Anthropic CEO Dario Amodei suggests OpenAI doesn't "really understand the risks they're taking"
Anthropic CEO Dario Amodei isn't sure that OpenAI "has written down the spreadsheet" when it comes to investing in AI compute.
In a recent podcast, he lays out why the math behind AI infrastructure spending is so complicated: he's convinced AI systems performing at the level of Nobel Prize winners could exist "within a few years" or even faster. So in that sense, it would make sense to throw every penny into more compute. But Amodei says it's far from clear how quickly that capability will actually translate into revenue.
Take curing diseases, one of the most promising economic applications. Even if AI could theoretically invent cures for everything, Amodei says those breakthroughs still need to go through biological discovery, drug manufacturing, and regulatory approval. So even if it's faster, it could take years before money starts flowing from the breakthrough.
Anthropic's own revenue went from zero to $100 million in 2023, hit $1 billion in 2024, and went to $9 to $10 billion in 2025, according to Amodei. Annualized revenue sits at $14 billion as of early 2026.
Being off by one year could mean bankruptcy
While Anthropic's growth has been roughly 10x per year, Amodei cautions against assuming that pace will hold. "I could buy $1 trillion of compute that starts at the end of 2027. If my revenue is not $1 trillion dollars, if it's even $800 billion, there's no force on earth, there's no hedge on earth that could stop me from going bankrupt if I buy that much compute," he says. Being off by just a single year in that growth rate—or seeing 5x growth instead of 10x—would be enough to sink the company.
Amodei says he gets the impression that some competitors "don't really understand the risks they're taking. They're just doing stuff because it sounds cool," adding that Anthropic has "thought carefully about it." While he only refers to "some of the other companies," the comment reads as a likely jab at OpenAI.
Anthropic is reportedly planning to invest in at least ten gigawatts over the next few years. OpenAI's ambitions are significantly larger: last year alone, it announced partnerships with Nvidia, Broadcom, Oracle, and AMD totaling over 30 gigawatts of compute capacity, though much about those deals remains unclear.
"We're buying a lot. We're buying a hell of a lot. We're buying an amount that's comparable to what the biggest players in the game are buying," Amodei says. "But if you're asking me, 'Why haven't we signed $10 trillion of compute starting in mid-2027?' First of all, it can't be produced. There isn't that much in the world. But second, what if the country of geniuses comes, but it comes in mid-2028 instead of mid-2027? You go bankrupt."
AI News Without the Hype – Curated by Humans
As a THE DECODER subscriber, you get ad-free reading, our weekly AI newsletter, the exclusive "AI Radar" Frontier Report 6× per year, access to comments, and our complete archive.
Subscribe now