What would Anthropic need to earn to justify spending ∼$1 trillion on computing?

Steve Rattner uses a Financial Times estimate: if Anthropic commits $1 trillion to compute, and wants a 20% after-tax return, at a 40% operating margin and 17% tax rate, it would need *$602 billion in annual revenue*.

Now compare that to reality:

- Anthropic today: *$35B* in revenue over the last 12 months, *$81B* annualized run-rate in October.
- Apple: *$478B*
- Alphabet: *$471B*
- Microsoft: *$346B*
- Meta: *$241B*

In other words, to make that $1T bet pay off, Anthropic would need to become *bigger than Apple in revenue* — and Apple took 50 years, selling iPhones to billions of people, to get there.

Anthropic would need to grow *7.4x from its current run-rate*, and *17x from its last-12-months revenue*.

This doesn't mean Anthropic will fail. It means the AI race is no longer about building the best model. It's about who can turn compute into _that_ much revenue, fast enough.

The chart is not a forecast. It's a reality check: the math of the AI boom is now Apple-sized.

#Anthropic