janky. What people really want is to be able to find a skill, know it’s somehow been “blessed” by Anthropic (ie put through security reviews, deployed within Claude behind their governance, etc), and quickly customize it for their own set of integrations / permissions.

There's another angle here that I think founders building AI products need to internalize. For the last decade of SaaS, "we integrate with X" was table stakes. You couldn't sell into the enterprise without a long list of logos on your integrations page. I think "we have a skill for that" is about to become the equivalent over the next 12 months. If your customer's primary surface for getting work done is Claude (or any agent), and your product can't be invoked as a skill from inside that surface, you're functionally invisible. It doesn't matter how good your product is - if the agent doesn't know to reach for you, you're not in the workflow.

Everyone today has a “blank canvas” problem. They know there’s skills they could be building, but they don’t know where to start. A marketplace like experience of skills just makes too much sense not to happen! When will we see our first venture funded “skill”??

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.1x

• Top 5 Median: 23.4x

• 10Y: 4.5%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 14.7x

• Mid Growth Median: 4.4x

• Low Growth Median: 2.3x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin 0%

• Median FCF Margin: 21%

• Median Net Retention: 109%

• Median CAC Payback: 33 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

When Machines Out-Eat Humans

Fun post this week. I want to write about my own (newly formed) definition of AGI. I think we'll hit AGI (or we can claim AGI) when as a society we decide the marginal unit of energy is better spent on a GPU (or whatever compute primitive exists at the time) than on a human. Said another way - when the energy consumed by compute becomes greater than the energy consumed by humans, we're making the implicit decision that we get higher utility out of sending energy to machines. All definitions of AGI are super wishy washy anyway, so why not through another into the mix! The reason I like this one is it's quite quantitative. I’ve run the math, and the answer is 2033 (as you’ll hear me describe later, it’s all a bit “funny math dragging assumptions to the right”) but that’s what makes it fun!

First - let’s start with some definitions. In today’s world people use power vs energy interchangeably, but they’re different. Power is a rate (how fast you’re using energy at any given time), while energy is a quantity (how much you’ve used over time). An oversimplified analogy - power is the speed your car is going (mph), energy is the total distance (miles) you’ve traveled. A car going 60mph for 2 hours has covered 120 miles. A data center pulling 100MW for 2 hours has consumed 200 MWh (mega watt hours).

From a unit standpoint, watts are the unit of power. My lightbulbs might be 9W (It needs 9W to turn on). I keep it on for 1 hour, and I’ve used 9Wh. Kilo / Mega / Giga / Tera are just scaling prefixes (1,000x, 1,000,000x, etc). When we hear power figures quoted for data centers we often describe them in mega watts or giga watts. For example, the Colossus 1 data center that xAI rented to Anthropic this week is quoted as 300MW. What does this mean? It means Colossus 1 needs transmission lines, substation(s), generation source(s) that can sustain 300 million watts of instantaneous draw. Think of this more as a physical infrastructure problem.

This data center won’t consume 300MW x 8,760 hours / year. Maybe it runs at 70-80% utilization (but AI training clusters are “running” constantly at higher utilizations).

The confusion most people have with power vs energy, is the “hour” component is not “division” it’s “multiplication.” It’s not MW per hour, it’s MW-hours (and yes, I just wrote a “it’s not X it’s Y sentence…tried to edit it out so it doesn’t sound like AI slop, but the it’s not X it’s Y just made too much sense). Super quick math to make this concrete - let’s say my house consumes 1KW for 10 minutes, then 1.3KW for 50 minutes. To calculate the total energy used:

• From 0 to 10 minutes: drawing 1 kW for 10/60 of an hour = 0.167 kWh

• From 10 to 60 minutes: drawing 1.3 kW for 50/60 of an hour = 1.083 kWh

• Total energy over the first hour ≈ 1.25 kWh

Let’s also use the 300MW Colossus site, running at 80% utilization(ie running 80% of the time)

• 300MW x 8,760 hours / year x 80% utilization = 2,102,400 MWh

• This translates to 2,102 GWh, or 2.1TWh

• So the energy consumption over a year is 2.1TWh

I walked through that lead in so that I could then bring it back to my new definition of AGI! When will total energy consumed by machines be greater than humans. To answer that, let’s first try to get a baseline on how much energy is consumed by humans to sustain human life! A very meta question…Obviously there’s no easy answer, but I used Claude to help frame it (ie guesstimate it). Focusing on just US.

Here was Claude’s answer: 27,600 TWh of total US energy consumption. This is how Claude broke it down:

Narrow definition:

• Residential energy use: 5,600 TWh

• Food systems (ag + processing + transport+ retail): 2,900 TWh

• Healthcare delivery: 590 TWh

• Water + sewage treatment: 290 TWh

• Total = 9,400 TWh

Moderate Definition. Everything in the narrow definition plus:

• Personal transportation: 9,400 TWh

• Commercial buildings: 4,100 TWh

• consumer goods manufacturing: 2,000 TWh

• Total = 20,500 TWh

Most Aggressive Definition

• According to the US Energy Information Administration (EIA) the US consumed 94.2 quads of energy in 2024. Converting that to TWh = 27,600 TWh

• This is all the estimate for all of the energy the US economy consumes, in any form, measured at the source. Every barrel of oil burned in a car. Every cubic foot of natural gas burned in a furnace. Every ton of coal or pound of uranium fed into a power plant. Every BTU of solar or wind generated. etc. A figure you may see quoted if you ask Claude how much electricity is consumed annually in the US is closer to 4,000 TWh (but that is just looking at the electric grid > offtake figure). Most energy is never electricity to begin with (and that’s the gap). This is the difference between the 4,000 TWh figure and the ~28,000 TWh figure

• The gap between the “moderate” definition and “aggressive” definition is largely military, freight, etc

So in summary, we can look at annual human energy consumption in the US as

• 9,400 TWh in most conservative case

• 20,500 TWh in the moderate case

• 27,600 TWh in the most aggressive case

What’s a good assumption for total annual data center energy consumption? I asked Claude, and the answer was ~300TWh. Here’s how the math breaks down

• Roughly 40GW of total data center capacity (this felt slightly high, but the Claude estimate was based on all hyperscalers, all labs and neolabs, and all neoclouds in the US)

• If you assume 80% utilization, then:

• This equates to 280TWh of data center energy consumption

So today, the data center energy consumption is 280TWh. Earlier, I said the “moderate” estimate for annual energy consumption to support humans was 20,500TWh. This means today, data centers consume 1% of the energy humans do. Long way to go to AGI :)

So when will we hit the 50% threshold?? Let’s do some more “funny math” :). Let’s say annual population growth is 0.5%, end users consume 1-2% more energy per year, but we get 1-1.5% energy efficiency gains / year. This leads us to ~0.5-1% annual growth in energy consumption, or 21,440 TWh in 2030.

Now, let’s assume OpenAI is going from 2GW in 2025 to 30GW in 2030. And let’s assume the rest of the industry grows at the same rate (maybe aggressive, but just for fun let’s make that assumption). This would get us to 12,600 TWh in 2030 used for data centers. The lines are getting closer! 12,600 TWh for data centers vs 21,440 TWh for human consumption = almost 40%!

Ok final assumptions :) Let’s assume data center growth continues at the same rate in 2030 and beyond as it’s projected to from 2025 to 2030. And let’s assume human consumption continues at ~0.75% / year beyond 2030. When do the lines cross?? The answer is 2033!! This would continue an exponential, which is very hard to do…But if we went from 40GW today to 600GW in 2030, at the same compounding annual growth rate we’d be at ~3TW in 2033. Doing all the funny-math converting that to TWh gets us to ~22,000TWh by data centers in 2033. Extrapolating the human consumption at 0.75% / year gets us to ~22,000 TWh in 2033. The lines have crossed!

Of course, this is all assumption upon assumption upon assumption, and I’m no energy expert at all, but thought it’d be fun to run through this exercise.

AGI by 2033!

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.2x

• Top 5 Median: 21.1x

• 10Y: 4.4%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 12.6x

• Mid Growth Median: 5.2x

• Low Growth Median: 2.5x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin 1%

• Median FCF Margin: 20%

• Median Net Retention: 110%

• Median CAC Payback: 36 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 14%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

The Death of Per-Seat Pricing?

All three Hyperscalers (Amazon, Google, Microsoft) reported earnings this week.

There was one quote from this earnings cycle that I think will get a lot less attention than it deserves. It came from Satya:

“The basic transformation of any per-user business of ours - whether it is productivity, coding, or security - will become a per-user and usage business. That is the best way to think about it.”

Big statement! The per-seat licensing model is the foundation that the entire modern SaaS industry was built on. It’s how so many IT budgets are structured. It’s how every renewal conversation goes. It’s how every comp plan is designed. Did Microsoft just say the seat is dying? Or at least, being repurposed? This isn’t the first time I’ve written about this topic...Nearly 2 years ago I wrote a similarly titled article, “Is seat based pricing dead“?

Amy Hood went further. She framed the new model as “a licensed business plus a consumption business applied far more broadly than I think people have thought about.” And later, Satya put a really nice frame on it - “the seat-based pricing is just entitlement to some consumption…there are some base usage rights that get bundled in or packaged into seats.” Said another way - the seat remains, but it becomes a packaging mechanism for prepaid consumption. Beyond a certain level, you’re paying per token, per agent action, per outcome.

They also announced a larger pricing change of their own! They moved GitHub Copilot to a usage-based model effective June 1. ~60% of their Dynamics 365 customer service customers are already buying usage-based credits. The Copilot credit-consumption offer was up nearly 2x QoQ. They’re already living in the new model.

There’s actually a tell buried in the bookings line. Hood called out that D365 bookings growth was impacted by weaker renewals “as customers balance spend between the traditional per-seat and the emerging seats-plus-consumption model.” The transition is already showing up as a drag on a legacy bookings metric. Customers are pausing on their old seat renewals because they’re trying to figure out the new pricing. It’s happening right now.

Why does this matter for founders building software companies?

A few reasons. First - if Microsoft is publicly committing to this shift, every other software company gets some air cover to do the same. The market will tolerate the transition pain because the largest player in the world is wearing the same lumps. If you’ve been afraid to introduce a usage-based component to your SaaS pricing, watching Microsoft go first is permission.

Second - the metric set you’ve been using to evaluate your business will need to change. ARR is a clean metric for a per-seat business. It is a much messier metric for a “per-seat + consumption” business. NRR will likely become more volatile (consumption ebbs and flows). Bookings will be lumpier. The “rule of 40” framework, the public market revenue multiple, the way you set sales quotas - all of it gets harder. Investors are going to have to learn a new language, and it’ll probably take some time to work its way through the system.

Third - the agent economy is fundamentally a consumption economy. An agent doesn’t buy a seat. It does a task. It uses tokens. It calls tools. If you’re building a software company today, and your business model assumes you’re going to charge a flat fee per human user, you’re solving for a world that’s about to get smaller. The world that’s growing is “per outcome, per task, per token.” Get there before you have to. Just about every startup I’ve worked with that went through some sort of seat > usage business model transition regretted not doing it sooner…

The thing I keep coming back to is - Microsoft is the least incentivized of any company in the world to do this. They have the largest installed base of seat-based subscriptions on the planet. Office, Windows, Teams, Dynamics. Every dollar they “convert” from seat to consumption introduces volatility into a model their investors have been comfortable with for two decades. The fact that they’re still leaning into it tells you everything about where they think this is going.

Seat-based pricing is becoming more of a wrapper than a product. The product is the work that gets done.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.2x

• Top 5 Median: 18.4x

• 10Y: 4.4%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 10.7x

• Mid Growth Median: 5.0x

• Low Growth Median: 2.4x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin 0%

• Median FCF Margin: 21%

• Median Net Retention: 110%

• Median CAC Payback: 33 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

The Impending Employment Explosion

I’m a perpetual optimist. It’s hard for me to see the world through any other lens! Sometimes I’m naive, but overall I think it’s a better way to live (and, generally, hard to bet against humanity’s resilience).

One of the larger debates surrounding AI relates to the impact it will have on employment. One side (booo, the pessimists!) argues we’ll see a collapse in employment as AI takes everyone’s jobs. The other side argues some form of “Jevon’s paradox” - with massive positive economic benefits. Given my intro, I think it’s clear what side of that debate I fall on!

There are many ways I’ve framed this in the past (I think I’ve even written about it in a prior week’s edition). However, I heard Jensen recently articulate it much more eloquently (surprise surprise, he’s better at framing this than me!). The crux of his argument (which I wholeheartedly agree with), is that the framing of “AI will take jobs” is wrong. In order to actually debate this, you have to separate the job, from the task. What does this mean? Let’s use an example. The job of a software engineer is to build high quality software. A task of a software engineer is writing code in an IDE. AI may in fact automate one task of a job (ie automating writing code which will no longer be done manually), but it won’t replace the job - building high quality software. Because there is more to the job than just one of its tasks. In fact, I’d argue AI will have the opposite impact! AI will make the task much easier, leading to an explosion of people doing the job! This isn’t anything novel (it’s a common claim to say AI tooling will lower the barrier to creating software, leading to lots more software created!). But I hadn’t heard it framed the way Jensen did - separating the job from the task.

What’s the common pushback to this? Well what if the one task represents a significant portion of the job! Saying software engineers are “software builders” is just putting lipstick on a pig! They’re really just code writers!” I get the pushback, but I think it misses something important. The hard part of being a software engineer was never the typing. It was never the syntax. It's understanding the problem deeply enough to know what to build. It's knowing how to architect a system, how to make the right tradeoffs, how to debug the thing when it breaks in some way nobody anticipated. The code is the output (not the job). And look, maybe the skeptic is right that writing code is 60-70% of the time spent today. Fine! But that just proves the point even more. If you can compress 70% of the time, you don't get rid of the person… you get that person shipping 3-4x more. And what does every company in the world want? More software. More tools. More internal apps. More automation. More everything. The demand for software is nowhere close to saturated. We've been supply constrained, not demand constrained. So when you make it much easier to supply software… you don't get fewer engineers. You get more. The nature of the software engineer job will change (probably a lot!). But the job itself? It gets bigger, not smaller.

