Elon Musk once said Anthropic "hates Western Civilization" and called its AI "woke" — now he's declaring the company "the leader in AI" while collecting $1.25 billion a month from them in compute fees. Funny how that works.
The reversal isn't just cosmetic. As AI models become the lens through which millions of Americans access information, the question of who builds them — and what ideological biases they bake in — is a question of liberty itself. Musk had it right the first time. Now he's got a business partner.
In an X post Thursday, Musk wrote that he was "clearly wrong about Anthropic," adding: "They are obviously currently the leader in AI. No company has released a model as good as Mythos/Fable and they will undoubtedly have Mythos 2 ready soon." This from the same man who in February accused Anthropic of stealing training data "at massive scale" and called it "smug, sanctimonious and hypocritical." In March, he asked: "Is there a more hypocritical company than Anthropic?" Last September, he wrote that "winning was never in the set of possible outcomes for Anthropic."
What changed? In May, the two companies announced a deal giving Anthropic access to SpaceX's Colossus 1 data center — more than 300 megawatts of capacity, over 220,000 Nvidia GPUs. SpaceX's S-1 filing revealed Anthropic agreed to pay $1.25 billion a month through May 2029. That's $15 billion a year. Either side can walk with 90 days' notice.
An X user pointed out that Anthropic is "completely reliant on the compute rented from SpaceXAI" and that Musk could "kill" the company if he wanted. Musk responded he'd "never cut them off in a way that hurt them badly, even as a competitor. That's not my style." He also wrote that he spent time with Anthropic's senior team and "no one set off my evil detector."
Here's the problem. Musk's original criticisms weren't wrong. Claude, Anthropic's flagship model, has been repeatedly documented exhibiting left-wing bias — the kind of algorithmic gatekeeping that decides what questions get answered and which perspectives get disappeared. When the most powerful AI in the room is "woke," as Musk himself once labeled it, ordinary Americans lose access to unfiltered information. That's not a technical issue. That's a speech issue.
And now the consolidation deepens. A handful of companies — OpenAI, Anthropic, SpaceXAI — are building the infrastructure that will mediate how citizens learn, debate, and think. Musk still sees the competitive landscape clearly: "Space(XAI) is only 3 years old," he wrote May 26. "Let's see where things stand 3 years from now." But competition between three oligarchs isn't competition. It's a cartel with better branding.
Business Insider framed the story as a charming rivalry warming into partnership. What got buried: the stakes. When the man who owns the compute infrastructure also controls a competing model — and can terminate his rival's access with 90 days' notice — Americans should ask who really holds the keys to the digital public square.
Musk built X into the closest thing this country has to a free-speech platform. He's earned the benefit of the doubt. But $15 billion a year has a way of changing what sets off your evil detector. The question isn't whether Anthropic's AI is good. It's who decides what "good" means — and whether any American still gets to decide for themselves.
