A disclosed agreement between SpaceX and Anthropic shows how access to large-scale computing power has become central to the artificial intelligence race, while also drawing investor attention to SpaceX’s governance, debt and AI spending ahead of its planned public listing.
SpaceX’s long-awaited public IPO filing has revealed that Anthropic has agreed to pay the company $1.25 billion per month for access to artificial intelligence computing infrastructure, in a deal that underlines the growing strategic value of data-centre capacity in the AI sector.
The agreement, disclosed in SpaceX’s S-1 filing and reported by WIRED, would amount to roughly $15 billion a year and is scheduled to run through May 2029. Anthropic, the developer of the Claude AI models, will use GPU capacity at SpaceX’s Colossus and Colossus II data centres, which were originally built for Elon Musk’s xAI business.
The scale of the agreement reflects one of the main constraints facing frontier AI companies: access to computing power. As demand for large language models and AI coding tools expands, the ability to secure large quantities of graphics processing units has become a major factor in commercial competition.
According to Reuters, the deal allows either SpaceX or Anthropic to terminate the arrangement with 90 days’ notice, while lower fees will apply during the initial capacity ramp-up in May and June. Reuters also reported that Anthropic expects second-quarter sales of at least $10.9 billion, compared with $4.8 billion in the March quarter.
The arrangement also gives SpaceX a new revenue stream as it prepares for what could be one of the largest stock market listings in history. The company is reportedly seeking to raise about $75 billion at a valuation of around $1.75 trillion, with a possible Nasdaq listing under the ticker SPCX. TechCrunch reported that SpaceX’s filing presents the company not only as a launch and satellite business, but as a broader technology group with major ambitions in AI infrastructure.
That shift is significant. SpaceX remains best known for reusable rockets, Starlink satellite internet and the development of Starship. However, the IPO documents suggest that AI infrastructure is becoming an increasingly important part of the company’s investment case. The company’s data centres, originally intended to support xAI and its Grok chatbot, are now being partly monetised through external compute contracts.
The filing also exposes the financial burden of SpaceX’s expansion. WIRED reported that SpaceX generated nearly $4.7 billion in revenue in the first quarter of 2026 but lost almost $4.3 billion. In 2025, the company generated $18.7 billion in revenue and posted a loss of $4.9 billion, with heavy spending on AI technologies and rocket development.
For Anthropic, the deal provides access to capacity at a time when compute shortages can limit growth. The company competes with OpenAI, Google DeepMind, Meta and xAI in a market where model performance, product availability and enterprise adoption are closely linked to infrastructure. Anthropic’s willingness to commit such sums to a rival-controlled platform illustrates the extent to which compute has become a scarce commercial asset.
For SpaceX, the agreement helps support a wider narrative around the IPO. The company is presenting itself as a business with multiple growth engines: space launch, satellite broadband, AI services and potentially space-based data centres. The filing identifies a possible $28.5 trillion market opportunity across SpaceX’s businesses, with a large share tied to AI.
The filing has also renewed scrutiny of Elon Musk’s control over the company. WIRED reported that the documents describe a structure in which Musk and his allies would retain significant voting power, while SpaceX would use provisions of Texas law to protect itself against hostile takeovers and attempts to remove executives or board members.
Investor groups have raised concerns about governance ahead of the proposed flotation. According to WIRED, public employee retirement fund leaders in California, New York and New York City criticised SpaceX’s proposed structure as unusually management-friendly. Their concern is that a company of this scale should offer stronger protections for long-term institutional shareholders.
The Anthropic agreement therefore has two sides. On one hand, it gives SpaceX a major external customer for infrastructure that might otherwise have remained largely internal to Musk’s AI operations. On the other, it highlights how closely the company’s future valuation is now tied to high-cost, capital-intensive bets in AI, alongside the continued development of Starship and Starlink.
The deal also marks a further convergence between space, cloud infrastructure and artificial intelligence. SpaceX’s IPO filing suggests that the company is no longer presenting itself simply as a launch provider. It is seeking to convince investors that its future lies in a broader industrial model built around communications, compute, aerospace systems and AI.
Whether public markets accept that valuation will depend on more than the scale of the Anthropic contract. Investors will also assess the durability of the compute deal, the profitability of Starlink, the cost of Starship, the losses in AI, and the governance terms under which Musk would continue to control the business.
For now, the filing shows that SpaceX’s public listing is not only a test of investor appetite for space technology. It is also a test of whether markets are prepared to value AI computing infrastructure as one of the central assets of the next technology cycle.