Here’s a hypothetical example I like to use. Let’s say we existed in a world before cars were invented. And if you wanted to travel from your home to the grocery store you used a horse drawn carriage. Let’s call the person operating the carriage the chauffeur. Then, let’s assume some technological innovation happened EXTREMELY quickly, and all of a sudden cars were hitting the road (the key part of the analogy here is the part in all caps - this change happened QUICKLY, just like AI has). A pretty pessimistic take may be “Ah! This is terrible! Think of all the chauffer’s who will loose their job!” But it’s pretty easy to see the world through a different lens. What it means to be a chauffer stays the same (you help people get from point A to point B). The task of a chauffer was operating a horse drawn carriage (and maintaining it, buying parts for it, etc). Now, the task of a chauffer changes! Their job is still helping people get from point A to point B. BUT the task changes - now they need to learn how to operate a vehicle (how to drive a manual transmission car) vs how to use the reins to help horses navigate.

And what happens in this world? We have way more chauffer’s because transportation has instantly become way more efficient (ie less productivity waste). At the same time, entire new industries are born servicing the automotive industry! Auto mechanics, oil and gas, auto manufacturers, etc.

Yes - the task of operating a horse drawn carriage, or manufacturing the carriage itself went away. But the job of a chauffer saw a spike in demand (leading to more employment), and a trickle down of job creation in adjacent industries.

I’m so convinced we’ll see the same thing happen with AI, and an explosion of employment is coming. And we do have some precedent to support this! ATMs came out in the late 1960’s. However, from that point in time to the mid 1980’s the number of bank tellers in the US doubled. Tellers went from cash-in / cash-out machines (the task) to relationship bankers who cross-sold financial products (the job). The task got automated, but the job got more valuable. At the same time, the task got way cheaper! According to Claude: “Before ATMs, you needed ~21 tellers to staff a branch. After ATMs hit saturation, that dropped to ~13. So the cost per branch went down meaningfully.” Cheaper branches lead to more branches. And more branches led to more tellers (in aggregate). I asked Claude to chart the number of bank tellers in the US. What do you see, number of bank tellers exploding at the same time the ATM came out)

You could make a similar argument around lawyer growth in the 70’s, 80’s and 90’s. The job of a lawyer is to protect and advance their client's interests, the some of the tasks include researching case law and drafting documents. Before the PC, a huge chunk of a lawyer’s time was spent on tasks that technology was about to automate. Legal research used to mean physically going to a law library, pulling case books off shelves, reading through indices, cross-referencing citations by hand. That could eat days or weeks. Then Westlaw (1975) and LexisNexis (late 70s) came along and compressed that to hours. Later, full-text search made it even faster.

Document drafting was similar. Contracts, briefs, motions - all typed up by hand or dictated to a secretary who typed them. Every revision meant retyping the whole thing. Word processors (and later PCs with WordPerfect / Word) made it so one lawyer could produce 5-10x the document output. So did things that automated the “tasks” of lawyers (PC, internet, LexisNexis, word processors, etc) lead to a reduction in the number of jobs held by lawyers? Here’s a chart from Claude:

I feel like there’s so many other examples I could use. Radiologists (this is an example Jensen uses). In 2016 Geoffrey Hinton (Godfather of Deep Learning!) predicted radiologists would go away because of AI. What happened? The job of a radiologist is patient care. The task is reading and interpreting scans. AI got really good at the task. But the job didn't shrink (it grew!). Because the demand for imaging exploded (more types of scans, more conditions screened for, aging population), and radiologists' job of providing patient care evolved toward more complex interpretive work, interventional procedures, and clinical consultation

The through line is the same: automating a task is not the same as automating a job. And more often than not, it leads to more of the job, not less. I expect AI to be no different.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.0x

• Top 5 Median: 18.4x

• 10Y: 4.3%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 10.6x

• Mid Growth Median: 4.9x

• Low Growth Median: 2.2x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin 0%

• Median FCF Margin: 21%

• Median Net Retention: 109%

• Median CAC Payback: 33 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 14%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

I have two topics I want to write about today:

1. Are we in an “over earning” period in AI

2. Follow up to last weeks post on the importance of the harness

Rising Tide, Hidden Risk

I’ve been thinking a lot recently about comments Satya made a few years back. If we rewind the clock to mid / late 2022, the biggest thing on software companies & investors’ mind was “when will the optimizations end.” The ZIRP period of 2020-2021 created a buying frenzy - no one was thinking about costs (when it came to cloud / cloud software spend), they were only thinking about growth and capturing more market share (oversimplification, but you get the main point I’m making). At the end of the day, the market was providing cheap and abundant capital (for public and private companies), and investors (for both public and private companies) were rewarding growth (ie placing the most emphasis on growth when determining valuation multiples). Once ZIRP ended and interest rates went up, capital got more expensive, investors started rewarding different behavior (efficient growth), and ultimately every CFO took a look at their P&L and realized they had a TON of unnecessary spend. Redundant spend, inefficient spend, or completely unnecessary spend was everywhere. After a couple years of “reckless spending” the tide turned, and every CFO was looking for “optimizations.” Way to reduce spend, consolidate spend, and in general reduce the wasteful spend. This was pain lasted quite some time! And this brings me to comments Satya made over and over again on earnings calls - cloud consumption trends tend to be quite cyclical. You have period of rapid expansionary spending, followed by periods of rapid optimizations - and this cycle repeats.

When I look at where we are in the AI spending cycle, we’re very clearly in a period of “rapid expansionary spending.” No one is really thinking about the ROI of the spend when budgeting. So many of the companies I work with say some version of “I don’t want to cap how much our employees use these tools, I want their creativity flowing, I want them trying out everything, and ultimately I want them becoming experts in these tools.” A common conversation went something like “our spend on AI tools (more often than not Claude) doubled from Jan to Feb, tripled from Feb to March, and is trending way up in April. And I don’t care if it 10x from here.” Then of course on top of this you have the “tokenmaxing” companies. Meta was in the news on this recently - the idea is “as an employee I need to show I’m using these AI tools a lot so I can show my boss I’m “AI Fluent" so I don’t get laid off. I want my token consumption to be super high, so I’m just going to have background agents running constantly, doing nothing, to give my boss the perception I’m using these tools a lot, even though they’re all doing empty tasks.” Clearly this kind of excess spend can’t last forever…

The common thread? No one is thinking about the costs. Just the potential.

I don’t know when the tide will turn, or what will cause it to turn. But at some point people WILL start looking at their AI spend and start thinking about “optimizations.” Should I be using smaller cheaper models vs larger ones? Should I be segmenting different use cases for different models? Should I run a particular background agent once per day vs once every hour? the list goes on.

In 2022, what caused the tide to turn was interest rates. Now? Interest rates won’t (I don’t think) be the reason the tide will turn. What most likely happens is the spend on AI tools balloons SO much, that companies have no choice but to think about the costs, and how to optimize. And when this happens, it could be painful for a number of AI companies experiencing hyper growth currently. There are just too many markets, that have too many competitors, that are all growing at ridiculous rates.

The thread I’m pulling at - a lot of companies in AI today are “over earning” just like many cloud companies were “over earning” in 2020 / 2021.

Just to call it out - I think the total token consumption will SKYROCKET over the next 10 years regardless of when this period of optimization happens. But today, everyone seems to be benefiting. When the tide shifts, the dispersion of who will truly benefit will shift, and a smaller subset of companies will benefit disproportionately.

Call out for founders - really introspecting your own customer / revenue / growth to understand “am I growing as part of the rising tide or am I differentiated” will be quite important…Many companies in the 2021 period got this wrong - they didn’t realize they were just catching the rising tide. They over hired, over fundraised, etc. And the unwind was painful.

Topic #2: Is the Harness Really the Moat?

In summary, last week I wrote a post that argues that the code and orchestration surrounding an AI model (the "harness," maybe formerly known as a "wrapper") matters a ton - maybe even more than the model itself? I’m back to write a quick follow up to this post.

I do think the harness matters a ton, and will continue to matter. BUT - it’s far from the only thing that matters. And we’re aready starting to see the evolution of some of the most successful application AI companies. (1) Start as a thin wrapper, (2) develop a complicated harness, (3) post train your own models, (4) lean into pre training your own models.

The vast vast majority of the market is still in phase (2). But we’re already seeing companies like Cursor (and other early AI winners) move into 3 and 4 (some form of developing their own model).

There are many reasons why AI companies may want to have their own model.

First - cost. Costs balloon when using someone else’s model. Your fixed costs will be higher developing your own, but variable costs may be lower. Second - capability. A purpose built model for your application and your domain could perform better than a general model. Third - control of your own destiny. No one wants to be beholden to someone else for the most critical layer of their infra…

And fourth - you may have no choice! There’s a very real world where the large labs (OpenAI / Anthropic) decide to keep their most powerful and current versions of their model for themselves, powering their own applications. And the models they release via API to customers may be “one version prior” of a model (ie one before the latest and greatest) or a distilled version of the current powerful model.

Why would they do this? Two main reasons I can think of. The labs are getting more and more into the application space. Why would they want their competitors (ie Cursor competing with Claude Code from Anthropic or Codex from OpenAI) to have access to the latest and greatest? Why not save the latest and greatest for their own product, and make their competitors use an older, less powerful version? I could certainly help the labs win more business.

Second (and this is all purely speculation), as these models get more and more powerful (ie Mythos speculation from Anthropic), the labs may view them more and more as a safety / security threat. Maybe they will want to be the only ones who “release” them to the world wrapped around a product? They could view themselves as the only ones truly capable of “protecting” users from the uncapped potential of the models. Said another way, they might trust themselves more to build applications around the powerful models, but not your average developer..

End of the day, this is all speculation. It may happen, it may not. But if you’re a breakout AI company, do you want to leave it to chance? Probably not. Developing your own model carries too many benefits, and could protect you in too many “downside cases” that I think we’ll see more and more companies move from the harness as the first moat, to developing their own model as the second moat.

The takeaway for founders I work with - are you building your own research / labs team? if not, maybe time to think about doing it…

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.2x

• Top 5 Median: 17.2x

• 10Y: 4.3%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 10.5x

• Mid Growth Median: 5.0x

• Low Growth Median: 2.4x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin 0%

• Median FCF Margin: 21%

• Median Net Retention: 109%

• Median CAC Payback: 33 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 14%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

Long Live the Harness (Wrapper?) !

In the early days of AI, we saw the rise of “GPT Wrappers.” Companies that created a product that resembled a thin layer on top of a model. People loved to mock these products, saying all the value was in the model with everything around it commoditized. “Why would I use your app when I can just use ChatGPT directly?” Years later, we have a new name for “wrapper” which is now “harness.” OK that’s a crude analogy and not exactly apples to apples... a harness is really the code that determines what information a model sees at each step, what to store, what to retrieve, and what context to present. It’s the scaffolding around the model. But the spirit of the comparison is directionally right: there’s an enormous amount of value in what sits around the model, not just the model itself. And we now have data to prove it.

Stanford just released a study called Meta-Harness that showed something pretty remarkable. Changing the harness around a fixed model (same model, same weights, nothing different about the model itself) can produce a 6x performance gap on the same benchmark. 6x! Same model. The only thing that changed was the code wrapping it. Their system, which uses an AI agent to automatically search for better harnesses, beat the best hand-engineered solutions by 7.7 points on text classification while using 4x fewer tokens. It hit #1 on an actively contested coding benchmark. And the harnesses it discovered for math problems transferred across five completely different models that were never seen during the search process.

The model race, while important, is not the whole game. For the last few years, the industry has been laser focused on who has the best model. Trillions of dollars of value creation tied to “my model is smarter than your model.” And yes, model quality matters. But this research suggests that the orchestration layer around the model, how you manage context, what you retrieve, when you retrieve it, what you store, how you handle errors, is where a massive amount of real-world performance also lives.

For founders, this is pretty freeing. You don’t need to train your own foundation model. [Side note - we ARE seeing a lot of leading application companies train their own models (either post trained or pre trained) which is contrary to what I just mentioned. I think that trend continues which is something I’ll write about next week] You don’t even necessarily need to pick the single “best” model. What you need is to build the best harness for your domain. If a 6x performance gap comes from harness quality on a generic benchmark, imagine what a deeply tuned, domain-specific harness could do in your vertical. There’s new value to the wrapper (harness)! To my earlier “side note” - I think we’ll see some of the most successful application companies start off by building a killer harness, and then overtime parlay that into amazing data collection, which is then parlayed into a post trained model, and ultimately a pre trained model of their own. BUT - i think the key first step for many will be a killer harness.

One of the findings from the paper that I think deserves more attention is about compression. The researchers tested what happens when you summarize the feedback given to the harness optimizer versus giving it the raw, uncompressed execution traces. Summaries made things worse. The full execution traces (the raw prompts, tool calls, model outputs, state updates from every prior run) outperformed the compressed version by 15 points at median. Now, this probably doesn't shock anyone who's been building agents... we all know intuitively that you lose something when you over-abstract (ie compress). But having it quantified at a 15 point delta is useful. It puts a number on the cost of being lazy with context. And it runs counter to a lot of the default patterns people are using today, where the first instinct is to summarize everything to save tokens and reduce costs. There's a real tradeoff there, and most teams are probably leaving performance on the table.

