Why Anthropic's I.P.O. Will Decide the Future of the Economy | Hard Fork
Anthropic's S-1 offers the first public look at an LLM business, showing fast but slowing growth, heavy compute risk, founder-controlled governance, and wider questions on AI politics, security, work and decision-making.
Watch on YouTube ↗Charlie's summary
- ▶ 2:04
Anthropic's S-1 is the first look inside a pure-play LLM business, and it looks healthier than skeptics expected
Erin says the 300-plus page filing finally opens the books on a boom that has drawn trillions. The scary $80 billion net loss is mostly an accounting quirk from convertible debt. The real operating loss in the first half was about $600 to $700 million, down from over $2 billion a year earlier. Revenue was $6.3 billion, which Erin says is growth like no company this size has shown.
- ▶ 6:07
The core risk is a compute bet: non-cancellable contracts signed against demand nobody can measure
Erin says Anthropic has signed huge infrastructure contracts, many non-cancellable, in anticipation of demand. If demand falls short, it is stuck with them; if demand outruns capacity, it loses business. Growth slowed to 44% in Q3 as OpenAI caught up, companies reined in token spending and cheaper models competed. Erin says the key is how Anthropic explains the slowdown: competition is fine for the boom, while a market-wide pullback would be “very very bad” for the whole ecosystem.
- ▶ 8:09
The road show needs a new growth story beyond coding, and biology is the pitch
Erin expects the road show to be where Anthropic explains what comes after its coding hit. The pitch now includes AI curing cancer and developing new drugs, plus legal tech and other ways AI could disrupt the economy beyond coding. Mike also noticed the filing opens with rosier, jazz hands language about curing cancer before backing into the risk factors.
- ▶ 9:40
The risk factors read like six years of AI Twitter, including existential risk and model welfare in passive voice
Max and Erin note the 93-page risk section warns that advanced AI could pose “catastrophic or existential risks to humanity,” written in passive voice, as if unnamed people raised it. Another clause says welfare considerations may “limit or delay the deprecation of older models.” Erin translates that as a sign the company may keep old models running because it thinks they might be alive. Such disclosure is rare in an IPO filing.
Give your email to unlock the rest of the key points — for every episode.
Do more with Charlie
Free — sign in with your email, no password.