Starcloud has raised $170 million in a Series A round that values the company at $1.1 billion, making it one of the fastest startups to hit unicorn status after graduating from Y Combinator. The round was led by Benchmark and EQT Ventures, TechCrunch AI reports, closing just 17 months after the company’s demo day presentation.
The pitch: build data centers in space.
The reality: it’s going to take a while.
What Starcloud Has Done So Far
The company launched its first satellite carrying an Nvidia H100 GPU back in November 2025. That satellite was used to train an AI model in orbit, a first, and run a version of Gemini. Total funding now stands at $200 million.
Later this year, Starcloud plans to launch a more powerful version, Starcloud 2, featuring:
- Multiple GPUs, including an Nvidia Blackwell chip
- An AWS server blade
- A bitcoin mining computer
- The largest deployable radiator ever flown on a private satellite
The Big Bet on Starship
Here’s where it gets interesting. Starcloud’s real play isn’t its current satellites. It’s Starcloud 3, a 200-kilowatt, three-ton spacecraft designed to launch from SpaceX’s Starship rocket using the same “PEZ dispenser” deployment system that launches Starlink satellites.
CEO Philip Johnston told TechCrunch AI that Starcloud 3 would be the first orbital data center that’s cost-competitive with ground-based facilities, with power costs around $0.05 per kWh. But there’s a catch: that only works if commercial launch costs hit roughly $500 per kilogram.
Starship isn’t flying commercially yet. Johnston expects access to open up in 2028 or 2029, and admits the math doesn’t work without it. “We’re not going to be competitive on energy costs until Starship is flying frequently,” he said.
Why This Matters
The interest in space-based compute is growing because terrestrial data centers face mounting obstacles. Energy constraints, permitting battles, political resistance, and water usage are all slowing construction on the ground. Space, in theory, offers unlimited solar power and natural cooling through radiators.
But the gap between ambition and reality is enormous. According to TechCrunch AI, the number of advanced GPUs currently in orbit is measured in the dozens, while Nvidia sold nearly 4 million to terrestrial hyperscalers in 2025 alone. SpaceX’s entire Starlink constellation of 10,000 satellites produces about 200 megawatts. U.S. data centers with over 25 gigawatts of capacity are currently under construction.
Those numbers tell you how early this industry really is.
The Competition
Starcloud isn’t alone. Several companies are chasing the same vision:
- Aetherflux is developing space data center infrastructure
- Google’s Project Suncatcher is exploring orbital compute
- Aethero launched Nvidia’s first space-based Jetson GPU in 2025
- SpaceX has applied for permission to build and operate a million satellites for distributed compute
That last one is the real wildcard. Going up against SpaceX is a tough proposition for any startup. Johnston sees room for coexistence, arguing that SpaceX is mainly building to serve Grok and Tesla workloads, while Starcloud positions itself as “an energy and infrastructure player” offering third-party cloud services.
The Hard Lessons
Johnston’s team has already learned that space is unforgiving. One GPU, an Nvidia A6000, failed during launch. The H100 worked but isn’t ideal for space, Johnston admitted. “The reason we did it is we wanted to prove that we could run state-of-the-art terrestrial chips in space,” he told TechCrunch AI.
What stands out here is the honest assessment of the timeline. Most space data center companies acknowledge that large-scale training workloads, which require hundreds or thousands of GPUs working in sync, won’t happen in orbit until much later. Simpler inference tasks will come first. The synchronization challenge alone, either building massive single spacecraft or reliable laser links between formations, remains unsolved.
This is a long game. Starcloud has early traction and real hardware in orbit, which puts it ahead of most competitors. But the entire business model hinges on rockets that don’t exist yet flying at prices no one has achieved. Worth watching, but the hard part hasn’t started.
Full details are available in the original report from TechCrunch AI.