There’s also something kind of wild happening here from a “meta” standpoint. The system Stanford built uses Claude Code as the agent that writes better harnesses. So you have AI... writing the orchestration code... for other AI systems. And it’s doing it better than humans. We hear a lot about recursive self-improving models lately, is this a good implementation of that?? It’s happening right now, in a practical engineering context, producing measurable results on real benchmarks. Agents improving agents. We’re going to see a lot more of this.

So if the harness matters this much, who owns it? This week Anthropic launched Claude Managed Agents, which is Anthropic saying “we’ll own the harness for you.” They’re productizing the entire orchestration layer: sandboxed execution, context management, error recovery, permissions, long-running sessions. According to their launch post, companies like Notion, Rakuten, Asana, and Sentry are already building on it. And at $0.08 per agent runtime hour plus model usage, they’re making it very easy to get started. [another side note - check out the pricing. We’re getting closer and closer to compute cycle runtime pricing! exact text from their blog: “Managed Agents is priced on consumption. Standard Claude Platform token rates apply, plus $0.08 per session-hour for active runtime.”]

Anthropic is turning into more and more of a platform. Provide the model AND the infrastructure around it. Make it so easy to build on your platform that switching costs compound over time. It’s SO smart. However, how will developers feel? Using Anthropic’s harness will certianly “lock you in” to using their model more and more. Which is what you’d expect! But as a developer, do you want the model itself to be fungible? Swappable for the latest and greatest?

But here’s the tension, and this is where it gets interesting for founders. The entire Stanford paper is about how harness optimization is domain-specific. The generic harness is good. The tuned one is 6x better. So if you’re building a deeply vertical AI product, say an agent that handles insurance claims or manages clinical trials or runs a supply chain, a managed, general-purpose harness from Anthropic is going to be fine out of the box. But fine isn’t the same as great. And in competitive markets, the gap between fine and great is the whole ballgame.

This maps directly to the build vs. buy question I wrote about a while back. For foundational infrastructure (sandboxing, auth, session management), buy it. Use Managed Agents or something like it. But for the actual orchestration intelligence, what context to surface, when to retrieve it, how to handle domain-specific edge cases, that’s where you build. That’s where your differentiation lives. The founders who understand this distinction are going to build much better products than the ones who treat the harness as an afterthought.

Maybe wrappers weren’t such a joke after all!

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.0x

• Top 5 Median: 16.2x

• 10Y: 4.3%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 9.9x

• Mid Growth Median: 4.8x

• Low Growth Median: 2.3x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin 0%

• Median FCF Margin: 21%

• Median Net Retention: 109%

• Median CAC Payback: 33 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

Zero Knowledge, Maximum Trust

What a week for security breaches... Claude Code source code leaked via a misconfigured npm package, exposing 500,000 lines of code and an entire unreleased feature roadmap. Mercor got hit through a compromised LiteLLM dependency, with Lapsus$ claiming 4TB of stolen data including source code, databases, and contractor video interviews. And the axios npm package, one of the most widely used libraries in JavaScript with 100 million weekly downloads, was hijacked by North Korean state actors who injected a cross-platform remote access trojan. All within about 48 hours.

The common thread? Trust in the software supply chain (and soon to be agent supply chain…) is incredibly fragile. A single misconfigured file, a single compromised maintainer account, a single poisoned open-source dependency...and the whole thing unravels. And these are just the breaches we know about, affecting tools and infrastructure that developers interact with directly.

Now imagine a world where AI agents are running autonomously. Booking flights. Executing trades. Signing contracts. Moving money. Managing supply chains. We’re not far from that world...we’re basically in it. And yet the trust infrastructure underpinning all of it is...what exactly? We’re basically just taking AI’s word for it, and trusting that the background agent working on it’s own is not only doing it’s job accurately, but also not acting maliciously. That’s fine when you’re asking ChatGPT to summarize an article. It’s less fine when an agent is wiring $50k on your behalf. Or when a compromised model is silently making decisions with tampered weights.

If a single npm package can be weaponized to deliver malware to millions of machines in under 3 hours, what happens when the attack surface is AI models and autonomous agents operating across every critical system in an enterprise?

This is where zero knowledge proofs come in. And before your eyes glaze over (”isn’t that a crypto thing?”), bear with me, because I think this could be one of the most important infrastructure layers of the agentic era. I wrote this tweet earlier this year, and wanted to flesh out the idea a bit.

So what are zero knowledge proofs? At a high level, they’re a way for one party to prove to another party that something is true...without revealing the underlying information. The classic analogy: imagine you’re colorblind and I want to prove to you that two balls are different colors. I can design a game where you hide the balls behind your back, sometimes swap them, sometimes don’t, and ask me whether you swapped. If I can consistently tell you correctly, you become convinced the balls really are different colors, even though you never “saw” the colors yourself. That’s the essence of a zero knowledge proof. Proof of truth without revelation of the details.

In crypto, ZK proofs have been a massive deal. They’re used for things like privacy preserving transactions (proving you have enough funds without revealing your balance) and scaling blockchains through ZK rollups. But here’s the thing...ZK proofs have basically stayed inside the crypto ecosystem. They haven’t broken out into the broader tech world. Why?

Two reasons: speed and cost. Generating a zero knowledge proof is computationally expensive. Like, really expensive. Historically we’re talking orders of magnitude more compute than just running the original computation. If you want to prove that a simple ML model ran correctly, the proof generation might take 100x to 1,000,000x longer than just running the model itself. That’s...not great for real time applications.

This worked in crypto for a couple reasons. First, blockchain transactions don’t need to be instant. Users are accustomed to waiting. A few minutes (or even longer) for settlement is totally fine. Second, you can batch transactions together, amortizing the cost of proof generation across many transactions at once. ZK rollups do exactly this...bundle hundreds or thousands of transactions, generate one proof, verify it on chain. The per-transaction cost becomes manageable. And third, the computations being proved in crypto are relatively simple compared to something like neural network inference. Verifying a token transfer is a different beast than verifying a billion-parameter model.

But for AI? Where you need inference in milliseconds? Where agents are making decisions in real time? Where the computations are massive and complex? The overhead of traditional ZK proofs was a total non-starter.

Here’s what’s changed. And this is what has me really excited...

I’ve been seeing research breakthroughs in the ZK space that are dramatically breaking down these performance barriers. The overhead is shrinking fast. We’re talking improvements from 1,000,000x overhead to 100,000x to 10,000x...and the curve keeps bending. New frameworks can now prove the inference of image classification models in just a couple seconds. New proof systems using recursive SNARKs (called “folding schemes”) are compressing proof sizes from gigabytes down to under 100 kilobytes. GPU acceleration, specialized ZK hardware (yes, people are building dedicated ZK chips), and better algorithms are all converging at once.

We’re not at “real time proof generation for every AI inference” yet. But the trajectory is undeniable. And we’re close enough now that it works for a growing number of practical use cases.

So why does this matter? What does ZKML (zero knowledge machine learning) actually unlock?

Model integrity. How do you know the model you’re using hasn’t been tampered with? When you call an API from an AI provider, you’re trusting that they’re running the model they say they’re running, with the weights they say they’re using. ZK proofs can verify this cryptographically. The provider generates a proof that inference was performed using a specific, committed set of weights. No trust required...just math. This is enormous for regulated industries. A bank using AI for credit decisions could prove to regulators that the model used only approved parameters without revealing the proprietary model itself. A hospital could verify that an AI diagnosis came from an FDA-approved model without exposing patient data.

Input integrity. It’s not just about the model, it’s also about the inputs. Were the inputs to a model tampered with before inference? Did someone inject malicious data into the pipeline? ZK proofs can verify the full chain...that a specific input went into a specific model and produced a specific output. The entire computation is provable end to end.

Agent verification. This is the one I keep coming back to. In a world where agents are executing multi-step workflows autonomously (querying databases, calling APIs, moving money, signing contracts), we need a way to verify that the agent actually did what it said it did. That it followed the correct logic. That it wasn’t manipulated mid-execution through prompt injection or some other attack vector. ZK proofs can provide a cryptographic receipt for every action an agent takes. Every decision...provable, auditable, verifiable. Without having to re-run the entire computation or expose proprietary model details.

Privacy-preserving AI. Today if you want to use a cloud AI service, you typically send your data to the provider. They see your inputs. With ZKML, inference can be proven correct without revealing the input data OR the model weights to either party. The AI provider doesn’t see your sensitive data. You don’t see their proprietary model. But you both can verify the output is legitimate. This unlocks AI adoption in domains where data sensitivity has been a blocker...healthcare, finance, legal, defense.

Agent-to-agent trust. This might be the most forward looking use case, but I think it could define the next era. As we move toward a world where agents interact with other agents (your purchasing agent negotiating with a supplier’s sales agent, your portfolio agent coordinating with a market data agent), how do those agents trust each other? Today, agent-to-agent interactions rely on the same old assumption...trust the platform, trust the API, hope for the best. ZK proofs could give each agent the ability to cryptographically prove its identity, its logic, and its outputs to every other agent it interacts with. That’s a fundamentally new trust primitive.

The timing of all this is not a coincidence. NIST launched an AI Agent Standards Initiative in February specifically focused on security and interoperability for autonomous agents. Microsoft just unveiled their Zero Trust for AI framework. Everyone is converging on the same realization...we need better trust infrastructure for agents. Fast.

The way I think about it: every major platform shift has required a corresponding trust layer. The internet needed SSL/TLS. Mobile needed app store review and sandboxing. Cloud needed IAM and zero trust networking. The agentic era will need its own. And I think ZKML is a strong candidate for what that looks like.

Trust has always been the bottleneck for autonomy. The more we trust agents, the more autonomy we give them. The more autonomy they have, the more value they create. Zero knowledge proofs could be what unlocks that loop.

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.2x

• Top 5 Median: 16.4x

• 10Y: 4.3%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 10.0x

• Mid Growth Median: 5.5x

• Low Growth Median: 2.5x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin (1%)

• Median FCF Margin: 21%

• Median Net Retention: 109%

• Median CAC Payback: 33 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

From GPU Hours to Token Dollars

One thing I’m starting to believe - the companies who figure out pricing and packaging the fastest will have a big edge in the early days of this AI phase shift. I think it’s one of the hardest problems right now for any AI company! What makes pricing so difficult in an entirely new (and expensive) line item has entered COGS - inference. Whether you’re paying OpenAI / Anthropic directly, or paying someone else to run open source models, inference costs are exploding (and we’re just getting started….). A big question becomes - how can you price your product such that you don’t torpedo your business into perpetual negative gross margin land (or said more positively, how can you price your product to more tightly align with value delivered).

A couple weeks ago I wrote a post titled “Get in the Token Path” which described one way to better align pricing models (and a way I think will ultimately lead to the most success). I think it’s so important to figure out pricing because we’re entering a world where your pricing model is your business model. In traditional SaaS, pricing was important but also forgiving. You had 80%+ gross margins, so even if your packaging wasn’t perfect, there was a huge cushion. You could underprice a seat, overprovision features, give away usage for free and it didn’t matter because the incremental cost of serving another user was basically zero. That’s not the case anymore. When every agent action, every “magic” feature triggers real inference costs on the backend, the gap between your pricing model and your cost structure becomes existential. Price too low and you’re literally paying customers to use your product. Price too high and you lose to the competitor who figured it out first. And the costs themselves are a moving target. Token prices are falling fast, hardware efficiency improves every generation, and the mix of models you’re calling changes the math entirely. You can’t just set a $50/seat/month price and forget it.

But here’s the flip side (and I think this is underappreciated). The companies that do get pricing right have a chance to capture way more value per customer than was ever possible in the old SaaS world. If you can tightly align price with value delivered, charging more when your product does more, scaling with outcomes rather than headcount th,e ceiling on what you can extract per customer goes way, way up. A traditional SaaS company selling seats was always capped by “how many humans work at this company.” An AI-native company selling on usage, outcomes, or work completed? The TAM within each account is theoretically limitless. Getting pricing wrong is fatal. But getting it right is a superpower.

So back to pricing…I think we’re about to see an explosion of pricing models based on pricing per token. It may not literally be “per token” - I think we’ll see a lot of credit based pricing structures emerge where on the back end credits buy you tokens in a somewhat obfuscated way. More multi product companies are coming out, and companies are layering on additional products faster and faster than ever before. Pricing and packaging for multi-product companies is really hard (you want to capture value for incremental products delivered). But most of these will all have token generation as a commonality, and I think we’ll see a lot of credit based pricing models emerge where credits basically buy you a certain amount of consumption across any of the products offered.

Related to all of this - I think we’re also about to go through a shift in how GPUs are monetized as the world moves from a training heavy one to an inference heavy one. For the last decade, the world has rented GPUs by the hour. You go to AWS, CoreWeave, Lambda, whoever - and you pay something like $2-4 per GPU hour. That’s it. That’s the business model. You’re basically renting a very expensive piece of silicon by the clock, the same way you’d rent a U-Haul truck. Doesn’t matter if you’re driving it across the country fully loaded or it’s sitting in your driveway. You’re paying by the hour.

This model made sense for training. Training is a big batch job and more of a cost center - you spin up a cluster, you run the job for days or weeks, and you’re done. There isn’t really direct “revenue” tied to this cost (obviously there is as you start charging for model access, but the direct cost of training isn’t paired with revenue). Hourly rental works fine. But we’re not in a training-dominated world anymore. We’re rapidly shifting into an inference-dominated world, and that changes everything about how these GPUs should get priced.

Jensen Huang spent a huge chunk of his GTC keynote this month hammering this point home. He showed what he called a “Pareto frontier” - basically a chart that maps the tradeoff between throughput (how many total tokens you can pump out) and latency (how fast each individual user gets their response). The key insight is that depending on where you sit on that curve, the economic value of a GPU hour changes dramatically.

