From “Misanthropic” to Business Partner: Why Musk Suddenly Needs Anthropic

With input from Axios, Anthropic, CNBC, Wired, Reuters, Bloomberg.

Three months ago, Elon Musk was mocking Anthropic online, calling the AI company “misanthropic” and accusing it of hating Western civilization. Now he’s handing over one of the biggest pools of AI computing power on the planet.

Welcome to the AI industry in 2026, where rivalries move fast, but demand for chips moves even faster.

Anthropic announced Wednesday that it signed a major compute deal with SpaceX, giving the company access to the full capacity of the Colossus 1 data center in Memphis – more than 300 megawatts of power and over 220,000 Nvidia GPUs. The extra capacity is expected to ease the growing strain on Anthropic’s Claude chatbot and API services after usage exploded far beyond the company’s forecasts.

CEO Dario Amodei said Anthropic expected revenue and usage growth of around 10x this year. Instead, the company saw “80x growth” in the first quarter alone. That surge left Claude users running into rate limits and reliability problems during peak hours.

SpaceX suddenly looked like the perfect solution.

The arrangement also gives Musk something he badly needs: money flowing from infrastructure that apparently wasn’t being fully used.

Musk’s AI company xAI – now being folded into what he’s calling “SpaceXAI” – built out enormous computing capacity around Memphis to compete with OpenAI, Google and Anthropic. But industry analysts say Grok’s user growth hasn’t come close to matching the scale of the hardware investment.

Instead of sitting on expensive idle GPUs, Musk can now rent that power to one of the hottest AI companies in the world. Timing matters here. SpaceX is reportedly preparing for a public offering filing next month, and turning excess compute into revenue helps clean up the financial story before investors start digging through the numbers.

Rather than explaining why thousands of chips are underused, Musk can point to Anthropic as a massive paying customer.

There’s also the Sam Altman factor hanging over all of this.

Musk is still locked in a legal war with OpenAI and Altman, the company he helped create before their relationship imploded. Anthropic, founded by former OpenAI researchers, is one of OpenAI’s biggest competitors. In Silicon Valley terms, the alliance almost writes itself.

“Enemy of my enemy is a compute partner,” tech researcher Ben Pouladian joked on X.

The shift in tone from Musk has been dramatic. After spending much of February publicly attacking Anthropic, he posted Wednesday that he’d spent time with the company’s leadership team and came away impressed.

“Everyone I met was highly competent and cared a great deal about doing the right thing,” Musk wrote on X. “No one set off my evil detector.”

That’s a notable change from the guy who recently accused the company of being “doomed to become the opposite of its name.”

Behind the scenes, though, this deal is mostly about survival in the AI arms race. AI companies are running into a brutal reality: there simply aren’t enough chips, data centers or power supplies to keep up with demand.

Anthropic has been scrambling to secure more infrastructure wherever it can find it. The SpaceX deal adds to a growing list of massive compute agreements, including partnerships with Amazon, Google, Broadcom, Microsoft and Nvidia. Altogether, the company has committed to tens of billions of dollars in AI infrastructure expansion.

Anthropic also hinted at something even bigger – orbital AI computing. As part of the agreement, the company said it’s interested in working with SpaceX on “multiple gigawatts” of compute capacity in space.

Yes, actual space.

The idea sounds futuristic, but power availability is becoming one of the biggest bottlenecks in AI development. Data centers are already straining electrical grids across the US, while local protests over pollution and energy use are growing louder. xAI’s Memphis operation has already faced backlash over gas-powered turbines tied to worsening air quality.

Meanwhile, Anthropic keeps growing. Fast.

The company is reportedly discussing fundraising at a valuation approaching $900 billion, a staggering leap from the roughly $380 billion valuation floated earlier this year. Claude has become one of the few serious challengers to ChatGPT, especially among developers and enterprise customers.

That growth is exactly why Anthropic needed more compute yesterday, not six months from now.

For Musk, the partnership may be less about friendship and more about pragmatism. Anthropic gets the GPUs. SpaceX gets the cash flow. And both companies get a stronger position against OpenAI.

In the AI business, apparently, even sworn enemies can become infrastructure partners if the economics line up.