The race to move computing infrastructure off Earth just got a major financial boost. Starcloud, a space compute startup, has closed a $170 million Series A — pushing its valuation to $1.1 billion. That makes it one of the fastest companies to reach unicorn status after graduating from Y Combinator.
The round closed just 17 months after Starcloud’s Y Combinator demo day. Benchmark and EQT Ventures led the investment. The company has now raised $200 million in total funding.
Why Investors Are Betting on Space Data Centers
Data center development on Earth is slowing down. Power shortages, land constraints, and regulatory hurdles are all creating friction. Investors are increasingly looking upward — literally.
Starcloud sits at the center of that trend. The company builds satellites equipped with high-performance GPUs. The goal is to run cloud computing workloads from orbit. It sounds futuristic. But Starcloud has already launched its first satellite.
In November 2025, the company put a satellite carrying an Nvidia H100 GPU into orbit. That satellite was used to train an AI model in space — a claimed industry first. It also ran a version of Google’s Gemini model. These are not small achievements. They represent real proof-of-concept milestones in a field that has more ambition than track record.
Meet the Founder: Philip Johnston
Philip Johnston, Starcloud’s CEO and founder, built the company around a clear thesis. Space-based computing will eventually be cost-competitive with ground-based data centers. Getting there requires the right rockets, the right chips, and the right timing.
Johnston is direct about where the business stands today. “An H100 is probably not the best chip for space, to be honest,” he told TechCrunch. “But the reason we did it is we wanted to prove that we could run state-of-the-art terrestrial chips in space.”
That hands-on learning is invaluable. One GPU — an Nvidia A6000 — failed during launch. That kind of hard data shapes every future design decision. No simulation replaces it.
What Starcloud Is Building Next
The company is not standing still. Three generations of hardware are either in progress or on the drawing board.
Starcloud 2 launches later this year. It carries multiple GPUs, including an Nvidia Blackwell chip and an AWS server blade. It also includes a bitcoin mining computer. Notably, it will feature the largest deployable radiator ever flown on a private satellite — a necessity for keeping hot-running chips from failing in the thermal extremes of space.
Starcloud 3 is the bigger bet. It is a 200-kilowatt, three-ton spacecraft designed to launch aboard Starship — the heavy-lift rocket built by Elon Musk’s SpaceX. Johnston says Starcloud 3 will be the first orbital data center that is genuinely cost-competitive with terrestrial alternatives. He estimates costs around $0.05 per kilowatt-hour of power, assuming commercial launch costs settle around $500 per kilogram.
The Starship Dependency
Here is the challenge: Starship is not flying commercially yet. Johnston expects commercial access to open up in 2028 and 2029. Until then, the company will continue launching smaller satellites on SpaceX’s Falcon 9.
“We’re not going to be competitive on energy costs until Starship is flying frequently,” Johnston said plainly. “If it ends up being delayed, we’ll just carry on launching the smaller versions.”
This is not unique to Starcloud. Every major space data center project faces the same bottleneck. Powerful, affordable space computing depends on a new generation of rockets flying at high frequency. That cadence may not arrive until the 2030s.
Two Business Models, One Long Game
Johnston describes Starcloud’s approach in two phases.
Right now, the company sells processing power to other spacecraft already in orbit. Its first satellite, for example, analyzes radar data collected by Capella Space. This is a real, revenue-generating use case that exists today.
In the future — once launch costs drop significantly — the vision expands. Distributed space data centers could handle workloads that currently run in terrestrial facilities. Think inference tasks first, then eventually larger training jobs.
That second phase requires solving a significant engineering problem: synchronization. Large AI training runs need hundreds or thousands of GPUs working together simultaneously. In space, that means either building enormous single spacecraft or developing reliable high-speed laser links between satellites flying in formation. Most experts expect that capability to arrive well after simpler workloads are already running in orbit.
How Small Is the Space GPU Market Right Now?
The scale gap between space and ground computing is worth putting in perspective.
Nvidia is estimated to have sold nearly 4 million GPUs to terrestrial hyperscalers in 2025 alone. The total number of advanced GPUs currently operating in orbit is measured in the dozens.
SpaceX’s Starlink network — the largest satellite constellation ever built, with 10,000 spacecraft — generates roughly 200 megawatts of power. Meanwhile, data centers currently under construction in the United States carry a combined capacity exceeding 25 gigawatts, according to research from Cushman & Wakefield.
The gap is enormous. But Johnston believes that gap is exactly the opportunity.
The Competition Is Already Forming
Starcloud is not alone in this space. Aetherflux, Google’s Project Suncatcher, and Aethero — which launched Nvidia’s first space-based Jetson GPU in 2025 — are all developing competing space computing businesses.
Then there is the biggest player of all: SpaceX itself has asked the U.S. government for permission to build and operate one million satellites for distributed space computing.
Johnston is not rattled. He sees a clear lane for Starcloud that SpaceX is unlikely to enter directly.
“They are building for a slightly different use case than us,” Johnston said. “They’re mainly planning on serving Grok and Tesla workloads. What I think they are unlikely to do is what we’re doing — as an energy and infrastructure player.”
It is a classic market positioning argument. SpaceX serves its own ecosystem. Starcloud serves everyone else.
Even Nvidia Is Moving Carefully Here
The immaturity of this market showed up at Nvidia’s annual GPU Technology Conference last week. CEO Jensen Huang unveiled the company’s new Vera Rubin Space-1 chip modules to significant fanfare. What he did not mention: none had been produced yet, and none had been shared with development partners.
That detail speaks volumes. Even the world’s leading GPU company is still in early stages when it comes to space-optimized silicon. Starcloud, with actual hardware in orbit and real operational data in hand, may have a more grounded view of what works than most.
The Enterprise Angle
For enterprise decision-makers, the Starcloud story is not just a space curiosity. It is an early signal of where edge computing infrastructure could be heading.
As AI workloads grow and terrestrial power capacity tightens, the economics of space-based computing will only become more relevant. The hyperscalers are already watching. AWS has a server blade aboard Starcloud 2. That is not a coincidence.
The timeline is long. The technical hurdles are real. But with $200 million raised, proven hardware in orbit, and major investors behind it, Starcloud is building the foundation now — before the window fully opens.