Let me make this concrete with some illustrative figures. Today, if you’re a GPU cloud provider, you’re renting out an H100 for roughly $2-4/hour (it can range more than that depending on what type of GPU with what specs, whether you’re buying on demand or on a committed deal, etc). That's your revenue per GPU hour. That's what the market pays. But now think about it from the other direction. A single GPU in the GB300 NVL72 generates roughly 15,000 tokens per second. The full rack — 72 GPUs working together — pushes that to over 1 million tokens per second, or roughly 4 billion tokens per hour. Now, renting that equivalent GPU capacity by the hour might run you $150-300/hour (72 GPUs at $2-4 each). But price those tokens instead. Even at rock-bottom commodity rates of $0.15-0.20 per million output tokens, 4 billion tokens generates $600-800/hour in token revenue — easily 2-4x the hourly rental value. And that's at the cheapest prices on the market. Mid-tier models charge $8-15 per million output tokens, which would push the math into the thousands per hour. The simple logic: a GPU rented by the hour is priced based on the cost of the silicon. A GPU monetized by the token is priced based on the value of the output. And it turns out the output is worth a lot more than the silicon.

Same GPU. Same hour. But when you price the output in tokens instead of clock time, the monetization more than doubles.

This is where Jensen’s Pareto chart becomes so powerful. On one end of the curve, you’re running maximum throughput - serving tons of users, batch workloads, cheap tokens. On the other end, you’re running low latency - fast, responsive, premium tokens for things like real-time agents. The further you push toward the premium end, the more revenue you extract per GPU hour. And with each generation of hardware pushing that Pareto frontier outward, the gap between “what you charge per hour” and “what you could charge per token” keeps getting wider.

So why does this matter for founders?

First, if you’re building on top of inference, you need to understand that the cost of tokens is going to keep falling (and fast). Every hardware generation pushes the frontier out further. NVIDIA’s own numbers show Vera Rubin delivering 5x the inference throughput of Blackwell, with token costs falling 10x. That’s deflationary pressure that will compound relentlessly. If your business model depends on tokens being expensive, you’re on the wrong side of this curve.

Second, and this is the more interesting part in my opinion. For the companies actually running the GPUs, this shift from hourly rental to token-based pricing is transformational. It turns GPUs from depreciating assets into revenue-generating factories. Under the old model, you buy a GPU, you rent it out by the hour, and you’re in a race against depreciation. The next chip comes out, your hourly rate drops, and you’re chasing a declining price (interestingly, H100 prices have actually been rising as the market remains so incredibly supply constrained). It’s a cost center. You’re basically a landlord watching your property value fall every 18 months.

Under the token model, the equation flips. You’re now selling output vs raw compute time. A token that helps an agent close a sales deal is worth a lot more than a token that summarizes a Wikipedia page, even though they cost roughly the same to generate. That’s the whole point of Jensen’s tiered token pricing vision - different points on the Pareto curve command different price points.

This is one reason why the Groq acquisition is so interesting. The Groq LPU is purpose-built for inference - ultra-low latency token generation. By integrating Groq into NVIDIA’s ecosystem (alongside their Dynamo inference orchestration software), they’re essentially building out the full stack to help GPU operators maximize token revenue, not just clock-time rental revenue. It extends the business model for everyone in the chain.

Think about the analogy to cloud computing. In the early days of AWS, you rented VMs by the hour. Then Lambda came along and you paid per function invocation. Then you started paying per API call, per request, per transaction. The pricing got more granular and more aligned with actual value delivered. The same thing is happening with GPUs. We’re moving from “pay per hour of silicon” to “pay per unit of intelligence produced.” And just like in cloud, the companies that figure out how to price on value (not just cost) will capture disproportionate economics.

One more thing. This shift has massive implications for the AI model providers too - the OpenAIs and Anthropics of the world. If you’re sitting on top of billions of dollars of GPU infrastructure and you’re selling tokens through APIs, you want to be on the high end of that Pareto curve. You want premium, low-latency, high-quality tokens that you can charge $10-15 per million output tokens for (not commodity batch tokens at $0.10). The model providers who can deliver the most valuable tokens (through better reasoning, faster responses, or more reliable agents) will extract the most revenue per GPU hour from their infrastructure. This is actually what’s going to drive these companies to profitability - not just scale (which will help!), but token monetization efficiency.

The world is moving from GPU hours to token dollars. For founders building in this ecosystem, the takeaway is this: understand where you sit on the Pareto curve, because that’s what determines your economics. Are you building for throughput (cheap, high-volume tokens)? Or for latency (premium, real-time tokens)? The hardware is getting better at both, but the pricing power lives at the premium end. And the companies that figure out how to monetize tokens based on value (not just pass through the cost of compute) are the ones that will have a big advantage.

And this is where the credit-based pricing model I mentioned earlier becomes so interesting. Credits give you a flexible abstraction layer that lets you sit on top of all of this complexity without exposing it directly to the customer. Token costs are falling, model mix is shifting, you’re running five different models at five different price points on the backend. The customer doesn’t need to know any of that. They buy credits, credits buy them outcomes across your product suite, and you manage the token economics underneath. It’s the same playbook the cloud providers eventually landed on (committed spend across a broad set of services), and I think it’s where the best AI-native companies will land too. The ones who figure out that abstraction, pricing on value and managing token costs as an internal optimization problem rather than a customer-facing one, will build the most durable businesses of this next era.

I had a fun time putting this post together! I’d love to speak with experts in the AI pricing and packaging space. Maybe even host a town hall with people who want to learn about pricing and packaging in the age of AI. Hit me up if you’d like to participate!

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.1x

• Top 5 Median: 16.1x

• 10Y: 4.4%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 9.7x

• Mid Growth Median: 5.5x

• Low Growth Median: 2.5x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin (0%)

• Median FCF Margin: 19%

• Median Net Retention: 109%

• Median CAC Payback: 33 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

Digital Twins

Every week I meet with founders building in the agent space. And lately, I keep hearing the same concept come up over and over - digital twins (or some version of this). When a concept starts showing up as frequently as this one, my ears generally perk up. Digital twins are the thing perking up my ears! And I think they’re about to become one of the most important concepts in AI. I think they could become a layer that helps scales AI to the masses (and consumption of AI).

So what actually is a digital twin? The term originally comes from manufacturing. You’d build a digital replica of a physical asset (a jet engine, a factory floor) to simulate and monitor it. With AI it’s the same core concept, but with a totally new application. In the AI era, a digital twin is just representing knowledge (from any source, in any form) digitally, so an agent can act on it. That knowledge could live in a person’s head, across a dozen siloed systems, in years of company history, or in the collective behavior of your customers. The twin is just the bridge between that knowledge and the agent that needs it to do work.

There are a couple main flavors of digital twins that I wanted to highlight here. If you’re building anything in the digital twin space, I’d love to talk to you!

1. Knowledge capture (workflows)

2. Institutional memory (knowledge retention)

3. The expert twin (scaling your best performers)

4. The customer twin (queryable customer knowledge, anytime)

5. Knowledge multiplication (1-to-many)

6. Personal monetization

The most obvious one is workflow knowledge capture. Think about something as simple as quote-to-cash, or sending a contract, or onboarding a new customer. Someone on your team knows exactly how that process works (and more importantly what to do in edge cases, when to ask for approval when something doesn’t look right, etc). They know which systems to pull from, whose approval you need, what order things happen in. But nowhere is it written down. It lives entirely in that person’s head because they’ve done it a thousand times. Until you can represent that knowledge digitally and hand it to an agent, the agent is going to keep getting stuck. Digital twins are how you fix that. I saw a cool launch from a company called Edra this week that reminded me of this concept. To automate workflows, we first need to understand what the workflow is!

Closely related is institutional memory. When a key employee leaves, their knowledge walks out the door with them. We’ve all felt that, and it can be quite painful depending on the employee. Digital twins give you a way to preserve not just the process documentation, but the actual judgment and pattern recognition that made that person valuable. You’re capturing the what, why and“when it gets weird, here’s how to handle it” knowledge.

Then there’s what I’d call the expert twin. Every company has a power law distribution of talent. There’s the sales rep who always finds a way to close, the SOC analyst who knows how to triage alerts in their sleep, the on-call SRE who’s seen every outage pattern and knows exactly what to pull. These people are incredible, but are also massive bottlenecks. There’s only one of them. A digital twin of your best performer will help you build better agents and raise the floor for everyone else. The new hire gets trained by the twin. The average rep gets real-time coaching from the twin. You stop hoping everyone eventually gets as good as your best person, and you just... give everyone access to that expertise directly.

The customer twin is another one I’m seeing more of. Instead of running a one-off survey or scheduling a round of user interviews every time you want to test a hypothesis, you build a persistent digital representation of your customer base that you can query anytime. What would our ICP think of this new feature? How would our mid-market segment respond to this pricing change? Ask the twin. Companies like Simile and Aru are already building in this direction (more for market research then b2b customer research) - the idea being you do the research once, build the baseline, and then run as many studies as you want against the digital version rather than constantly recruiting new participants.

That’s a good example of a broader pattern I’d call knowledge multiplication. Taking something that used to be 1-to-1 and making it 1-to-many. One survey becomes unlimited studies. One expert becomes a resource available to the entire company. Which brings me to the most interesting flavor of all of this.

What if you’re the expert?

If you’re a graphic designer with a distinct style and aesthetic, you used to be capped by time. You could only take on so many projects. You had to turn clients away. Now imagine you build a digital twin of yourself (your taste, your process, your design sensibility) and others can hire “you” without needing calendar time with the real you. Same thing for an executive coach. You’ve spent years developing a methodology and a point of view that genuinely helps people. Historically, you could only serve so many clients. A digital twin changes that entirely. You go from a time-constrained practice to an infinitely scalable one.

This is where I think the job displacement narrative gets it wrong. Everyone asks “will AI take my job?” But the better question is “can I build a digital twin of myself before someone else does it for me?” The people who win in this world are generally the ones who move fastest to adopt new technologies. With AI, this could mean they’re the ones who figure out how to package and distribute their own knowledge and taste at scale. I’ll always remember my couple friends in high school who learned how to program iPhone apps before the app store got bloated. Their early apps printed money! (at least for a high schoolers standard).

The throughline across all of these is the same: the bottleneck to the agentic era isn’t model intelligence. The models are already good enough. The bottleneck is knowledge representation. And specifically, how do we represent that knowledge digitally. Agents can only act on knowledge they have access to. And right now, most of the world’s most valuable knowledge is locked in people’s heads, scattered across systems, or sitting undocumented in the institutional memory of companies. Digital twins are how you unlock it. That’s why I keep hearing about this concept in meeting after meeting. And it’s why I think we’re just getting started.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.3x

• Top 5 Median: 17.7x

• 10Y: 4.3%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 10.4x

• Mid Growth Median: 5.9x

• Low Growth Median: 2.7x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin (0%)

• Median FCF Margin: 20%

• Median Net Retention: 109%

• Median CAC Payback: 33 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

In Defense of Model Lab Profitability

There seems to be endless debate around AI companies, and whether they have “upside down P&Ls” that will forever lose money, or if they will turn into cash cows in the future. Whether this sentiment is pointed at the large labs like OpenAI and Anthropic, or upstarts like Cursor, I hear it all the time! And I can’t tell if the bears just want to confirm their priors on AI negativity, if the bulls just have blind naive optimism, or if anyone really has a pov grounded in real analysis. As an early stage VC I certainly fall into the “perpetually optimistic” camp, so you can apply the appropriate filter to this post :) But for this post I wanted to focus on the profitability debate centered around the large labs, and why I think they’ll turn into wildly profitable business.

There’s three main parts to this debate. 1) gross margins, 2) training costs, and 3) retention / commoditization. The “AI labs will never be profitable” crowd will claim gross margins of labs are structurally capped, the hamster wheel of training larger and larger models (whether that’s a capex or opex expense) will always lead to FCF negative businesses, and the switching costs from one model to the next are super low. So (as the bears would say), even if you can solve one of those 3 issues, you’ll never solve all three, and thus these businesses will never be profitable.

To start, I want to set the stage by analyzing how the profitability of infrastructure software companies (like Confluent, Snowflake, Mongo, etc as well as the hyperscalers) have evolved over the years. It’s an imperfect analogy for so many reasons, but I think it’s helpful to lead with it.

Let’s start with gross margins, as this is the part of the debate I’m confident on (one could describe my views as “naively certain” :). Just about every infra software company started off with negative gross margins, that then scaled to something in the 60-70% range with scale. Why do infra companies typically start of with negative / very low gross margins?

• You’re buying infrastructure at retail and selling at a discount. Early on you have no scale with your cloud provider. You’re paying list price (or close to it) for compute/storage, but you’re pricing your product aggressively to win customers. The math is upside down by design, you’re subsidizing adoption.

• Utilization is terrible. You’ve provisioned capacity for customers you don’t have yet. Your clusters are running at 10-20% utilization but you’re paying for 100%. Every incremental customer improves the ratio, but early on you’re eating dead capacity.

• Fixed costs of launching new cloud regions. Early on, you may only make your product available on one cloud, in one cloud region. “AWS East.” over time, you add more clouds, and more regions. There’s a real fixed cost to adding new regions (and even more adding more clouds). Early on, you are amortizing these costs over a very small number of customers. As your customer base grows, you can spread those annual fixed costs across a larger cross section of customers.

• The product isn’t optimized yet. Early infra products are architected for speed-to-market, not efficiency. Queries are wasteful, storage isn’t tiered properly, caching layers don’t exist yet. Engineering is focused on features, not cost optimization. The “make it work, then make it fast” cycle means COGS is bloated in the early years.

