Although Elon Musk may have presented the market with a vision of space-driven Artificial Intelligence, Wall Street analysts indicate that SpaceX's immediate and real value lies on Earth, where the company develops essential infrastructure for the growth of artificial intelligence, according to Reuters analysis.
Current Computing Revenues
The computing contracts signed by SpaceX with companies such as Anthropic, Google, and Reflection AI, intended for its Colossus supercomputer clusters, are projected to generate over US$ 28 billion in annual revenue. This amount exceeds the approximately US$ 3.2 billion in registered AI revenue for the company in 2025 and is higher than the individual revenues from rocket launch services and Starlink connectivity.
Investment in Terrestrial Infrastructure
In 2025, SpaceX directed an investment of nearly US$ 18 billion towards AI infrastructure and development. This figure includes US$ 12.7 billion in capital expenditures related to AI and US$ 5.1 billion in research and development activities, significantly exceeding spending on the space and connectivity segments.
J.P. Morgan projects that SpaceX will increase its terrestrial processing capacity to approximately 9 gigawatts by 2029. This capacity would be equivalent to about four times the energy produced by the Hoover Dam in the United States. The Colossus and Colossus II complexes together offer about 1 gigawatt of computational capacity, positioning SpaceX among the leading global AI infrastructure providers.
Expansion into Corporate Software
Additionally, analysts point to the acquisition of the AI coding startup Cursor for US$ 60 billion as proof that SpaceX is diversifying its operations, moving from mere infrastructure to corporate software. This allows the company to monetize both AI applications and the computing capacity that supports them.
Prospects for Orbital AI
Musk's vision of using computing satellites in orbit remains a long-term possibility, but it depends on technological advancements that have not yet been fully achieved. Bank of America analysts, cited by Reuters, stated that the long-term viability of orbital data centers has not yet been proven and is strongly tied to pending technological milestones.
Anthony Milovantsev, a partner at the consulting firm Altman Solon, commented to Reuters that the idea that orbital AI will drastically revolutionize terrestrial data centers is an exaggeration. He estimated that any change in terrestrial data centers is more than ten years away.
J.P. Morgan predicts that SpaceX will keep its terrestrial clusters operational after 2029, gradually supplementing them with orbital capacity. Although solar-powered satellites could reduce costs related to energy, cooling, and physical space—major challenges for conventional data centers—this possibility depends on Starship achieving very low launch costs and a high reuse rate.