Key Points:
- Wall Street analysts project that SpaceX’s near-term AI financial payoff remains firmly tied to Earth-based data centers, not space.
- Deals with Anthropic, Google, and Reflection AI for its Colossus supercluster are expected to generate over $28 billion in annual revenue.
- The projected $28 billion in terrestrial compute revenue far surpasses SpaceX’s individual launch and Starlink connectivity business segments.
- J.P. Morgan expects SpaceX to expand its terrestrial AI compute capacity to 9 gigawatts by 2029 before transitioning to orbital projects.
The financial community believes the near-term payoff for the commercial space industry remains firmly rooted on Earth, despite corporate leadership frequently pitching a highly speculative future where solar-powered satellites run massive data centers in outer space. The primary financial driver for the newly public aerospace giant is its rapid construction of land-based digital infrastructure. By building and leasing out massive computing clusters on the ground, the firm has turned its computational overhead into a highly lucrative commercial business. This strategic pivot positions the company as a dominant player in the physical plumbing of the artificial intelligence boom, even as its ambitious orbital concepts remain a distant, decade-long goal.
The physical reality of this terrestrial strategy is delivering extraordinary financial rewards. The company’s newly signed hosting and compute contracts with major tech firms—including Google, Anthropic, and open-source startup Reflection AI—are expected to generate more than $28 billion in annualized revenue. This massive computing pipeline easily eclipses the firm’s 2025 artificial intelligence revenue of approximately $3.2 billion. Crucially, the $28 billion figure also surpasses the annual sales of the company’s traditional launch and Starlink satellite connectivity business segments individually, making terrestrial computing the fastest-growing and most profitable division within the entire corporation.
This rapid commercialization follows an unprecedented wave of internal capital deployment. The firm poured nearly $18 billion into artificial intelligence infrastructure and development in 2025 alone. This capital allocation included approximately $12.7 billion in direct, AI-related capital expenditures and an additional $5.1 billion dedicated to advanced software research and development. This massive budget dwarfed the company’s spending on its core launch vehicles and satellite manufacturing networks combined, proving that the firm restructured its long-term financial priorities long before its public market debut.
The physical hub of this computing power is the massive Colossus supercomputer complex located in Memphis, Tennessee. Originally built at a breakneck speed to train the firm’s own conversational software, the underutilized capacity of the campus has recently transitioned into a highly profitable, public-facing rental product. The campus houses approximately 555,000 state-of-the-art Nvidia graphics processors, acquired for an estimated $18 billion. To keep these power-hungry processors running continuously without straining the local municipal grid, the company deployed 19 on-site natural gas turbines capable of generating 350 megawatts of electricity, creating an independent, self-contained energy oasis.
The commercial viability of this massive campus was recently highlighted by a landmark, long-term leasing agreement. Open-source artificial intelligence startup Reflection AI agreed to pay the company a staggering $150 million every month, starting July 1, 2026, for daily access to the newest Nvidia GB300 chips inside the Memphis facility. If the contract runs its full course through 2029, the cumulative payout will reach approximately $6.3 billion. For a pre-revenue startup that has yet to release a commercial model, committing such a massive sum proves how desperately artificial intelligence labs require immediate access to high-performance, ready-to-run silicon.
Looking ahead, the company is preparing to scale its physical footprint to unprecedented heights. The firm will expand its terrestrial computing capacity to approximately 9 gigawatts by 2029. To put that scale into perspective, a 9-gigawatt power pipeline is roughly equivalent to four times the electrical output generated by the Hoover Dam. This massive land-based capacity will drive the company’s overall revenue and earnings growth for the remainder of the decade, providing a highly reliable, high-margin cash engine that can comfortably fund its more speculative space initiatives.
This massive terrestrial expansion effectively debunks the speculative narrative that space-based computing will soon make Earth-bound data centers obsolete. While placing servers in low Earth orbit offers seductive advantages—including abundant solar energy and frictionless vacuum cooling—the physical and financial hurdles of operating in space remain extreme. The average cost of constructing, launching, and maintaining an industrial-scale data center in orbit is currently three times higher than on the ground. Any meaningful commercial displacement of land-based data centers is at least ten to fifteen years away, making the orbital concept a long-term research project rather than a near-term revenue driver.
Furthermore, operating highly sensitive, high-performance electronics in the harsh environment of space introduces severe technical and regulatory challenges. Outer space is an unforgiving environment where cosmic radiation constantly threatens to damage microprocessors, and orbiting space debris poses a continuous risk of physical destruction. In addition, international space treaties designed in the late 20th century do not contain clear legal frameworks for digital property rights, data sovereignty, or tax liabilities in low Earth orbit. Until developers can resolve these complex technical and regulatory hurdles, global financial institutions will remain highly hesitant to fund space-based server farms.
Ultimately, the massive success of the company’s terrestrial computing business demonstrates its incredible agility and industrial strength. By leveraging its unique ability to rapidly procure, construct, and power massive server clusters on the ground, the firm has secured an irreplaceable position in the global artificial intelligence economy. While the long-term vision of a spacefaring, solar-powered digital network remains an exciting prospect for the next decade, the company’s immediate financial viability relies on the massive, land-based data centers humming in Tennessee. The coming years will reveal how successfully the firm can ramp its capacity from the ground to orbit, but for now, the physical foundations of its AI empire are firmly tethered to Earth.
