Oct 29, 2026 · 1h 12m
Anthropic’s S-1 tests the economics of AI’s capital race
Is AI the new cigarettes? Our VC panel digs into Anthropic’s S-1 | E2356
Anthropic’s filing offers a rare view into whether explosive AI usage can translate into durable profits despite immense compute and infrastructure costs.
- 1Anthropic’s 900% net revenue retention signals extraordinary expansion within existing customer accounts, but raises questions about pricing durability.
- 2The reported $80.7 billion loss largely reflects non-cash valuation changes, separating accounting results from the company’s underlying operating performance.
- 3The panel sees workflow integration, infrastructure exposure, and founder conviction as bigger long-term questions than raw demand for AI.
Don't miss
The panel unpacks Anthropic’s extraordinary 900% net revenue retention and asks whether such customer expansion can sustain long-term pricing and profits.
The brief
Jason Calacanis, D.A. Wallach, Dave McClure, and Benjamin Black use Anthropic’s S-1 to examine a company growing rapidly while spending heavily on compute and infrastructure.
The filing’s $80.7 billion net loss looks alarming until the panel separates non-cash fair-value changes on convertible notes from operating performance and cash economics.
Anthropic’s reported 900% net revenue retention becomes the episode’s sharpest datapoint: existing customers spent roughly nine times more year over year, suggesting powerful but untested demand.
The panel compares AI with tobacco’s addictive, high-margin economics, then argues that the near-term bottleneck may be employees integrating AI into workflows rather than access to compute.
The debate widens to Dario Amodei’s caution, Sam Altman’s aggressive strategy, hyperscaler spending, and whether neocloud investors are underwriting infrastructure they cannot control.
The closing thesis is that AI could make entrepreneurship dramatically cheaper, even as today’s capital race leaves public markets and infrastructure investors exposed to a costly reversal.
What was said on this episode
37 statements · 24 positive · 8 negative · 4 mixed · 1 neutral
Anthropic appears operating-profitable for the last two quarters.
“Company looks like they are profitable, looks like they've been... on an operating basis, profitable for the last two quarters.”
Listen at 4:08
Anthropic continues to grow despite slower recent growth.
“I would still say that Anthropic is growing.”
Listen at 6:28
Anthropic added at least $5 billion in quarterly revenue.
“I think they added, you know. at least $5 billion of new revenue in the quarter.”
Listen at 6:45
Anthropic may reach $150–$200 billion run-rate next year, possibly $300 billion.
“I'm still a believer that they're going to hit $300. I might guess it's going to be between 150 to 200 next year.”
Listen at 7:11
Anthropic is attractive around a $1.5 trillion valuation, but not $2–$2.5 trillion.
“I think I'm still a believer in Anthropic, probably at one and a half trillion, not at two or two and a half trillion.”
Listen at 7:26
AI businesses require new analytical measures beyond traditional SaaS metrics.
“we have to have a whole new set of measures for these businesses that really help us understand what's going on.”
Listen at 9:01
Anthropic is the fastest-growing company in business history.
“It's the fastest growing company in the history of business.”
Listen at 10:33
Anthropic is the greatest business humans have ever created.
“I think this is the greatest business created by humans on planet Earth ever.”
Listen at 11:30
AI breakthroughs could create currently unmeasured revenue categories.
“There's potentially new lines of revenue that we're not even measuring.”
Listen at 11:54
Higher corporate AI spending could reach several trillion dollars annually.
“if you just double it or triple it you can easily start to see how we get to a couple trillion dollars a year of spend”
Listen at 14:44
Low corporate AI spending mainly reflects uncertain utility rather than high cost.
“my hypothesis would be it's mainly a utility question.”
Listen at 15:17
Employees are currently unable to deploy AI effectively in organizations.
“the people, the meat sacks at organizations are not able to deploy it.”
Listen at 18:12
AI productivity gains will spread from technology firms to ordinary companies.
“I think you're going to see AI productivity gains hit not just tech companies, but normal companies.”
