Elon Musk is downplaying xAI's compute partnership with Anthropic, claiming the arrangement is short-term and easily terminated. Yet SpaceX's S-1 filing paints a different picture, explicitly detailing contractual payments extending through May 2029.
The discrepancy matters because it reveals competing narratives around one of AI's biggest infrastructure bets. SpaceX committed substantial compute resources to Anthropic, Claude's maker, in what amounts to a major capital deployment. The official SEC filing from SpaceX locks in obligations that run roughly five years out. That's not casual infrastructure sharing.
Musk's public reframing suggests he views the deal as less binding than the contractual terms suggest. This creates tension between what SpaceX's legal and financial disclosures state and what Musk says publicly. For Anthropic, which has scaled aggressively on the back of this compute commitment, the divergence raises questions about deal durability and Musk's actual confidence in the partnership.
xAI remains Musk's competing AI venture, operating separately from OpenAI, the company he co-founded but no longer controls. Giving Anthropic access to SpaceX compute is unconventional leverage in the AI arms race, letting Anthropic train and run inference at scale while SpaceX gains visibility into frontier model development.
The contradiction between Musk's casual dismissal and SpaceX's binding S-1 disclosure suggests either Musk doesn't fully appreciate the contractual obligations he's committed to, or he's signaling to the market that xAI might exit the arrangement earlier than May 2029 if business priorities shift. Neither scenario instills confidence for Anthropic's long-term compute planning.
Anthropic has raised over $7 billion and competes directly with OpenAI and Google's