There are more, but these were the first couple I could think of. And what happens? Every company figures all of these out, and the gross margins scale predictably…If there’s one thing I’m almost never worried about when evaluating early stage infra companies, it’s their gross margins…

Now let’s look at the model companies. A huge knock early on was “their gross margins suck! They’ll never be profitable!” But these model companies face a lot of the same challenges classic infra companies did, and they’re already showing a similar trend of expanding. The Information recently reported Anthropic had -94% gross margins in 2024, and a target of ~40% gross margins in 2025. I won’t comment on the validity of those figures, but the trendline should be very clear…How have the labs been able to expand margins like this? Their model architectures improve and serving becomes more efficient (think adding things like prompt caching). How efficiently they auto-scale and utilize GPUs in the background gets more efficient. Overall, the inference cost per token drops dramatically as architectures get more efficient. I’d argue the marginal inference call is already quite gross margin profitable. My partner Clark posted an analysis related to this you can find here. I also think there’s a good chance the labs offer higher gross margins products down the line (like applications, ads, etc). I’m really not worried about gross margins being structurally lower for the labs then classic infra companies of the cloud era.

The second debate around “these will never be profitable companies” centers around training costs. Cloud infra companies don’t really have a parallel to analyze here. However, the real question (in my opinion) is more about how long does it take to payback the training costs. For that, we do have an analogy for the cloud infra companies - their S&M payback (or CAC payback as I like to call it). It’s pretty well understood what great, good, bad looks like when it comes to CAC payback, and what’s sustainable from a “profitability” standpoint. To define the metric, CAC payback is calculated as: “if you spent $x to acquire a customer, how many months does it take to generate $x of cumulative gross profits from that customer.” Anything less than 12 months is fantastic, 12-24 months is good, 24-36 months is ok, and >36 months gets dicey (these are very general rules of thumb I usually apply to earlier stage companies. As you get larger and larger these metrics aren’t quite as relevant). Companies with <24 months of CAC payback have shown the ability to be wildly FCF profitable. The reason this metric is relevant is because after you’ve paid back the cost to acquire, all future gross profits from that customer should largely flow straight to the bottom line (again, a broad strokes comment that isn’t entirely accurate, but directionally is).

So the right comparison to the labs is “what is your pre-training payback period” and more importantly, how is that metric trending as you release future models. The way to calculate “training payback period” would be to look at the fully loaded cost to train a model, and then calculate how long it takes to generate an equivalent amount of gross profits. Without getting into private data, I can say the dynamics of the labs training payback period doesn’t look that different from infra companies CAC payback periods. I have all the comfort in the world that the training payback periods are not that long at all.

Two pushbacks to this.

1. CAC payback for infra co’s is a one time thing. You spend once to acquire a customer, then farm the revenue. Model companies don’t just spend once to pre-train a model, they do it repeatedly, over and over again. So the real question is can they pay back the training costs (and then generate profits) BEFORE they get to their next model. This is the strongest bear argument in my opinion, so it’s worth digging into in a little more depth.

   First — people dramatically underestimate the commercial lifespan of a model. There’s this perception that labs ship a new model every few months and the old one immediately becomes worthless. That’s not how it works. Sonnet 3.5 was Anthropic’s workhorse for well over a year. GPT-4 was the backbone of OpenAI’s revenue for ages. And these models don’t get “replaced” — they get layered. New frontier model comes out at a premium, old model slides down to serve the high-volume cost-sensitive tier. The revenue from the older model “shifts” way more than it just “disappears.” So the window to payback training costs is longer than the bears assume.

   Second — revenue growth between model generations has been significantly outpacing training cost growth. Yes, training costs are going up something like 3-5x per generation. But the labs have been roughly doubling (or more) annually, and each new model unlocks entirely new use cases and customer segments. Said another way, if it takes you 6 months to pay back Model A’s training cost on a $1B revenue base, and Model B costs 4x more to train but you’re on a $3B revenue base… the payback period actually shrinks. The ratio is getting better, not worse.

   Third — training efficiency itself is improving. The labs aren’t just throwing more compute at the same problem. Better data curation, synthetic data, architectural improvements, all of these reduce the FLOPs needed to hit a given capability level. So “4-5x more expensive each generation” is probably overstated if you measure cost per unit of capability rather than raw spend.

   Longer model lifespans than people think, revenue growth outpacing training cost growth, and improving training efficiency. I have a hard time seeing training payback periods become unsustainable. If anything, the data suggests they’re getting better over time, not worse.

2. Well, great, the training payback costs aren’t crazy, but the model companies still have to pay back CAC!! I’d respond to this by arguing that model labs will have structurally lower CAC payback periods because their market is more of an oligopoly vs a super crowded SaaS market. That is maybe the part of my analysis that will garner the most push back… There used to be many more large lab competitors. The dust has settled, and there’s really 3 that stand out (that are in the business of selling inference tokens themselves). OpenAI, Anthropic and Gemini from Google. I think the CAC payback period of labs will look more like CAC payback of the hyperscalers given they both look like oligopolies. In time, neo-labs will be real challengers. But I think the neo-labs wont structurally change the payback periods of the large labs.

The obvious pushback to the oligopoly framing is open source. Llama, DeepSeek, Mistral. These models are good and getting better, and they're basically free. Don't they structurally cap the labs' pricing power? I don't think so, for a few reasons. Open source models are great for experimentation and for use cases where "good enough" is fine. But enterprise customers overwhelmingly want a vendor, they want SLAs, they want support, they want someone to call when things break, they want compliance certifications, and they want a roadmap they don't have to build themselves. Running and serving open source models at scale is not free. You still need the infra, the ops team, the fine-tuning pipeline, all of it. By the time you've built all of that, you've basically rebuilt a worse version of what the labs already offer. Open source keeps the labs honest on pricing (which is a good thing), but I don't think it fundamentally breaks the profitability thesis. The thing for the labs to be “worried” about. I do see a pattern where companies will use the labs models for quick prototyping, and then look for an open source (cheaper) model for the larger deployment. This could be a canary in the coal mine for my thesis.

And finally, the “labs will never be profitable” crowd argue the switching costs are zero, retention will be all over the place, and eventually the models will commoditize themselves by offering the same features. First (to get it out of the way), people have been saying that about the hyperscalers (AWS, Azure, GCP) forever….And guess what, they’re all wildly profitable DESPITE massive price cuts along the way. I wrote about this a few years ago here. The hyperscalers have added a structural layer of lock-in the labs don’t benefit from - egress fees! Despite not having that tool, I think the switching costs for labs are building faster than people realize, and will be much higher than the bears think within a couple years.

Think about what a serious enterprise deployment on a lab actually looks like today. You’ve got fine-tuned models trained on your proprietary data that don’t transfer to another provider. You’ve got eval suites built around a specific model’s behavior and quirks. Your engineering team has built muscle memory around a specific API, specific prompting patterns, specific tool-use conventions. You’ve got system prompts that have been iterated on for months. You’ve probably got committed spend contracts with negotiated pricing. And increasingly you’ve got features like memory, context windows, and agent tool ecosystems that are deeply provider-specific. None of that stuff ports cleanly.

And it’s only going to get stickier. As the labs move further into platform territory — hosting agents, managing long-running workflows, storing persistent context, offering fine-tuning and retrieval infrastructure - switching starts to look less like “swap out an API key” and more like “migrate off a cloud provider.” That’s a very different conversation. The model itself might be commoditizing (and honestly, for a lot of use cases, the frontier models are pretty interchangeable). But the platform around the model is not. And that’s where retention gets built.

The other thing I’d point out - the market structure itself supports retention. This is basically an oligopoly. Three serious players (OpenAI, Anthropic, Google), and the gap between them and everyone else is meaningful (and growing). In an oligopoly, you don’t see the aggressive price wars and customer churn you see in fragmented markets. You see stable market share, rational pricing, and high retention. The hyperscalers are the perfect analogy here - AWS, Azure, and GCP compete intensely, but none of them have low retention. It’s the same dynamic.

So in summary, why do I believe the labs will all be quite profitable?

1. I believe their gross margins will (at maturity) look a lot more like classic infra software gross margins. And the data supports this

2. I believe the training payback and CAC payback periods of the labs are sustainable, and ultimately the marginal gross profit dollar will be >> the marginal training & CAC dollar. And the data supports this.

3. I believe the labs will have much higher retention than the bears think, and simultaneously won’t commoditize.

Ultimately, none of this matters if the labs themselves can’t stay at the frontier. If they fail at that, nothing else in this post matters. However, IF you believe they will, I think there’s a very clear path to very profitable businesses.

**Disclaimer, Altimeter is an investor in both OpenAI and Anthropic**

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.3x

• Top 5 Median: 18.2x

• 10Y: 4.3%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 10.8x

• Mid Growth Median: 6.4x

• Low Growth Median: 2.6x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 75%

• Median Operating Margin (1%)

• Median FCF Margin: 20%

• Median Net Retention: 109%

• Median CAC Payback: 34 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

Get in the Token Path

As always, these posts are more of a brain dump of “what I’m thinking” about…And lately I’ve been thinking about a pattern that keeps showing up when I study the biggest infrastructure winners of the cloud era, and what it means for AI companies today.

Here’s the general idea: the biggest infrastructure winners of the cloud era monetized the core consumption primitive of the platform. In the cloud era, that primitive was compute, storage, and network I/O. In the AI era, it increasingly looks like tokens.

Let’s unpack.

When cloud computing first started taking off, the core primitive of the platform was very clear: compute. Everything that happened in the cloud ultimately boiled down to compute cycles running somewhere inside a data center. Storage, networking, and databases all mattered of course, but the engine driving the system was compute.

That had an interesting downstream effect. The companies that ultimately became the biggest infrastructure businesses of the cloud era found ways to align their revenue directly with compute activity (or they charged directly for compute). They owned the meter.

AWS and the other hyperscalers obviously did this, their business’ are literally selling compute hours. The more workloads move to the cloud, the more compute gets consumed, the more AWS and the hyperscalers made. But it wasn’t just the cloud providers.

Let’s look at some of the infra leaders of the cloud buildout.

Databricks monetizes job compute. Every time a customer runs a data pipeline, trains a model, or processes a workload, Databricks' revenue grows automatically. Snowflake monetizes query compute, similar story. Every new query, every new dataset, every new workload meant more revenue without having to sell a single new seat. Datadog monetizes telemetry generated by compute workloads. Every new microservice, every new container, every new cloud instance generates incremental Datadog revenue. More cloud compute = more Datadog revenue. Cloudflare monetizes requests generated by applications running on compute. MongoDB charges based on storage and compute consumed through Atlas.

The details vary, but the pattern is remarkably consistent. The biggest companies ended up sitting directly in the execution path of workloads, with pricing models that scaled automatically as compute activity increased (and compute was one of the core primitives of the cloud buildout).

And this is the key insight - these companies didn’t just have “consumption pricing.” Lots of companies have consumption pricing and grow slowly. What made these companies special is that their consumption unit was the same unit the entire ecosystem was scaling. When the world spun up more compute, these companies grew without doing anything. Their revenue was structurally coupled to the platform’s growth vector.

That might seem obvious today, but during the early years of the cloud buildout many infrastructure companies were still trying to monetize software the old way. Perpetual licenses, term licenses, maintenance contracts, support subscriptions layered on top of open source software. These models worked well in the on-premise world where infrastructure growth was slow, predictable, and tightly controlled.

But cloud computing fundamentally changed the underlying economics. Workloads could scale instantly. Compute consumption could grow by orders of magnitude. The companies that adapted to this new world quickly built the biggest outcomes. The ones that didn’t…well, the contrast is striking.

Docker might be the most instructive example. Docker was containerization. They were one in the same. It was the technology that made cloud-native development possible. Millions of developers used it. Arguably the most important developer tool of the cloud era. But Docker never figured out how to monetize the primitive. They couldn’t connect their massive developer adoption to the underlying compute spend that containers enabled. Kubernetes (open sourced by Google) ate their orchestration business, and every hyperscaler ended up monetizing Docker’s own innovation through managed container services. Docker enabled billions of dollars in compute spend…they just never captured any of it (they have been doing a much better job over the last few years, however. This commentary is more geared towards their origins).

The common thread across Docker and others who gained massive adoption but ran into some sort of business model wall was similar. Most were deeply embedded in the cloud infrastructure stack and were critical tooling. But when they failed, it was often because they didn’t figure out how to make their revenue a derivative of the core consumption primitive. They monetized adjacently, through seats, support contracts, consulting, and the market rewarded them accordingly. Or rather, didn’t.

Compare these outcomes to Snowflake, Datadog, and Cloudflare. Same era. Same underlying platform. But radically different business models, and radically different outcomes. The difference? The winners owned the meter. They put themselves directly in the path of the underlying compute primitive.

Now Map this to AI

If cloud infrastructure was built on the primitive of compute, AI infrastructure is being built on a different primitive: tokens.

Every AI workload ultimately boils down to tokens being generated, processed, and consumed by models. Prompts become tokens. Context becomes tokens. Responses become tokens. Agents running multi-step workflows can generate enormous volumes of tokens as they reason through tasks. Tokens are the atomic unit of work in modern AI systems.

And once you start looking at the ecosystem through this lens, a very familiar pattern emerges.

The model providers like OpenAI and Anthropic - they literally are the token primitive (like the hyperscalers were the compute / storage primitive for the cloud buildout). They charge per token in, per token out.

But it’s not just the model providers. The fastest-ramping AI companies today are the ones sitting directly in the token path.

Coding agents are the standout. Cursor reportedly hit $2B ARR recently according to online reports. Every keystroke, every code completion, every agent action triggers inference, and their business model has evolved from simply charging per seats (these seats now come with usage limits!). Their revenue is structurally coupled to token consumption.

Companies in the business of inference more directly (Inferact, Baseten, Fireworks, Together, etc) are essentially selling the raw primitive.

