SpaceX Enters the AI Compute Race With Rocket-Grade Ambition
The numbers came first, and they were hard to ignore. SpaceX reported AI revenue of $2.6 billion in the second quarter, up 213% sequentially, and disclosed $14.1 billion in contracted cloud services agreements — figures that landed on SpaceX's inaugural earnings call as a publicly traded company.
Total company-wide capital expenditure reached $18.4 billion in the second quarter, up from $10 billion in the first quarter. Executives told investors that spending is expected to hold at roughly the current level over the next two quarters, signaling that SpaceX is treating its AI infrastructure push as a permanent strategic commitment, not a one-cycle bet.
The Musk Thesis: Rocket Science as Competitive Moat
CEO Elon Musk used the call to make an unusually blunt case for why SpaceX can outcompete established cloud giants. 'We're taking a small amount of the expertise that we use for rockets and satellites, and applying that to scaling terrestrial data centers,' he said, describing the company's engineering advantage as 'the New York Yankees going in and playing a Little League team.'
Data centers are a 'trivial problem' compared to what SpaceX engineers handle daily, Musk argued — noting that every rocket flight involves keeping a vehicle from doing what it 'desperately wants to do,' which is blow itself apart. Applying even a fraction of that discipline to AI infrastructure, he said, produces 'an amazing outcome.'
Musk made the comparison without naming specific rivals. SpaceX is a late entrant into a market long dominated by Amazon, Microsoft, and Alphabet's Google — all three of which reported record growth in their cloud divisions in their most recent earnings. A wave of 'neocloud' startups is also competing aggressively for AI compute contracts.
Nvidia Partnership and the Gigawatt Targets
A key pillar of SpaceX's AI strategy is its access to Nvidia hardware. Musk confirmed that SpaceX will build its AI computing platform exclusively on Nvidia's Vera Rubin architecture, calling it the industry's best AI computing system. 'Our understanding with Nvidia is that we will receive a very significant percentage of their GPUs next year,' he said.
The company ended the second quarter with 1.4 gigawatts of nameplate compute capacity. SpaceX told investors it expects to exceed 2 gigawatts by year-end. Pressed further by an analyst, Musk said the company's 'tentative target' is 20 gigawatts of power and cooling capacity by the end of 2027 — and that even accounting for delays, SpaceX would likely land 'close to 15 gigawatts.'
Among the contracts anchoring SpaceX's cloud business are deals with Anthropic and with Google. The company entered the AI space in February when it acquired xAI, an AI company also owned by Musk.
CEO Times Take
The market has already voted on what happens when a company with genuine engineering density enters a capital-intensive industry: it competes on unit economics that incumbents cannot easily replicate. SpaceX's $14.1 billion contracted backlog and its Nvidia GPU allocation are not marketing claims — they are hard assets that translate directly into future cash flow.
Free enterprise rewards exactly this kind of vertical integration: proprietary talent, secured supply chain, and a customer list that includes Google and Anthropic. Whether Musk's 20-gigawatt target proves achievable or not, the Q2 numbers confirm that SpaceX is no longer a rocket company dabbling in AI. It is an AI infrastructure company that also happens to launch rockets — and that distinction is worth watching very closely.
