SpaceX's chief financial officer believes the company's fully reusable Starship rocket could make computing in orbit cheaper than running data centers on Earth within the next year, fundamentally reshaping where artificial intelligence infrastructure gets built. The claim hinges on SpaceX achieving something it has never done before: reflying both the first and second stages of Starship, which would dramatically lower launch costs and make orbital compute economically viable at scale.

Why Does Orbital Computing Matter More Than You Might Think?

Terrestrial data centers face mounting pressure. Power consumption, cooling systems, real estate, and construction costs keep climbing, making it increasingly expensive to build and operate the massive facilities that train and run artificial intelligence models. Orbital infrastructure, by contrast, could sidestep many of these constraints. If SpaceX can reduce the cost to launch cargo to space dramatically enough, companies might find it cheaper to deploy computing power in orbit than to build new data centers on the ground.

SpaceX's prospectus outlines an ambitious vision: Starship V3 is expected to carry 100 metric tons to low Earth orbit, and the company believes full and rapid reusability could eventually reduce launch costs by 99 percent or more compared to NASA's historical average of $18,500 per kilogram. That kind of cost reduction would be transformative.

"When you get into, hopefully as soon as next year, where you're reflying both first and second stage of Starship, the cost curve goes down dramatically," said Bret Johnsen, CFO at SpaceX.

Bret Johnsen, CFO at SpaceX

How Does Starship Reusability Compare to What Competitors Are Doing?

SpaceX already has more than a decade of experience reflying Falcon 9 boosters, giving it a significant head start. The company's track record with booster recovery and reuse demonstrates that the concept works at scale. However, Starship is far more ambitious because it aims to refly both stages, not just the first stage booster.

China's private space industry is making progress on reusable rockets. LandSpace successfully landed the first stage of its Zhuque-3 orbital-class rocket last month, becoming China's first private launcher to achieve ground-based orbital-booster recovery. The methane-fueled rocket can carry about 14.2 metric tons to low Earth orbit. While impressive, this achievement still falls short of Starship's capabilities and ambitions.

"I think the Chinese are trying to get there, right? To their credit, they have got an amazing space program. But not on a Starship level anytime soon, let us put it that way," said Bret Johnsen.

Bret Johnsen, CFO at SpaceX

What Would Make Orbital Computing Practical?

Three key factors would need to align for orbital compute to become economically competitive with terrestrial infrastructure:

• Launch Cost Reduction: Full Starship reusability must bring per-kilogram launch costs down significantly, making it affordable to deploy and maintain computing hardware in orbit.
• Satellite Deployment Efficiency: As launch costs fall, the cost to deploy satellites and computing infrastructure to orbit drops proportionally, improving the overall economics of orbital operations.
• Terrestrial Cost Escalation: Power, cooling, real estate, and construction expenses on Earth continue rising, making orbital infrastructure relatively more attractive by comparison.

SpaceX executives have previously argued that Starship could make large-scale orbital compute practical as terrestrial data centers face power, land, and cooling constraints. The company sees reusable rocketry as a way to "drive our cost of compute dramatically lower than others in the industry".

When Could This Actually Happen?

The timeline is aggressive but not unrealistic given SpaceX's recent progress. CEO Elon Musk has said the company could attempt its first upper-stage tower catch within months, with a first reflight of both stages possible by late 2026 or early 2027. If those milestones hit, orbital compute could reach cost parity with terrestrial infrastructure "maybe as soon as next year," according to Johnsen.

The challenge is proving that full Starship reuse actually works at the scale and frequency SpaceX envisions. The company has successfully recovered and reflown Falcon 9 first stages hundreds of times, but Starship's upper stage is a different beast. Getting both stages back safely and ready for rapid reuse would be a watershed moment for the space industry and could unlock entirely new business models for computing infrastructure.

If SpaceX pulls this off, it would represent a fundamental shift in how the world thinks about data center placement and AI infrastructure deployment. Instead of building massive facilities on Earth, companies might lease computing power from orbital platforms, fundamentally changing the economics of the technology industry.