The companies that sit closest to the generation and consumption of tokens are the ones whose revenue naturally scales with AI activity. Meanwhile, other parts of the AI ecosystem are experimenting with more traditional SaaS pricing models, seat-based developer tools, platform subscriptions, enterprise licenses layered on top of open source frameworks.

Those businesses may still succeed. Many developer tools companies built valuable franchises during the cloud era. But if history is any guide, the biggest infrastructure companies tend to emerge where the core unit of platform activity is measured and monetized. And we already have a pretty clear set of counterexamples from the cloud era showing what happens when you don’t.

Now, being in the token path is necessary but not sufficient. This is an important nuance.

The pure-play CDN companies of the cloud era were technically “in the compute path.” They charged based on bandwidth and requests. Traffic was exploding. They should have been massive winners. But bandwidth turned out to be a commodity. Prices compressed relentlessly. Limelight Networks had record traffic during the streaming boom of 2020-2021 and declining revenue at the same time. They eventually rebranded (to Edgio) and went bankrupt. Meanwhile Cloudflare, which started in a similar spot, layered on security, developer tools, and edge compute, building real differentiation and switching costs on top of the primitive. Same starting point, radically different outcomes.

The lesson for AI founders: get in the token path, but build something differentiated on top of it. Don’t just be a pipe that tokens flow through. Be the layer that makes those tokens more valuable, whether that’s through better developer experience (Cursor), specialized vertical models, security and compliance tooling, or proprietary data moats.

There’s also a timing dimension. In the cloud era, the companies that established themselves as defaults early in the compute path captured the most value. Datadog, Snowflake, and Cloudflare all got to scale before the primitives became fully commoditized. The implication: the window to get into the token path is now. Inference costs are dropping rapidly (which is great, it means more tokens consumed, but it also means per-unit economics compress over time). You need to be in the path and build a moat before that happens.

The biggest infrastructure winners of the cloud era monetized the core consumption primitive. The ones that didn’t, even the ones with massive adoption, beloved products, and deep integration into the stack, produced dramatically smaller outcomes.

The primitive has changed. It’s no longer compute cycles. It’s tokens. And if the pattern holds, the most valuable AI infrastructure companies of the next decade will be the ones that find a way to get themselves directly into the token path.

If you own the meter, growth takes care of itself.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.5x

• Top 5 Median: 17.7x

• 10Y: 4.1%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 10.4x

• Mid Growth Median: 6.5x

• Low Growth Median: 2.7x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 13%

• Median LTM growth rate: 15%

• Median Gross Margin: 75%

• Median Operating Margin (1%)

• Median FCF Margin: 20%

• Median Net Retention: 109%

• Median CAC Payback: 34 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

The Poison of Inertia

By now I’m sure everyone has seen Jack Dorsey’s tweet. If not, you can find it here. For those who haven’t seen it (or don’t care to read it), he announced a ~40% headcount reduction at Block (formerly Square). This is a massive move… You rarely see headcount reductions this large. Throughout the post he used the word “intelligence” - which really can be replaced with “AI.”

“we're already seeing that the intelligence tools we’re creating and using, paired with smaller and flatter teams, are enabling a new way of working which fundamentally changes what it means to build and run a company. and that's accelerating rapidly”

and later:

“we're going to build this company with intelligence at the core of everything we do.”

I’ve broadly seen two different reactions to this post:

1. AI is coming and it will take all our jobs!!

2. Block massively inflated their headcount in the COVID period, never reverted, was running hyper inefficiently, and is now unwinding the inefficiencies.

Most will want to ignore #1 and blame the large RIF on #2, but the reality is it’s a bit of both (and kind of a related #3). The related #3:

3. Large organizations are riddled with inertia. And during a tectonic platform shift (none bigger than what we’re going through with AI), that inertia can become fatal, beginning with the innovator’s dilemma and ending in irrelevance. The only way to avoid this inertia is to forcefully reduce it.

In the past, I’ve compared the rise of AI to the rise of the internet, using GDS platforms like Sabre as an example of how incumbents were reshaped by that shift. Article here. To summarize - the internet “squeezed” the value the GDS systems captured and OTAs (Priceline, Booking, etc) took the majority of the incremental profits earned. With the help of some ChatGPT research, GDS systems used to make ~$20 / booking. Post internet (and rise of OTAs) that dropped >10x.

The natural question - why didn’t Sabre and other GDS systems end up owning the OTA layer? Surely they saw the internet coming, and understood the risk of someone sitting on top of the GDS layer and capturing their value? Why didn’t they innovate?

The short answer is of course they tried! The more interesting part is how they tried. Instead of setting up their own OTA inside of Sabre, they created an entirely new entity - Travelocity! New brand, that was born technically inside of Sabre, but lived as it’s own thing. In 2000 Travelocity was carved out entirely and went public as it’s own standalone entity. So why did Sabre structure it this way? Why not just make the “Sabre OTA?” Because of inertia! (I say that definitively, but of course I don’t actually know, this is just my speculation)…

Simply “adopting” a new platform technology (in this case the internet) wasn’t the hardest part for an incumbent to do. It was reversing the inertia that existed everywhere in the business. The only way to fight that inertia was to start something new untethered by the existing org. Business models changed (and therefore sales commission structures needed to change). Marketing shifted from trade relationships and agency channels to consumer acquisition and performance marketing. Distribution moved from proprietary terminals and long term contracts to open web traffic and price comparison. Product development had to become software driven and release cycles accelerated from years to weeks. Even the culture had to evolve, from protecting existing margins to being willing to cannibalize them.

There are tons of other examples I could give - but to pull out a common thread it’s this: the playbooks in the old world had to be rewritten for the new world. And it’s realllllly hard for people to rewrite playbooks. Most can’t. Sometimes the only way to do it is to start fresh.

I think the same thing is playing out now in software. Playbooks are being rewritten. Take something like developer relations. How can you create broad developer love. Create word of mouth virality that leads to strong product lead growth (PLG). Well, if agents become the customer, you now need agent to agent signaling not “word of mouth.” You’re now marketing to a new audience. Hiring structures change. For an enterprise sales driven company, maybe you used to have 2 SDRs for every 1 AE. Well, in a world of agentic AI, maybe you have 100 agentic SDRs for every 1 human SDR, who supports 10 AEs (and then continue this loop once AEs become more automated!)

No longer can you just hire the sales exec who “took company X from $10m to $100m in revenue because we’re now at $10m and we want him to help us scale like he did company X.” That exec knows the old playbooks, not the new ones. This doesn’t mean he can’t adapt, but you really have to evaluate him critically.

So back to Block and their ~40% RIF. Sometimes the only way to transition into the new world is to remove the “old way group think.” Force new ways of thinking by bringing in fresh blood. The natural question - is 40% enough, or should it have been larger?

Every large incumbent SaaS company should consider a similar cut. It sounds harsh, but a) you probably have bloat and inefficiency EVERYWHERE, and b) It’s going to be very hard for you to move quickly with all the inertia that exists.

And I’ll end with this - large technological shifts often come with huge changes to cost structures. A crude analogy I love is looking at a world before mass production existed. If you wanted to buy a dining room table, someone had to hand craft it for you, Carving the wood by hand, 3 layers of paint by hand, etc. Because of the effort required, the merchant had to charge a price that compensated them for the time / effort to produce. Then mass production came along - that same table could be made for a fraction of the cost, and therefore sold at a fraction of the cost. The old artisanal craft vendor simply couldn’t compete on price, because the more modern vendor was structurally able to produce the unit at a lower cost, and therefore sell it at a lower cost. If the artisanal craft vendor matched the new market price, they’d go out of business.

Something similar will happen in software - modern AI vendors will be able to produce software at a materially lower cost. Therefore allowing them to sell it at a materially lower cost. If the incumbents want to compete, they will have to materially lower their cost basis. There’s really only one way to do this…They should all get ahead of this before it’s too late.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.3x

• Top 5 Median: 16.6x

• 10Y: 4.0%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 10.3x

• Mid Growth Median: 6.3x

• Low Growth Median: 2.6x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 12%

• Median LTM growth rate: 15%

• Median Gross Margin: 76%

• Median Operating Margin (1%)

• Median FCF Margin: 20%

• Median Net Retention: 109%

• Median CAC Payback: 34 months

• Median S&M % Revenue: 35%

• Median R&D % Revenue: 24%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

The SSD / Memory Reckoning

Memory stocks have taken over recently. If the early AI “trade” was compute, the current trade is memory! Over the last year:

• SK Hynix is up >300%

• Samsung is up >200%

• Kioxia is up ~1,000%

• Sandisk is up >1,200%

• Micron is up >300%

• Western Digital is up >400%

This is by no means an exhaustive list of memory related stocks, but it should give you a flavor of what’s happening in the stock market for memory related companies. So what’s happening? Why are memory stocks having a moment? I asked David Flynn, founder and CEO of Hammerspace to share his perspective. The below post was written together by David and myself. David is a repeat founder and long-time storage/memory architect who built Fusion-io, the company that put flash on the server fast path and helped change the industry’s performance playbook (later acquired by SanDisk). He also helped invent the architectural precursor to NVMe, so his perspective on why memory stocks are ripping is grounded in first-hand experience building the systems that made this era possible, and it’s only getting more critical as AI evolves and memory bandwidth, latency, and supply become the limiting factors for scaling compute.

Let’s dig in!

Most of the conversation around AI infrastructure centered on GPUs over the last two years. That’s understandable; compute was the obvious driver of AI innovation and in a very short period of time has had a profound global economic effect. GPU supply has begun to normalize relative to two years ago (but still remains quite constrained!), however a new constraint has emerged that should have been foreseen and is potentially longer lasting: memory and storage due to global shortage in NAND flash memory supply. In many ways, memory and solid state storage now feels like GPUs did two years ago.

From Compute Bound to Data Bound

The memory and storage markets have exploded recently, but why? Early AI was training heavy. What does this mean? You loaded datasets, ran repeated passes, and optimized weights. It was compute bound and AI innovators were racing to deploy more GPUs, as fast as possible. Once the data was staged, GPUs did most of the work (and you hammered those GPUs with data…). Of course, over time, as training runs got larger, we added more and more data to the runs. However, inference is different. It’s more interactive / concurrent / session based. Models are pulling from large, diverse datasets, often stored in different locations, in real time. That makes the system increasingly data bound. It is not practical to colocate GPUs in every location data is generated and stored, the data needs to be “known to exist” and moveable to place it with a GPU when it is needed for a processing job. Results depend less on raw compute and more on how efficiently data can move to the model (from wherever it resides). Manual data identification, preparation, and management is too slow and cumbersome. And legacy storage systems are inefficient at this type of workload (and rely on overprovisioning both memory and solid state disk drives (SSDs)). Optimal AI architectures require all data to be able to be delivered at high performance to feed the GPUs for inference faster.

But there’s another accelerant: AI is now generating more data than it consumes. Inference is producing new artifacts continuously -synthetic training data, augmented datasets, embeddings and vector indexes for retrieval, logs/telemetry for evaluation and safety, and agent outputs that get stored, versioned, and re-used. Even when the underlying “source” data doesn’t change, AI pipelines create multiple derived representations (chunks, summaries, features, indexes) that have to be stored somewhere and be refreshed as models evolve. The result is a compounding data footprint: more reads and more writes, more metadata, more movement, and more capacity needing to be high performance.

Now add the KV cache (which is stored in memory). Every time a model processes a prompt, it stores intermediate attention state so it can generate the next tokens efficiently without recomputing prior context. What does this mean to an AI architect? Longer prompts? Bigger KV cache. More concurrent users? Bigger KV cache. Larger context windows? Bigger KV cache. Longer running agents & tasks? Bigger KV cache. All of these trends are pushing the limits of the KV cache (and creating incredible demand for memory). Unlike model weights, which are fixed, this memory footprint scales dynamically with usage. Memory demand rises meaningfully! That memory typically lives in GPU memory first, then spills into host DRAM and fast storage as systems scale. So inference does not just consume compute. It consumes memory across the entire hierarchy (from storage nodes to controller nodes to caching tiers to GPU nodes).

The more AI gets used, the more memory the system requires.

The Memory Crisis

The increased use of data that already exists, coupled with an increased amount of data generated and stored has created a global NAND shortage. NAND is used in memory and SSDs used for high-performance data storage. The NAND shortage isn’t just a typical quarterly fluctuation, it’s quickly becoming the underlying driver of a broader memory crisis. We’re coming off a massive downturn where manufacturers slashed production and returned to extreme capital discipline only to have AI demand explode at a rate that the market did not account for. Hyperscale clusters are now consuming exabytes of data annually. The multi-year supply agreements being signed today cover capacity that won’t even exist until 2027. That’s a structural shift, not a pricing cycle.

The “Flash Tax” on Innovation

The industry’s default response to AI scale has been overprovisioning capacity. When flash was abundantly available and relatively cheap, inefficiency was tolerated. Now that flash is in short supply and the price has skyrocketed, inefficiency painfully amplifies the problem.

The scarcity of flash changes the math for the entire infrastructure stack. AI clusters are tightly integrated systems including GPUs, storage, networking, and power integrated as a single unit. When flash costs spike and/or allocation is unpredictable, the “all-in” price per deployed GPU goes up forcing many customers to alter their deployment strategies, with profound implications.

Hyperscalers will just write a bigger check and pass the cost down to their tenants. But for Enterprises trying to stand up their next generation AI clusters, the “flash tax” is potentially a project killer. It forces harder, more conservative decisions about what customers can actually fund and how fast they can actually scale.

