SpaceX Aims for $100B Revenue Through AI Compute Expansion

SpaceX is scaling its AI infrastructure by using a new hosting agreement to reach $100 billion in annual recurring revenue. The firm is using orbital compute to avoid terrestrial power and regulatory bottlenecks. It is expanding its compute capacity to reach 10 gigawatts of terrestrial power by the end of 2025.

"We could not find supply-chain partners that could match its pace or quality standards," said Bret Johnsen, Chief Financial Officer at SpaceX, during the Goldman Sachs Communacopia + Technology Conference 2026.

Scaling Infrastructure and Revenue

The company’s latest hosting agreement is projected to generate roughly $1.1 billion in monthly recurring revenue beginning December 1, 2026. This financial push relies on an increase in terrestrial compute capacity. The company intends to grow this capacity from just over 2 gigawatts at the end of 2024 to between 5 and 10 gigawatts by the end of 2025.

SpaceX currently carries a market capitalization of approximately $2 trillion, according to InvestingPro data. This valuation reflects the company's transition from an aerospace manufacturer to an infrastructure provider. Johnsen attributed this shift to the necessity of internalizing supply chains to meet production timelines.

The Pivot to Orbital Compute

Terrestrial expansion remains the immediate priority. SpaceX is also developing orbital compute capabilities as a long-term strategy to bypass the power-grid constraints and permitting delays that limit land-based data center construction. The company has targeted 2025 for its first satellite launches dedicated to this mission, with plans for larger-scale orbital deployments continuing through 2028.

Strategic Implications

For the tech sector, SpaceX's expansion represents a shift in how high-performance computing environments are architected. By integrating launch capabilities with data center hosting, the company aims to offer a compute solution that is decoupled from traditional terrestrial utility infrastructure. Because power demand for AI training strains regional grids, the ability to deploy satellite-based compute could provide an alternative for data-intensive applications. This depends on the company successfully navigating the technical challenges of space-based thermal management and latency.