Space Data Centers Are Here — And Starcloud Just Raised $170M to Lead the Race
Space is no longer just about exploration. It is becoming the next frontier for computing power, and one startup is moving faster than anyone expected. Starcloud has raised $170 million in a Series A funding round, reaching a $1.1 billion valuation just 17 months after its Y Combinator demo day. The round was led by Benchmark and EQT Ventures, bringing its total funding to $200 million.
 |
| Credit: Starcloud |
Why Space Data Centers Are Suddenly Big Business
The idea of moving data centers to orbit might sound like science fiction, but the pressure on Earth-based infrastructure is making it a serious commercial proposition. Energy costs are rising, political friction is slowing permits, and the land required for hyperscale data centers is becoming harder to secure. Investors are betting that the economics of space computing will eventually catch up, and Starcloud is positioning itself at the front of that wave.
What the company is building is not a single satellite. It is an evolving architecture of orbital compute platforms designed to grow in capability as rocket technology matures. That long game is exactly what attracted institutional backing at this scale.
From Y Combinator to Unicorn Status in Under Two Years
Starcloud's rise is one of the fastest in recent startup history. The company went from demo day to unicorn status in just 17 months, a trajectory that reflects both the urgency investors feel around AI infrastructure and the credibility of the team executing the vision.
In November 2025, Starcloud launched its first satellite carrying an Nvidia H100 GPU, making it the first company to deploy a state-of-the-art terrestrial graphics chip in orbit. The satellite was used to train an AI model in space, a claimed industry first, and also ran a version of Gemini. Those milestones gave the company proof-of-concept credibility that is rare at this stage of a frontier technology.
CEO and founder Philip Johnston acknowledged the H100 is not optimally designed for space conditions, but said deploying it was deliberate. "We wanted to prove that we could run state-of-the-art terrestrial chips in space," he explained. That learning came at a cost: a second GPU, an Nvidia A6000, failed during launch, underscoring how brutal the operational environment still is.
What Starcloud 2 and Starcloud 3 Will Actually Do
The next satellite, Starcloud 2, is scheduled to launch later this year. It will carry multiple GPUs, including an Nvidia Blackwell chip, an AWS server blade, and a bitcoin mining computer. It will also feature the largest deployable radiator ever flown on a private satellite, a critical engineering milestone given how much heat advanced chips generate in a vacuum.
Starcloud 3 is an even bigger leap. Designed to launch aboard SpaceX's Starship rocket, it will be a 200-kilowatt, three-ton spacecraft configured to fit the same deployment system SpaceX uses for Starlink satellites. Johnston believes Starcloud 3 will be the first orbital data center genuinely cost-competitive with ground-based alternatives, targeting power costs of around $0.05 per kilowatt-hour, assuming commercial launch prices settle near $500 per kilogram.
The catch is that Starship commercial access may not open until 2028 or 2029. Johnston is pragmatic about this. "If it ends up being delayed, we'll just carry on launching the smaller versions on Falcon 9," he said. "We're not going to be competitive on energy costs until Starship is flying frequently."
Two Revenue Streams, One Long-Term Vision
Starcloud's current business model operates on two layers. In the near term, the company sells processing power to other spacecraft on orbit. Its first satellite is already doing meaningful work, analyzing data for a radar imaging company operating its own constellation. That is real, recurring revenue derived from an asset already in space.
The longer-term play is more ambitious. As launch costs decline and the company deploys more capable hardware, Starcloud wants to pull workloads from terrestrial hyperscalers, offering space-based compute as a cheaper or more strategically resilient alternative. That second model depends on infrastructure that does not yet fully exist, but the groundwork is being laid now.
Johnston is also clear-eyed about the synchronization challenge. Training large AI models requires hundreds or thousands of GPUs working in tight coordination. Doing that in space either demands extraordinarily large spacecraft or ultra-reliable laser links between vehicles flying in formation. Most industry experts expect inference tasks to come first, with large-scale training workloads following much later.
The Space Compute Race Is Heating Up
Starcloud is not operating in a vacuum, figuratively speaking. Several well-resourced players are now pursuing similar ambitions. Aetherflux, Aethero, and a Google-backed initiative called Project Suncatcher are all developing orbital compute strategies. Aethero notably launched Nvidia's first space-based Jetson GPU in 2025.
Then there is SpaceX, which has petitioned the U.S. government for permission to operate a constellation of one million satellites for distributed space compute. Competing directly with SpaceX is an intimidating proposition by any measure.
Johnston, however, sees a path to coexistence rather than collision. "They are building for a slightly different use case than us," he said. His view is that SpaceX's orbital compute strategy is primarily designed to serve its own internal workloads, including Grok and Tesla applications, and is unlikely to build out the kind of open energy and infrastructure platform Starcloud is targeting.
Whether that assessment holds as the market matures remains to be seen, but it reflects a coherent positioning strategy for a company that cannot afford to fight every battle simultaneously.
How Far Away Is Practical Space Computing, Really?
The honest answer is that meaningful scale is still years away. The gap between orbital and terrestrial compute is enormous by almost every measure today. The entire Starlink network, with roughly 10,000 spacecraft, generates around 200 megawatts of power. Meanwhile, data centers under construction in the United States alone are expected to draw more than 25 gigawatts when completed.
The number of advanced GPUs currently in orbit is measured in the dozens. Nvidia is estimated to have shipped nearly four million units to terrestrial hyperscalers in 2025 alone. Even Nvidia's own Vera Rubin Space-1 chip modules, unveiled with considerable fanfare at a recent GPU Technology Conference, had not yet been produced or distributed to development partners at the time of the announcement.
These are sobering figures, but they also represent the opportunity. The infrastructure does not yet exist because the technology and economics have not been ready. That moment of readiness is approaching, and companies like Starcloud are racing to be positioned when it arrives.
What This Funding Round Signals for the Industry
A $170 million Series A led by Benchmark and EQT Ventures is not a speculative bet on a distant future. It is a statement that serious investors believe the foundational work being done now, in satellite design, chip hardening, thermal management, and orbital operations, will translate into a defensible market position before the window closes.
Starcloud's path forward is still laden with technical risk, capital intensity, and dependence on third-party launch infrastructure that is not yet operating at the required cadence. But the company has something most competitors do not: a real satellite in orbit, with a real GPU, doing real work, with real lessons learned from failure. In frontier technology, that is worth more than any roadmap.
The race to put data centers in space is no longer theoretical. It is funded, it is launched, and it is accelerating.
