SpaceX's artificial intelligence empire is being built on the ground, not in the sky. Wall Street analysts say the company's immediate AI revenue will come from Earth-based data centers rather than the orbital computing infrastructure that Elon Musk has long promised. This reframes SpaceX as primarily an infrastructure company focused on powering AI models today, while space-based computing remains a speculative long-term bet.

Where Is SpaceX Actually Making Money From AI?

SpaceX has signed enterprise contracts for its Colossus supercomputer clusters with three major clients: Anthropic, Alphabet's Google, and Reflection AI. These deals alone are expected to generate more than $28 billion annually. To put this in perspective, SpaceX's total AI revenue in 2025 was approximately $3.2 billion, meaning these new contracts represent a dramatic expansion of the company's compute business. When counted separately, these compute contracts have already outperformed SpaceX's earnings from rocket launches and Starlink, the company's satellite internet service.

However, analysts caution that these contracts include termination clauses, so they should not be treated as guaranteed recurring revenue. The deals signal demand for SpaceX's computing capacity, but they are not ironclad commitments stretching indefinitely into the future.

SpaceX has invested heavily to support this business. In 2025, the company spent approximately $18 billion on AI infrastructure and research. This breaks down into roughly $12.7 billion in capital expenditures for building and upgrading data centers and $5.1 billion on research and development. Notably, this spending outpaced investment in SpaceX's space and connectivity divisions.

How Powerful Is SpaceX's Current Computing Infrastructure?

Colossus and a second cluster called Colossus II together provide approximately 1 gigawatt of AI compute capacity. This makes SpaceX one of the world's largest compute operators. To understand the scale, J.P. Morgan predicts that total terrestrial AI capacity will reach approximately 9 gigawatts by 2029, which is equivalent to four times the output of the Hoover Dam. SpaceX's current infrastructure already represents a significant slice of the global AI compute pie.

Beyond hardware, SpaceX is also moving into software. The company is acquiring Cursor, an AI coding startup, for $60 billion. Multiple brokerages cited this acquisition as evidence that SpaceX wants to sell AI applications and services, not just the computing machines that run them. According to reports, Musk's lab, now renamed SpaceXAI, built its Grok 4.5 model jointly with Cursor.

Why Is Orbital Computing Still Years Away?

Musk has long promoted a vision of computers running in space, avoiding the costs of ground-based data centers such as energy consumption, cooling systems, and land use. However, analysts view this as a much later chapter in SpaceX's AI story, not an immediate revenue driver. The timeline for orbital computing faces significant hurdles that have not yet been overcome at scale.

Several key technological milestones must be achieved before space-based computing becomes viable:

• Starship Launch Frequency: SpaceX must achieve frequent and cheap Starship launches to make orbital computing economically feasible.
• Satellite Technology: Better satellites designed specifically for computing workloads do not currently exist at the scale needed.
• Cost Reduction: Launch costs must fall dramatically to justify moving compute infrastructure into orbit.

"The narrative that orbital will fundamentally disrupt terrestrial data centers is a little bit overblown," said Anthony Milovantsev, a partner at consultancy Altman Solon, who estimated that any real displacement of ground-based data centers would take "ten years plus."

Anthony Milovantsev, Partner at Altman Solon

Bank of America analysts were even more direct, calling the long-term viability of orbital data centers "unproven and heavily reliant on key technological milestones that have yet to be realized". These are not minor engineering challenges; they represent fundamental unknowns about whether the concept will ever work at commercial scale.

What Does This Mean for SpaceX as an AI Investment?

The shift in focus from space to Earth reveals an important truth about SpaceX's current business model. The company is generating substantial revenue from compute infrastructure today, while orbital computing remains a speculative future opportunity. For investors or customers evaluating SpaceX as an AI play, the immediate value proposition is terrestrial data center capacity, not space-based computing.

Ground-based clusters will continue to operate and generate revenue regardless of whether orbital computing ever materializes. This makes the current Colossus business a more reliable revenue stream than the long-term space vision. If SpaceX's engineering breakthroughs do eventually enable cheap, frequent Starship launches and orbital computing becomes viable, that would represent an additional opportunity. But analysts are clear that this is a "if" scenario, not a "when" scenario.

The reality is that SpaceX has become a major player in the AI infrastructure market by building and operating data centers on Earth. The company's ability to attract contracts from Anthropic, Google, and other AI leaders demonstrates that its compute offerings are competitive today. Whether SpaceX can eventually extend that dominance into space remains one of the most ambitious and uncertain bets in the technology industry.