Listen at 20:14
Anthropic’s product is arguably the most useful product ever created.
“this is the most useful product I think that's ever been created.”
Listen at 23:39
Apple Intelligence will eventually match leading AI products.
“at some point in the future apple intelligence is going to be at parity with these things”
Listen at 28:44
Apple’s distribution puts it in a leading position in consumer AI.
“I think they have the distribution. So they're in the pole position here.”
Listen at 29:23
Canva’s revenue growth has slowed somewhat.
“Canva's growth on a revenue basis has slowed down a little bit”
Listen at 35:09
Canva’s secondary-market price fell 25% after a growth slowdown report.
“the price dropped 25% in the secondary market.”
Listen at 36:08
Canva and Figma may remain good businesses but no longer transform the world.
“They're going to be nice businesses, but they're not going to change the world anymore.”
Listen at 36:40
Figma’s network effects and designer workflows remain stronger than AI alternatives.
“they have network effects. Turns out graphic designers still do better graphics work for real companies than OpenAI or Gemini does.”
Listen at 37:59
AI tools will progressively capture more graphic-design workflow.
“They're going to progressively eat more and more of that workflow.”
Listen at 38:09
Offering free AI intelligence represents an exceptionally large competitive opportunity.
“It's the biggest pot of gold in human history is to figure out how to offer intelligence for free right now.”
Listen at 38:52
If Anthropic’s leaders are wrong, stronger business operators will eventually defeat them.
“if they're wrong, they're just religious extremists who happen to be engaged in the greatest business of this moment, but they will ultimately be beaten by people who are really good at business.”
Listen at 40:30
Sam Altman’s aggressive compute strategy was intended to win the AI business race.
“Sam was playing this game to win.”
Listen at 44:34
Securing compute capacity is necessary to win the AI business competition.
“when it became clear that Sam was right, That you had to have compute locked up to win the business game.”
Listen at 44:38
The AI infrastructure cycle may ultimately make Microsoft and Google much larger, or Oracle the winner.
“it will probably end with Microsoft and Google being much bigger maybe Oracle will even win”
Listen at 47:25
The AI labs, cloud providers, and hardware providers form a coherent circular economy.
“I'm convinced that it makes sense within itself as one organism.”
Listen at 54:09
The AI infrastructure ecosystem remains viable if demand grows rapidly.
“as long as as long as that demands growing like gangbusters the organism's gonna be fine”
Listen at 55:09
Hyperscalers are resilient because their dominance lets them adjust infrastructure spending.
“I think the hyposcales are totally fine. I think they're so dominant. They can always adjust.”
Listen at 56:47
Neocloud companies face uncontrollable financing costs and project timelines.
“with all... the Neo labs, your cost of capital is out of your control. And also how long it takes you to implement your projects is out of your control.”
Listen at 57:04
Crypto’s strongest current case is currency-debasement protection, not speculation.
“the bull case for crypto at this point probably does fall away from this speculative ponzi scheme thesis and is merely a debasement trade thesis.”
Listen at 1:02:04
Stablecoins have a strong practical use case.
“stablecoin has a really solid use case.”
Listen at 1:03:19
Tokenized real-world assets may have legitimate utility.
“another area that's possibly not bullshit is tokenizing real world assets”
Listen at 1:03:29
DeFi lending could have durable utility over the long run.
“I think even DeFi in the long run is not crazy. I think lending makes sense.”
Listen at 1:03:39
Ten to twenty VC firms control roughly 75% of invested capital.
“At the top of the market, there's 10 VC firms, maybe 20, that are taking down 75% of all the capital.”
Listen at 1:04:30
AI will enable top future entrepreneurs to raise less capital than prior founders.
“the best entrepreneurs of the next few years are going to raise less money than the best startup founders have ever raised”
Listen at 1:05:50
Teenagers building AI companies may become the greatest entrepreneurs in history.
“These are going to be the greatest fucking entrepreneurs in human history.”
Listen at 1:08:54
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.