The Problem with “Enterprise-Grade” Storage

What can no longer be ignored is that most “enterprise-grade” storage architectures weren’t built for AI. They were built for traditional enterprise IT—transactional apps, VM farms, home directories, backup, and predictable throughput—not for feeding GPUs at line rate with high concurrency, low latency, and massive metadata activity… often with data stored in different data centers and clouds.

There are two main challenges historical storage architectures face in the Age of AI. These are what are creating such a crippling demand for memory as organizations look to fix it:

First, the tiering model breaks down. Historically, storage was broadly organized into three tiers: a performance tier for fast access (flash), a capacity tier for “cold” data you still need (HDDs), and an archive tier (often tape). In the Age of AI, data centers increasingly need performance across every tier—because AI pipelines don’t neatly separate “hot” vs “cold,” and retrieval/inference workflows routinely reach into long-tail datasets. But the capacity and archive tiers can’t deliver GPU-class performance because the physical media and architectures weren’t designed for it. What are we seeing? A shift toward all-flash (or flash-heavy) data centers and that pushes memory demand up because flash systems carry significant memory/metadata overhead to deliver performance and manage large namespaces at scale.

Second, the legacy NAS data path has too many hops—and too many copies. Traditional NAS architectures typically involve three roles: the storage node, a controller (or head), and the server/client. Each has CPUs, memory, NICs, and switches between them. If you trace the path from where data physically sits (let’s start at NVMe) to a GPU, the data is repeatedly staged, buffered, and copied: NVMe → CPU/memory → NIC (storage node) → switch → NIC → CPU/memory → NIC (controller) → switch → NIC → CPU/DRAM (client) → GPU (see image below). That’s easily 8–10 copies/hops depending on the design. When performance requirements were modest, this “hop tax” was tolerable. In the Age of AI—especially inference, where latency and concurrency dominate—it becomes a structural bottleneck. The common “fix” is to throw more memory at caching and buffering across the stack, which drives even more demand and cost.

The old-school response to this has been simple: overbuild. Throw more NVMe at it, make more copies, and hope the hardware masks the inefficiency. When flash was cheap and abundant, that waste was just noise. But in a constrained market, that inefficiency becomes a massive and debilitating liability (as well as a massive cost as prices rise!).

In an AI environment, you have unpredictable data reuse, massive and rapid scale, and since GPU servers have far greater gravity than data the orchestration of that data needs to be transparent and global in scale.

The GPU Utilization Trap

The demand for GPUs isn’t slowing down, but they don’t live in a vacuum. A 1000-GPU cluster might have hundreds of petabytes of SSD storage sitting behind it. If that SSD storage is delayed your GPUs spend their time idling.

The SSD crisis doesn’t eliminate GPU demand, but it can shift revenue timing, slow activation cycles, and introduce friction into the AI expansion curve. The GPU economy is now interdependent with SSD availability.

Solving for Efficiency, Not Just Supply

The global NAND issues are not easily fixable, but we can stop being so wasteful with the silicon we already have. At Hammerspace, we look at this as a data orchestration problem. Some of the benefits of the Hammerspace data platform:

• Leverage GPU Server Storage: GPU servers typically ship with local NVMe. Hammerspace enables you to use it as strategic Tier-0 shared storage with Enterprise-class reliability. This enables you to leverage an asset you have already invested in, immediately move forward with your AI projects and since the data is exactly where the compute lives, you will get the highest levels of performance. In the “legacy” graphic shared above data takes 8-10 hops. With Tier-0 (graphic below) it takes only 1.

• Consider HDD Storage: Not all data is “hot.” Orchestrating “cold” sets to high-capacity HDD systems saves the fast flash for the work that actually needs it.

• Recover Stranded Capacity: Most enterprises have petabytes of storage locked in legacy silos. We federate those assets into one logical system so you can stop buying new SSDs to create yet another storage silo for AI and instead continue using what you already own.

• Hybrid Cloud: The public cloud has supply of both GPUs and SSD storage, and we provide a single logical system that enables a hybrid cloud approach that many of our customers are already using. You can leverage GPUs both on premises and in cloud and orchestrate data seamlessly to where you GPUs are without copying data.

• Intelligent Tiering: We orchestrate data between all of these different tiers live and transparently leveraging policies that eliminates any impact to your operations.

• Geo-Deduplication: You shouldn’t have to replicate a dataset three times just to access it in three regions. Our global namespace lets you access one instance of the data from anywhere, instantly reducing your flash footprint.

In Summary

We’re moving into a phase where AI success won’t only be defined by who has the biggest purchase orders for GPUs. It will also be determined by those who manages their data most intelligently regardless of physical constraints. Flash isn’t infinite and that has become glaringly apparent. It’s time we started building architectures that not only reflect that reality but thrive in spite of it.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.2x

• Top 5 Median: 18.4x

• 10Y: 4.1%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 10.7x

• Mid Growth Median: 7.0x

• Low Growth Median: 2.7x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 12%

• Median LTM growth rate: 13%

• Median Gross Margin: 76%

• Median Operating Margin (1%)

• Median FCF Margin: 19%

• Median Net Retention: 108%

• Median CAC Payback: 36 months

• Median S&M % Revenue: 37%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

Build vs Buy

Another week and software continues to grind lower. However, despite all of the carnage, there was another big winner this week! Fastly is up ~100% over the last week. The week prior, 8x8 had the big week (they were up ~70% in a week). Always an opportunity somewhere…

I thought I was done talking about “is software dead” after the last couple weeks Clouded Judgement posts, but I just had more thoughts I wanted to share… I think two things are true. I think people are simultaneously under and over estimating the impact AI will have on the existing software complex. The difference is the timing. Overestimating in the short term, and underestimating in the long term.

I see a lot of arguments claiming software is dead because everyone will just vibe code their own software. I don’t buy this at all… This is really just another iteration of the “build vs buy” debate. Historically, people have chosen not to build internal tools for a few reasons:

• Total cost of ownership: Internal software looks “free” until you account for maintenance, infrastructure, support, upgrades, and opportunity cost of diverted talent.

• Speed to market: Vendors were already built and production ready. Internal builds often took quarters or years, especially once edge cases surfaced.

• Focus on core differentiation: Engineering time is scarce. Companies preferred to spend it on what made them unique, not rebuilding commoditized infrastructure or applications.

• Ongoing maintenance and technical debt: Software is never done. Security patches, compliance updates, performance tuning, and feature requests turn into permanent internal burden.

• Economies of scale and feature velocity: Vendors amortize R&D across thousands of customers and ship improvements continuously. A single company rarely matches that pace.

• Reliability, security, and compliance expertise: Enterprise grade uptime, SOC2, GDPR, industry regulations, audit logs, monitoring. Vendors build teams around this. Internal projects often underinvest until something breaks.

Now, being able to “vibe code” certainly eliminates some of the historical challenges of building your own software. But a lot remains…We can’t underestimate how good vibe coding will get, but I don’t think it changes the overall viewpoint on build vs buy.

However - I think there is a different (but very related) reason that software is challenged. The cost of creating software is going to zero. The risk isn’t that someone will vibe code a internal CRM replacement…The risk is that 10 companies could now create a new CRM, from the ground up, built for a new end user in mind (agents vs people), with a business model for the AI world (consumption / usage vs seats), and now all of a sudden the market is flooded with offerings and the legacy space commoditizes.

This, to me, is the real risk. Software broadly commoditizes, with a new crop of software / value emerging. A big constraint to the development of software is engineering resources. Before the cloud, a constraint was how quickly could you stand up racks of servers to support user growth. In the cloud era that was commoditized, and engineering resources became the constraining factor (how quickly could you develop software). With AI, that constraining resource (engineering velocity) is going away.

So what happens from here…The world is about to be flooded with software. For companies that can’t innovate and capture this next S-Curve of innovation, they will slowly fade to irrelevance. The will be valued as companies in a post-growth industry, and receive a post-growth valuation multiple (see ya revenue multiples…). For those who can, a new vector of growth lays ahead of them.

I’ll end this weeks post with (hopefully) two final anecdotes about the topic of “is software dead.”

First - when looking for examples of historical “major disruption” periods, one many people point to was the iPhone. The first iPhone came out in 2007. In that year, Nokia had a market share of ~40%. They were the king. Well, we all know what happened from there. Apple became one of the most dominant tech companies in history. However, it might surprise people that Nokia still has a ~$40b market cap today. Far from dead! This is down from ~$150b in 2007 (and they’ve had to re-invent themselves). The point is, despite Apple seemingly “killing” them, they never fully died.

Second - the market will typically discount stocks facing major disruption potential far before earnings are impacted (ie before the disruption shows up in the numbers). If we bring this back to the “is software dead” conversation, many are pointing to the recent Q4 earnings reports (we’re in the middle of earnings season right now) as “evidence” that AI isn’t eating software. For the most part, earnings have been good! Retention figures don’t seem to show any sign of cracking. However, I found an awesome graphic floating around X this week (copied below). It showed an index of newspaper companies stock performance and earnings over time (starting in 2002). What you’ll see, is the voting machine of the market saw the disruption coming from the internet, and started to discount the newspaper stocks right away. From 2002 to 2009 those stocks basically went down in a straight line. However, if you look at earnings estimates for that same set of companies, they actually grew for about 5 straight years! During that time, the stocks continued to drop. It wasn’t until 2007 when the earnings really started to get disrupted. Earnings then fell off a cliff. All of this to say - don’t take too much comfort in the short term quarterly results :) Disruption generally takes a bit longer

Of course, there is so much different when comparing the newspaper industry in 2002 to the software industry in 2026…. First - back then legacy newspaper players were slow to adapt. Today, legacy software companies are embracing AI rapidly. Second, newspapers lost their entire core business (print ads) to zero-cost alternatives. Software disruption via AI is more nuanced: I believe the downside is more about trimming growth rates, not erasing demand completely. Third, newspapers were low-barrier commodities (anyone could print and distribute) with fragmented markets and razor-thin margins once ads fled. Software has more network effects, data moats, and higher switching costs. There are many other differences, these were just a few call outs.

There’s a lot that’s quite different about the disruption newspapers faced vs what software companies are facing today. However, the point that could be similar - short term performance may not be the “sign of safety” we think. Software is more likely to die a slow death than an instant one. The threat AI poses is very real (one of biggest risks I wrote about a few weeks back). Most companies face a real commoditization risk as the cost of creating software craters. Only some will capture the next tailwind. At the end of the day, most “AI companies” are really “software companies",” so really the important question is how has a path for durable predictable growth into the future.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.4x

• Top 5 Median: 17.9x

• 10Y: 4.1%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 9.7x

• Mid Growth Median: 6.3x

• Low Growth Median: 2.7x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 12%

• Median LTM growth rate: 14%

• Median Gross Margin: 76%

• Median Operating Margin (1%)

• Median FCF Margin: 20%

• Median Net Retention: 110%

• Median CAC Payback: 36 months

• Median S&M % Revenue: 36%

• Median R&D % Revenue: 24%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

.

.

.

.

Just kidding…Software dying hasn’t taken down Clouded Judgement, not yet at least! Subscribe while you can, legend has it if the median software multiple dips below 3.0x then Clouded Judgement spontaneously combusts. And in the phoenix’s ashes will rise…Hallucinated Judgement, and ode to our new AI overlords. Kidding again (maybe), that will never be the name

Subscribe now

Software Is Dead...Again...For Real this Time...Maybe

Last week I wrote a post titled “Software is Dead…Again.” Since then, IGV is down ~20% (in just 1 week!). If software was dead a week ago what is it now, down an incremental 20%?!

First - some fun stats

The median NTM revenue multiple (cue all the comments “he’s still talking about revenue multiples?!?”) is 3.6x. This is the lowest it’s been in the last 10+ years. For the revenue multiple haters, the median FCF multiple is 16x NTM FCF, for median growth rate of ~20% (alas, once again, cue another set of haters, saying none of the FCF is real and it’s all sitting in SBC). Can’t escape it, maybe software is a zero with no valuation support. Was good while it lasted.

39% of my software index is trading <3x NTM revenue

The median “high growth” software co is trading <10x revenue

Hyperscalers (specifically AWS and Google Cloud) CRUSHED it. Google cloud accelerated from 34% last quarter to 48% this quarter, and AWS accelerated from 20% last quarter to 24% this quarter. Their CapEx projections also exploded. Google guided to ~$180b of CapEx for the upcoming year vs estimates of ~$120b. Meta ~$125b vs estimates of $110b. Amazon $200b vs estimates of ~$150b. Microsoft is run-rating at $120b of CapEx (quarterly capex x 4). Here’s a fun tweet (chart below) showing how steep the ramp in CapEx has become.

Despite the massive ramp in expected CapEx, both Nvidia and TSM were down this week.

Then there’s the few software companies who have already reported earnings. ServiceNow, Atlassian, Palantir, Bill.com, Paylocity (and a few others). Results have actually been pretty solid. Median Q4 quarterly beat is 3.0%, and median guidance raise for next quarter is 2.0%. Both would be the highest mark in the last ~3 years. So far, the numbers aren’t deteriorating (but doesn’t matter, market voting machine moves quick!).

And finally there’s 8x8 - not all hope is lost. They are up ~70% in the last week. There’s always the diamond in the rough somewhere.

So how do I synthesize everything going on with public software companies? The voting machine is saying “shoot now ask questions later.” The voting machine is putting all of software in the “too hard” bucket. A new Anthropic model comes out, and the “AI will replace all software” narrative gets even more vociferous.

When I talk to software bears they say three things (among others):

1. They will all be vibe coded away

2. Classic software has hit its “late maturity” phase. TAMs have been exhausted, competition is up, and the entire industry’s growth rate will collapse, eventually to that of GDP growth, faster than most think. Because of this, the entire sector should trade on GAAP P/E multiple, like the other “late maturity” industries facing disruption.

