Public transcript

Avoiding Bankruptcy Before AGI – Dario Amodei

By: Dwarkesh Clips
Length: 5 min
Language: English
Text from: the video's captions
Transcribed: 12 October 2026

Read the full transcript Open on YouTube

Notes in English.

Summary

Dario Amodei explains why Anthropic cannot simply buy compute as if its 10x annual revenue growth will continue indefinitely. Even if powerful AI arrives around 2026 or 2027 as he suspects, economic value like curing diseases takes years to diffuse, and overbuying data centers by even a year or two could bankrupt the company. He contrasts Anthropic's thoughtful approach with what he sees as other companies spending recklessly.

Key points

Dario Amodei estimates powerful AI, a 'country of genius in the data center', could arrive in 2026 or 2027, though he could be off by a year or two.
Even after powerful AI exists, curing diseases takes time due to manufacturing, regulatory processes, and access, as shown by the year and a half it took to distribute COVID vaccines.
Anthropic's revenue has grown about 10x a year, reaching a 10 billion dollar annualized rate at the beginning of the year.
If Anthropic bought compute assuming revenue of 1 trillion dollars and actual revenue came in at 800 billion, no hedge could prevent bankruptcy.
Amodei says Anthropic supports hundreds of billions in spending, not trillions, accepting risk on both demand and slower growth.
He criticizes other companies for spending without understanding their risk, saying they seem to be 'yoloing' rather than being thoughtful.

Questions it answers

00:00Timing of powerful AI

When does Dario Amodei expect powerful AI to arrive?

His hunch, drawn from his essay Machines of Loving Grace, is that a "country of genius in the data center" could arrive in 2026, maybe 2027, though he could be off by a year or two.

Powerful AI may arrive in 2026 or 2027.
Amodei admits he could be off by a year or two.

00:39Diffusion of cures takes time

Why doesn't curing diseases follow immediately from powerful AI?

It requires biological discovery, drug manufacturing, and regulatory approval, and getting access to everyone is hard — the COVID vaccine took a year and a half to distribute, and polio still isn't eradicated after 50 years.

Diffusion will be faster than anything seen before but still has limits.
COVID vaccines took a year and a half to reach everyone.
Polio eradication remains unfinished after 50 years.

01:59Compute buying math

How does revenue growth interact with data center purchases?

Anthropic has grown revenue about 10x a year, from a 10 billion dollar annualized rate, but data centers must be bought a year or two ahead, so a wrong growth assumption can be ruinous.

Revenue has grown roughly 10x per year.
Data centers take a year or two to build or reserve.
Being off by a couple of years can be ruinous.

02:52The bankruptcy risk

Why doesn't Anthropic buy compute assuming trillion-dollar revenue?

If he bought a trillion dollars a year of compute and revenue came in at 800 billion instead of a trillion, no hedge on earth could stop bankruptcy — so Anthropic supports hundreds of billions, not trillions.

A trillion-dollar compute bet risks bankruptcy if revenue falls short.
Growth at 5x instead of 10x would also cause bankruptcy.
Anthropic accepts risk of both excess demand and slower growth.

04:26Spending responsibly

What does Amodei mean by behaving responsibly?

It means being thoughtful about spending rather than "yoloing"; he suspects some other companies haven't written down the spreadsheet and don't understand the risk they're taking.

Responsibility is about thoughtfulness, not absolute spending amounts.
He believes some competitors spend because it sounds cool.
Anthropic spends somewhat less than some other players.

Notes

Powerful AI timing

Dario Amodei's hunch, from his essay "Machines of Loving Grace", is that powerful AI — a "country of genius in the data center" — could arrive in 2026 or 2027.
He acknowledges he could be off by a year or two.

Diffusion takes time even after the breakthrough

The arrival of powerful AI is a "starting gun", not the finish line.
Curing diseases requires biological discovery, drug manufacturing, and regulatory approval — the COVID vaccine took a year and a half to reach everyone.
Polio vaccine has existed for 50 years, yet eradication in remote corners of Africa is still incomplete, though Amodei calls this the most difficult case.
His conclusion: diffusion will be faster than anything we've seen, but still has limits.

The data center bankruptcy dilemma

Anthropic has seen roughly 10x revenue growth per year, at a 10 billion dollar annualized rate at the beginning of the year.
Data centers must be bought a year or two in advance.
If he assumed 10x growth continued, revenue would be 100 billion at end of 2026 and 1 trillion at end of 2027, justifying buying roughly 5 trillion dollars of compute (a trillion a year for five years).
But if revenue came in at 800 billion instead of 1 trillion, "there's no hedge on earth" that could stop bankruptcy.
Being off by a year, or growing at 5x instead of 10x, would also mean bankruptcy.

Behaving responsibly

Anthropic ends up supporting hundreds of billions, not trillions, accepting risk both of excess demand and of slower growth.
Amodei distinguishes responsible behavior from absolute spending amounts — the key is being thoughtful rather than "yoloing".
He suspects some other companies haven't written down the spreadsheet and don't understand the risk they're taking, doing things because they "sound cool".

Text from the video's own captions. Summary and notes written by AI from the transcript, so check anything important against the recording.