Anthropic is about to go public, and right before that happens, the numbers on how much compute it has locked in just came out. It is a lot.
Since last October, the company signed deals adding up to roughly 14.8 gigawatts of computing capacity. Put a dollar figure on all of it over the life of those contracts and analysts land around 517 billion dollars. That is not money spent yet. That is money committed, spread across the next decade.
Google and AWS make up the biggest chunk, together accounting for around 11 gigawatts of that capacity. SpaceX shows up in the deal list too, which tells you how far outside the usual cloud playbook this company is reaching just to keep up with demand.
Why the rush. Claude Code and Cowork blew past what Anthropic expected this year, and running those tools for millions of people takes an enormous amount of raw compute. When a product suddenly gets that popular, you either find the power to run it or you watch it slow down for everyone using it.
A number like 517 billion is easy to read as reckless. My first reaction was close to that too. But think about what it actually is: a bet that demand for AI tools keeps climbing for the next ten years, made by a company that is about to ask public investors to make the same bet right alongside it.
The market will get to weigh in soon enough. The IPO prospectus and an investor day are both expected within weeks. Until then, this is the clearest look yet at what it actually costs to keep an AI company running at the scale people now expect from one.
——
Follow: @Ali Demi
Book your free AI clarity call, NOW!
https://buff.ly/TpWy277
——
Sources:
https://www.theinformation.com/articles/anthropic-clinched-517-billion-compute-deals-11-months
https://www.datacenterdynamics.com/en/news/anthropic-signed-517bn-in-compute-agreements-in-past-11-months/
https://finance.biggo.com/news/28be15e2-a525-4cce-94e8-f7b6c822bc19
Repost this. Thanks.
