Anthropic Pays Musk $4B After He Called Them ‘Evil’

Anthropic just handed Elon Musk a $4 billion check. The same Musk who called Anthropic 'evil' three months ago is now their landlord, collecting rent for every Claude query that runs through SpaceX’s Colossus data center in Memphis.

The deal, revealed by Fortune on May 7, turns personal beef into corporate necessity. Anthropic gets exclusive use of the 300-megawatt facility, doubling Claude’s capacity overnight. Musk’s companies pocket enough cash to fund another Starship program.

Users already feel the difference. Pro and Max subscribers report immediate usage limit increases as Anthropic pushes compute through Musk-owned silicon. The same infrastructure that powers Grok now serves Claude’s 7% of users who’ve been hitting session limits during peak morning hours.

Three months ago, Musk sat in an X Spaces session calling Anthropic 'evil' and claiming they were 'trying to destroy humanity.' His exact words: 'These guys are worse than OpenAI, and that's saying something.'

Now he controls their compute destiny. The SpaceX-xAI merger earlier this year means every Claude response runs through property Musk effectively owns. It's the AI equivalent of renting office space from someone who thinks you're Satan.

The speed of this reversal stunned industry watchers. One venture capitalist told Fortune: 'He went from moral panic to monthly invoices faster than most founders pivot their startup ideas.'

Anthropic needed compute so badly they paid their harshest critic. The numbers explain why: 7% of users hitting session limits during 5am-11am Pacific translates to millions of frustrated customers and potential churn.

The $4 billion figure isn't just rent — it's protection money against Musk throttling their access. When your competitor owns the pipes, you pay whatever they ask or watch your service degrade.

This explains why Anthropic announced capacity increases immediately rather than waiting for gradual rollout. They're burning cash to patch a bleeding user base, even if that cash lines Musk's pockets.

The deal sandblasts any remaining myth about AI compute commoditization. When sophisticated companies pay $4 billion to their enemies just to keep servers running, that's not a healthy market.

Smaller AI startups watching this unfold now know the game is over. It's not just about money anymore — it's about whether Elon Musk thinks you're evil or merely annoying. The barrier to entry just became personal politics plus unlimited capital.

Anthropic's move also highlights how the AI stack has verticalized into personality-driven fiefdoms. Companies aren't just building models; they're navigating the whims of billionaires who control both infrastructure and public opinion.