Infrastructure providers, especially data centers, are set to be significant beneficiaries of the accelerating adoption of AI technology. This spans applications from software development and robotics to everyday activities like shopping and scheduling.
SpaceX, a key player in the AI landscape, has begun to monetize computing power by forming partnerships with enterprise clients, such as Anthropic, for its Colossus supercomputer clusters. Conversely, its orbital AI initiative is still a long-term prospect reliant on the development of Starship rockets, reduced launch expenses, and technological advancements.
Documents reviewed by Reuters and insights from research firms reveal that revenues from new computing contracts are expected to substantially surpass income from other areas this year.
“The idea that (orbital) will radically disrupt terrestrial data centers is somewhat exaggerated. Any displacement of land-based data centers is more than a decade away,” remarked Anthony Milovantsev, a partner at consulting firm Altman Solon.
J.P. Morgan anticipates that SpaceX will scale its terrestrial AI compute capacity to approximately 9 gigawatts by 2029, which is roughly equivalent to four times the energy output of the Hoover Dam.
“Post-2029, we expect SpaceX to shift focus to orbital computing for additional capacity, while still managing and operating its terrestrial compute networks,” the brokerage added.
SpaceX’s partnerships with Anthropic, Google from Alphabet, and Reflection AI concerning its Colossus compute facilities are projected to yield over $28 billion in annual revenue.
This figure significantly surpasses SpaceX’s forecasted AI revenue of around $3.2 billion by 2025, and it exceeds earnings from both its launch services and Starlink connectivity individually, based on calculations by Reuters.
Analysts noted that these contracts include termination clauses and should not be considered assured long-term recurring income.
In 2025, SpaceX is expected to invest nearly $18 billion in AI infrastructure and development, which includes about $12.7 billion in AI-related capital expenditures and $5.1 billion in AI research and development—far exceeding expenditures on its space and connectivity ventures, per company documents analyzed by Reuters.
Colossus and Colossus II together deliver roughly one gigawatt of AI compute capacity, positioning SpaceX among the largest AI compute providers globally. Analysts predict this presence will multiply significantly over the coming years.
Several firms also highlight SpaceX’s $60 billion acquisition of AI coding startup Cursor as proof the company is branching out beyond infrastructure into enterprise software, enabling it to monetize both AI applications and the cloud computing capabilities that support them.
Orbital AI remains a long-term bet
Most brokerages that began coverage of SpaceX after its IPO view orbital AI as a longer-term opportunity, focusing their short-term financial projections on terrestrial AI infrastructure.
“We believe the long-term feasibility of orbital data centers is untested and highly contingent on crucial technological advancements that have not yet been fully achieved,” noted analysts from BofA.
Analysts generally agree that terrestrial computing will fuel SpaceX’s growth and profitability through the end of this decade, with orbital AI hinging on Starship achieving rapid reusability, cost reductions for launches, and progress in satellite technology.
The majority indicate that Starship might eventually enable the deployment of solar-powered computing satellites, potentially lowering costs related to energy, cooling, and land—significant concerns for terrestrial data centers.
The discussion, analysts assert, has shifted from whether SpaceX can develop and monetize AI infrastructure to how swiftly it can transform that opportunity into a lucrative enterprise beyond Earth’s boundaries.
