SpaceX Now Makes More From AI Than From Space: Revenue Triples
SpaceX's AI division brought in $2.6 billion in the second quarter of 2026, three times the figure from a year earlier. The main driver is a pair of contracts to lease GPU capacity: one with Anthropic (May 2026) and one with Google (June 2026). SpaceX has entered the AI cloud market, competing with CoreWeave, and now earns more from AI than from rockets.
AI-processed from The Verge; edited by Hamidun News
In the second quarter of 2026, SpaceX's AI division generated $2.6 billion in revenue, more than triple the figure for the same period a year earlier. Two large compute-leasing contracts drove the growth: one with Anthropic, signed in May, and one with Google, signed in June 2026. In the quarterly results, the AI business outearned the core rocket and satellite business for the first time.
How much did SpaceX earn from its AI cloud?
The AI division posted $2.6 billion in quarterly revenue but remained unprofitable: it lost $1.5 billion, only slightly less than in the same quarter of 2025. Tripling revenue while still losing money is a classic sign of an investment phase, in which infrastructure is built out faster than it can pay for itself.
Key figures for the quarter:
- AI division revenue: $2.6 billion (more than 3x year over year)
- AI division loss: $1.5 billion over the same period
- Anthropic contract: May 2026
- Google contract: June 2026; according to CNBC, Google pays ~$920 million per month
- In its IPO filing, SpaceX calls the AI division "the main source of the company's value"
How SpaceX entered the neocloud market
SpaceX provides GPU compute to third-party AI companies, competing with CoreWeave and other neoclouds, providers that specialize exclusively in AI infrastructure. The Google contract shows how deep the move goes: according to CNBC, the search giant pays SpaceX around $920 million per month for compute capacity.
Neoclouds have become a strategically important class of infrastructure. Companies such as CoreWeave raise tens of billions of dollars precisely because the GPU clusters that serve large AI models require specialized management. SpaceX enters this market with a strong card: in its very first operating year, it signed agreements with two of the world's leading AI companies.
The growth of the AI business is also changing SpaceX's capital strategy. The company has sharply increased spending on data center construction and equipment purchases, which explains the gap between revenue and profit.
Why are investors looking at SpaceX differently?
SpaceX is deliberately rethinking its positioning ahead of a public offering. In its IPO filing, the company stated outright that the AI division is the "main source of its value", an unusual phrasing for a traditional rocket operator. Falcon 9 and the Starlink network have long been seen as SpaceX's main value drivers; now the narrative has shifted.
"The AI division accounts for the bulk of the company's value," says
SpaceX's 2026 IPO filing.
The quarterly data supports this narrative: AI revenue exceeded launch revenue. The signal to the market is that SpaceX should be valued first and foremost as an AI infrastructure company, not just as a rocket operator.
What this means
In a single year, SpaceX has become a notable player in the AI cloud market, and it did not start from zero: the company used existing relationships with leading labs and its existing infrastructure to scale up compute capacity. A $1.5 billion loss against $2.6 billion in revenue is an investment bet, not a crisis. SpaceX is deliberately entering a segment where competition with CoreWeave and the hyperscalers is likely to be fierce in the coming years.
Frequently asked questions
How much does Google pay SpaceX for compute capacity?
According to CNBC, Google pays about $920 million per month for access to compute capacity under an agreement signed in June 2026.
Is SpaceX's AI division profitable?
No. On $2.6 billion in quarterly revenue, it posted a $1.5 billion loss; the company is investing heavily in expanding its AI infrastructure.
Access Frontier LLMs & Multimodal AI in One Workspace
Discover NomadNet AI (nomadnet.ai) — unified access to frontier language models, research synthesis, and multimodal workflows.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.