3. Not only is are they late in the cycle, but they also won’t capture any “AI revenue.” They won’t innovate fast enough, and can’t attract the right talent to innovate. All future growth in software will come from AI agents, none of which the legacy SaaS providers will capture.

4. Operationally, they’re run terribly. SBC is rampant, and is a way of enriching insiders at the expense of shareholders. They all have way too many employees and operate very inefficiently (relative to how they could operate)

When I talk to software bulls they say in response:

1. Good luck vibe coding a system or record. Good luck vibe coding the classic “enterprise” features like RBAC, SSO, audit logs, data residency, SOC2 (and all other security certifications), audit reporting, etc. Good luck encoding all business logic and edge cases (what has taken the classic systems of record years to accumulate). Good luck offering support to your internal users. Good luck maintaining / upgrading / keeping up with the latest best of breed tech. And if you actually are able to do all of this, good luck doing it in a way that has a total TCO lower than just buying the original piece of software from a vendor…At the end of the day, companies aren’t in the business of building internal tools, they’re in the business of selling solutions to customers.

2. Software isn’t in it’s late cycle, it’s no where close (as an industry) to growing at the same rate as GDP growth, it’s way less cyclical than other industries, so still should trade at mor premium multiples

3. Existing players have a distribution and trust advantage, already being installed in the largest enterprises. They’ll capture the agents revenue, or at least their fair share

4. Ya, the bears probably right on point 4. For the most part these companies are run inefficiently, and all the SBC is quite the scheme.

Where do I sit? Personally I don’t think the “vibe code” risk is real…at all…At least not in the short term. I will acknowledge that it’s really hard to predict what the future looks like. And these models improve SO quickly. So what seems impossible to vibe code today will be entirely possible a year from now (and much more). The rate of change is extraordinary. I think it’s naive to underestimate the possibilities of what vibe coding will be able to achieve a year from now (let alone 2-3 years from now). But, I don’t think the overall attitude of “let’s build internally vs pay someone to manage internal software” will change…We’ll always settle on companies outsourcing internal software to vendors (be it legacy SaaS or new AI native startups).

What I do believe - there’s real “front door” risk. I wrote about it here in December. In summary, classic software could be reduced to something that looks more like middleware with agents capturing all of the incremental value on top of them. I mean, look at this graphic from OpenAI…It’s screaming “the classic system of records will be pushed down the stack.” The system of record is literally at the bottom…Down the stack = lower growth potential, and smaller profit pool to capture.

In this world, which I believe we will get to eventually (but the path from here to there is probably longer than what the market is pricing in, but sooner than most industry critics think…), the next growth vector of software won’t be captured by legacy cloud software businesses. And I don’t mean this in an absolute. Some value will, but most (vast majority, I think) of the incremental AI value created in software will go to agents. Incumbents just move to slow. Can’t attract the best talent (what great AI engineers want to work for a legacy company vs cutting edge startup?), have innovators dilemma, have to evolve their business models, etc. Said another way, expansion revenue will be MUCH harder to come by. There will be a small percentage of “legacy SaaS” companies that innovate and capture this new S-Curve. Maybe 10%. The rest won’t…But, not capturing

But, this doesn’t mean the doom of software is at our doorstep…Not capturing the next phase shift of explosive growth doesn’t mean that all growth disappears…I do think there is still a lot of growth left in classic cloud migrations. I was talking to the CEO of a very large business this week. They recently started a migration to SAP. The way they described this migration: I know SAP isn’t a modern company by any stretch, let alone an AI company, but what else was I going to do? There are no “AI-native SAPs.” And I’m not going to have my engineers building out our own SAP. I hear this sentiment all the time.

Said another way, I don’t think we’re necessarily “late cycle” of the cloud software market…I still think there is tons of room for classic cloud migrations. So I don’t think we can call cloud software a melting ice cube, yet. However, maybe the AI native co’s end up stealing the “first” cloud migrations for the industry laggards (that would be an anti-pattern, the laggards will probably not jump straight to AI native challengers).

The other reality - software has a longgggg half life…It takes a long time to “kill.” maybe this is why we’re on the “Software is Dead…Again…Again” blog post title. The software cat has lost 3 of it’s 9 lives! But it’s still alive. For now. With a few lives left.

So the right question - what is the right multiple now for this industry? A “not yet a melting ice cube, but future growth prospects look hampered, and more terminal risk with lots more uncertainty.”

Well, it’s certainly deserves a lower multiple than where it was! When uncertainty goes up, the discount rate must follow. And the uncertainty is WAY up. Dust hasn’t even come close to settling yet.

I do think growth prospects have changed. The voting machine sure believes it. I think it’s gone too far. But at the same time I think some of software universe is still over-valued. There will (could) be some epic returns from here if you pick the 10% who become more of an “AI winner” who can avoid being pushed down / reduced to middleware. Those looking at my multiple charts and saying “but the pre-covid average multiple was 8x NTM rev, we’ll bounce back eventually” will probably be disappointed. If we ever get back there (as an industry median) it will probably be because a new crop of “AI native Agent” companies in hyper growth mode go public and bring the median up. The basket of existing cloud software companies may never get back to a median of 8x NTM rev (mark that statement in case we do, and this is the bottom!)

We can analyze and debate this until we’re blue in the face, but the reality is uncertainty up, discount rate up (I’ll stop beating that dead horse).

If you want a “top line” (revenue, gross profit, etc) vs a “bottom line” (earnings, FCF) valuation metric, you have to be operating in hyper / high growth - because the assumption is if you (and the industry) is in (hyper) growth mode, then you’re early in the S-Curve. The only way top line valuation metrics can ever make sense is if you can grow fast enough to eat the multiple compression that will inevitably come as the industry matures and re-rates. Either that, or the market will just always place high multiples on industries early in a cycle :) We did value internet companies on a multiple of users / eyeballs at one point!

So the question becomes - will legacy SaaS co’s re-enter growth mode and capture a new S-Curve of growth with agents? Will their growth fall off a cliff as agents eat their lunch and turn them into middleware? Time will tell. There will be dispersion. Good luck! For those who don’t capture the AI Agent tailwind, their “death” will probably happen much slower than people think. For those who do, it’s been an elevator down, and will be a slow escalator up. Takes a while for the market to rebuild confidence in an entire sector.

In summary - I think the market is getting it wrong in the short term. These companies have stronger moats and will be more resilient than anticipated. However, I think the market is probably right in the long term. Agents will create a ton of creative disruption, hamper expansion revenue of legacy vendors, and history is against the incumbents capturing this new vector of hyper growth.

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 3.6x

• Top 5 Median: 16.4x

• 10Y: 4.2%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 9.7x

• Mid Growth Median: 6.4x

• Low Growth Median: 2.6x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.

Companies with negative NTM FCF are not listed on the chart

Scatter Plot of EV / NTM Rev Multiple vs NTM Rev Growth

How correlated is growth to valuation multiple?

Operating Metrics

• Median NTM growth rate: 12%

• Median LTM growth rate: 13%

• Median Gross Margin: 76%

• Median Operating Margin (1%)

• Median FCF Margin: 19%

• Median Net Retention: 108%

• Median CAC Payback: 36 months

• Median S&M % Revenue: 37%

• Median R&D % Revenue: 23%

• Median G&A % Revenue: 15%

Comps Output

Rule of 40 shows rev growth + FCF margin (both LTM and NTM for growth + margins). FCF calculated as Cash Flow from Operations - Capital Expenditures

GM Adjusted Payback is calculated as: (Previous Q S&M) / (Net New ARR in Q x Gross Margin) x 12. It shows the number of months it takes for a SaaS business to pay back its fully burdened CAC on a gross profit basis. Most public companies don’t report net new ARR, so I’m taking an implied ARR metric (quarterly subscription revenue x 4). Net new ARR is simply the ARR of the current quarter, minus the ARR of the previous quarter. Companies that do not disclose subscription rev have been left out of the analysis and are listed as NA.

Sources used in this post include Bloomberg, Pitchbook and company filings

The information presented in this newsletter is the opinion of the author and does not necessarily reflect the view of any other person or entity, including Altimeter Capital Management, LP (”Altimeter”). The information provided is believed to be from reliable sources but no liability is accepted for any inaccuracies. This is for information purposes and should not be construed as an investment recommendation. Past performance is no guarantee of future performance. Altimeter is an investment adviser registered with the U.S. Securities and Exchange Commission. Registration does not imply a certain level of skill or training. Altimeter and its clients trade in public securities and have made and/or may make investments in or investment decisions relating to the companies referenced herein. The views expressed herein are those of the author and not of Altimeter or its clients, which reserve the right to make investment decisions or engage in trading activity that would be (or could be construed as) consistent and/or inconsistent with the views expressed herein.

This post and the information presented are intended for informational purposes only. The views expressed herein are the author’s alone and do not constitute an offer to sell, or a recommendation to purchase, or a solicitation of an offer to buy, any security, nor a recommendation for any investment product or service. While certain information contained herein has been obtained from sources believed to be reliable, neither the author nor any of his employers or their affiliates have independently verified this information, and its accuracy and completeness cannot be guaranteed. Accordingly, no representation or warranty, express or implied, is made as to, and no reliance should be placed on, the fairness, accuracy, timeliness or completeness of this information. The author and all employers and their affiliated persons assume no liability for this information and no obligation to update the information or analysis contained herein in the future.

Subscribe now

Share Clouded Judgement

Leave a comment

Subscribe now

Software is Dead…Again!

Well, software is dead again! At least investor confidence is dead… The median NTM revenue multiple for the cloud software universe is 4.1x. That’s the lowest it’s been in 10 years (it was about the same very briefly in 2016, when the fed started hiking rates for the first time after the GFC ZIRP period).

The current median FCF multiple is 18.9x. The previous low in the last 10 years was ~26x!

However, the “narrative violation” metric here is the growth adjusted revenue multiple median is still 0.35x vs the pre covid average of 0.28x (you can see the graph below, I post it every week). So while multiples are at historical lows, so are growth rates. The FCF multiple is the most telling, however. With the current median ~30% lower than the prior low point in 2016

So what’s going on?! I think it’s a couple things.

Mainly, confidence in the SaaS business model has shattered. SaaS businesses were long thought of as “cash flow annuities.” Loose money early on, flip profitable, and then every year print cash predictably. You could then calculate the “intrinsic value” of a SaaS business by summing the present value of every annual cash flow, with a terminal value assumption. More specifically, calculate the present value of the next 10 years of cash flows (discounted back to today), and make an assumption of the terminal value (ie year 11 onward).

There are two big assumptions in this kind of analysis (ie a DCF). There are of course more than two, but I’ll call out two main ones.

The first - you are assuming retention rates remain high and stable. You need this to be true in order to predict stable cash flows in that 10 year calculation. If retention rates drop, your cash flows drop precipitously.

Second - you are assuming there IS terminal value! Said another way - you are assuming the terminal value is not 0 :)

So what’s happening right now? Those two big assumptions are being questioned, which is leading to cratering valuations.

AI is creating huge questions about what the future retention rates of these “stable” software companies will be. Software bears will say this platform shift will lead to deteriorating retention rates as companies leave behind legacy SaaS vendors for modern AI native alternatives. At the same time (and related), this is increasing the probability that the terminal value is in fact 0 for some companies.

Regardless of what you believe, the discount rate has gone up. The probability that retention craters, or that the terminal value for some of these companies is actually 0 is higher today that it was a year ago. That SHOULD translate into lower multiples. I will say, I don’t really agree with the “why” this all happened recently. It feels like the prevailing market sentiment is that it will be easy to vibe code replacement software…I don’t buy that for a number of reasons. However, what I think is actually happening is that the marginal cost to create software has cratered. This will certainly lead to an explosion of competition, and an explosion of choice for software buyers. This could certainly put a hamper on future growth as many market become commoditized quickly by a flood of similar looking solutions.

Even if you vehemently disagree with what’s happening in the market today, the real question becomes “what will change the markets mind.” In my opinion it will take a few quarters of showing “stable” retention rates in the face of AI challengers to give the market confidence back. So far we’ve only had a few companies report Q4 earnings. ServiceNow (a stalwart cloud software business) was one of them. Their retention rates haven’t taken a hit yet!

Quarterly Reports Summary

Top 10 EV / NTM Revenue Multiples

Top 10 Weekly Share Price Movement

Update on Multiples

SaaS businesses are generally valued on a multiple of their revenue - in most cases the projected revenue for the next 12 months. Revenue multiples are a shorthand valuation framework. Given most software companies are not profitable, or not generating meaningful FCF, it’s the only metric to compare the entire industry against. Even a DCF is riddled with long term assumptions. The promise of SaaS is that growth in the early years leads to profits in the mature years. Multiples shown below are calculated by taking the Enterprise Value (market cap + debt - cash) / NTM revenue.

Overall Stats:

• Overall Median: 4.1x

• Top 5 Median: 19.4x

• 10Y: 4.2%

Bucketed by Growth. In the buckets below I consider high growth >22% projected NTM growth, mid growth 15%-22% and low growth <15%. I had to adjusted the cut off for “high growth.” If 22% feels a bit arbitrary, it’s because it is…I just picked a cutoff where there were ~10 companies that fit into the high growth bucket so the sample size was more statistically significant

• High Growth Median: 12.4x

• Mid Growth Median: 7.0x

• Low Growth Median: 3.1x

EV / NTM Rev / NTM Growth

The below chart shows the EV / NTM revenue multiple divided by NTM consensus growth expectations. So a company trading at 20x NTM revenue that is projected to grow 100% would be trading at 0.2x. The goal of this graph is to show how relatively cheap / expensive each stock is relative to its growth expectations.

EV / NTM FCF

The line chart shows the median of all companies with a FCF multiple >0x and <100x. I created this subset to show companies where FCF is a relevant valuation metric.