Musk softens stance on Anthropic as SpaceX signs Colossus 1 deal and SpaceXAI branding surfaces

The merged SpaceXAI entity is preparing for an IPO expected as soon as next month, touted as the largest in corporate history.

The article notes Anthropic's other compute deals — a 5 GW agreement with Amazon, a 5 GW partnership with Google and Broadcom, $30 billion of Azure capacity from Microsoft, and a $50 billion Fluidstack build-out — won't start online until late 2026 at the earliest. The two companies also begin talking about building multiple gigawatts of AI compute in orbit.

Musk says the Anthropic team he met was highly competent and serious about doing the right thing, and that Claude would probably turn out fine so long as the lab continues questioning itself.

Anthropic signs a deal to use the entire Colossus 1 supercomputer in Memphis, gaining more than 300 megawatts of capacity and access to over 220,000 Nvidia GPUs (H100, H200, GB200). Inference workloads are expected to begin shifting onto the cluster within days. Ami Vora unveils the partnership at Code with Claude in San Francisco; Tom Brown notes that meeting AI demand requires moving a lot of atoms, and that SpaceX is fast at doing so.

SpaceX announces consolidation of xAI into SpaceX, publicly adopting the SpaceXAI branding; Musk confirms xAI will be dissolved as a separate company with its AI products folded into SpaceX.

Musk sat down with senior Anthropic executives last week and walked away impressed; he wrote on X that 'No one set off my evil detector.'

Anthropic clashed with the Pentagon; the lab was blacklisted in March as a supply chain risk.

Elon Musk spends February criticizing Anthropic, calling it misanthropic and accusing its models of demographic bias after its $30 billion funding round, and predicting it would never win.