SpaceX's AI compute division grows to $2.6B in revenue amid aggressive neocloud expansion

SpaceX reported AI division revenue of $2.6 billion for its latest quarter—a 3x increase from the prior-year period—driven predominantly by compute supply agreements with Anthropic and Google, according to SpaceX’s quarterly earnings disclosure reported by The Verge AI. Despite the revenue surge, the division remains operationally unprofitable, recording $1.5 billion in losses, a marginal improvement from $1.6 billion in losses in the same quarter last year. The company’s overall net loss narrowed to $143 million as scaling revenue offset rising capital intensity.

SpaceX’s Pivot to Neocloud Infrastructure

SpaceX is positioning itself as a direct competitor to established neocloud providers like CoreWeave by monetizing compute capacity through enterprise contracts. The May 2026 Anthropic deal and June 2026 Google partnership represent the company’s first major wins in this vertical. According to The Verge AI, SpaceX highlighted the AI division as the principal driver of enterprise valuation in its public-company filings—a bet that positions compute services as a core business rather than a secondary revenue stream.

The neocloud strategy requires substantial capital commitments. SpaceX’s capital expenditures reached $18.37 billion, reflecting dual investment tracks: AI data center infrastructure expansion and advancement of the Starship heavy-lift vehicle, which Elon Musk has positioned as essential to deploying next-generation satellite payloads and theoretical space-based compute assets.

Starship Development and Connectivity Expansion

SpaceX’s space division costs increased by $389 million year-over-year, with Starship as the primary driver, The Verge AI reports. The vehicle is foundational to Musk’s plan to scale Starlink—currently the company’s only profitable business segment—into a broader connectivity and eventually compute infrastructure play. SpaceX has already manufactured and launched 20 new Starlink satellites designed for heavier payload capacity, according to the earnings documents.

The full deployment target is 60 satellites deployed simultaneously, though The Verge AI notes the timeline for complete rollout remains unclear. This phased expansion suggests Starlink revenue growth will accelerate incrementally rather than unlock a revenue cliff.

Why This Matters

SpaceX’s aggressive neocloud expansion signals that vertically integrated space-infrastructure companies are willing to absorb near-term operating losses to capture market share in enterprise AI compute. For Anthropic and Google—both facing GPU scarcity and rising inference costs—SpaceX’s deal terms likely offered either novel satellite-based compute topologies, favorable pricing via Musk’s capital deployment, or both.

However, The Verge AI notes that despite beating analyst estimates, SpaceX’s stock declined after an initial surge, suggesting investor skepticism about the path to neocloud profitability. Teams evaluating multi-year compute partnerships should monitor whether SpaceX’s capital intensity continues to grow faster than revenue, and whether Starship delays push expected space-based compute deployments beyond 2027.