SpaceX doubles revenue on Anthropic and Google AI compute deals

SpaceX said its revenue roughly doubled year-over-year in its first quarterly earnings report since the company went public in June, with growth driven largely by new compute deals with Anthropic and Google and continued expansion of its Starlink satellite internet business. The results mark the first time investors have had a detailed public look at the finances of a company long known primarily for rockets and satellites, and they show a business that now generates a substantial share of its revenue from selling computing capacity to AI companies rather than from launches alone.

The compute business, which SpaceX has built around unused capacity in its data infrastructure, saw revenue grow roughly threefold from a year earlier to about $2.6 billion, according to the company's earnings materials. That growth followed two deals signed earlier this year: one with Anthropic in May and one with Google in June, both providing the AI companies with computing capacity in exchange for revenue that flows back into SpaceX's broader business, including Starlink and its rocket programs.

Despite the revenue growth, the AI and compute division posted a loss of about $1.5 billion for the quarter, a figure the company described as slightly smaller than the loss it reported in the same quarter a year earlier. The combination of fast-growing revenue and continued losses is typical of infrastructure-heavy AI compute businesses, which require large upfront investment in hardware and power capacity before reaching profitability at scale.

SpaceX's move into AI compute puts it in direct competition with a category of companies known as "neoclouds" — firms such as CoreWeave that specialize in renting out GPU capacity to AI labs rather than operating as traditional cloud providers like Amazon Web Services or Microsoft Azure. Unlike most neoclouds, SpaceX did not build its compute business from scratch to serve AI customers; it repurposed capacity originally built to support Starlink's satellite network and its own rocket operations.

Starlink itself continued to be a significant growth driver, according to the earnings materials, as the satellite internet service keeps expanding its subscriber base across the United States, the United Kingdom and other markets, alongside newer efforts such as direct-to-phone connectivity. The company's overall revenue picture now spans three distinct lines of business — launch services, Starlink, and AI compute — a diversification that executives have pointed to on the earnings call as a hedge against volatility in any single segment.

The earnings call was also SpaceX's first as a public company, following its listing in June, a shift that requires the company to disclose financial details it withheld for years while privately held. Chief Financial Officer Bret Johnsen and President Gwynne Shotwell both spoke on the call, laying out the company's growth priorities across its various business lines to an audience of public shareholders for the first time.

For Anthropic and Google, the compute deals with SpaceX are part of a broader pattern of AI companies diversifying where they source computing capacity beyond the traditional big three cloud providers. As demand for AI training and inference computing has outpaced the ability of established data center operators to build capacity fast enough, AI labs have increasingly signed deals with a wider range of infrastructure owners, including companies like SpaceX that were not originally built as cloud providers.

Analysts covering the space and AI sectors have noted that SpaceX's ability to monetize otherwise idle infrastructure capacity gives it a cost advantage that dedicated neoclouds, which must build compute capacity specifically to sell it, do not have in the same way. That advantage, however, comes with a ceiling: SpaceX's compute capacity is ultimately constrained by what its Starlink and launch operations need first, meaning its ability to scale the AI business further may depend on new, dedicated infrastructure investment going forward.

The results come as public markets have shown particular interest in how newly listed space and satellite companies are positioning themselves within the broader AI infrastructure boom, with investors watching to see whether SpaceX's compute revenue proves to be a durable business line or a shorter-term way to monetize excess capacity while demand for AI compute remains unusually high.

SpaceX's first earnings report as a public company gives investors a data point rather than a verdict: the company's AI compute segment is growing quickly and generating meaningful revenue, but it remains unprofitable, and its long-term trajectory will depend on decisions the company has not yet detailed publicly about how much new, dedicated capacity it plans to build specifically for AI customers going forward.