The data center capacity crisis just went extraterrestrial.
Starcloud, a Redmond, Washington startup building solar-powered AI data centers in orbit, raised $170 million in a Series A round led by Benchmark and EQT Ventures, reaching a $1.1 billion valuation. The round closes just 17 months after the company's Y Combinator demo day - making it the fastest graduate in YC's history to reach unicorn status. Total funding stands at $200 million.
What Starcloud Has Already Done
Founded in January 2024, Starcloud launched its first satellite, Starcloud-1, in November 2025 with just $3 million in pre-seed capital. The satellite carried an Nvidia H100 GPU - the most powerful compute ever deployed in orbit by approximately 100 times - and achieved several industry firsts: training an AI model in space, running inference on a version of Gemini, and completing model fine-tuning off-planet. The company is now processing data from Capella Space's radar satellites in orbit, its first commercial use case.
CEO Philip Johnston acknowledges the early skepticism was loud. When Starcloud announced, critics said it was impossible. The early results say otherwise.
Why Space for AI Infrastructure
The pitch is grounded in a real problem. Earth-based data center development is grinding against hard constraints - permitting delays, land scarcity, power grid limitations, and growing community resistance. Sanders and AOC's data center moratorium bill covered earlier this week reflects exactly this pressure at the political level.
Space removes those constraints entirely. In low Earth orbit, solar power is near-continuous and effectively free once deployed. Cooling is passive - waste heat radiates into deep space at around -270°C with no water required. No planning permissions, no grid connections, no land acquisition battles.
What Comes Next
Starcloud-2 launches in October 2026, carrying 100 times the power generation capacity of the first satellite, an Nvidia Blackwell B200 chip, an AWS Outposts server blade, and a Bitcoin mining computer. The satellite will carry the largest deployable radiator ever flown on a private satellite and will run commercial workloads for customers including Crusoe, AWS, Google Cloud, and Nvidia.
The longer-term roadmap is ambitious. Starcloud-3, designed to launch from Starship, will be a 200-kilowatt, three-ton spacecraft targeting cost parity with terrestrial data centers at around $0.05 per kilowatt-hour - but only if Starship commercial launch costs reach roughly $500 per kilogram. Johnston expects that window to open in 2028 or 2029. The ultimate vision is an 88,000-satellite constellation.
The competitive field is intensifying. SpaceX acquired xAI in February and has applied to the FCC to build a million-satellite orbital compute network. Blue Origin has expressed similar ambitions. Google has Project Suncatcher. Aetherflux recently raised at a $2 billion valuation. But Johnston argues Starcloud holds a meaningful lead - it is the only company with actual telemetry data on how advanced chips perform in orbit, hard-won after one Nvidia A6000 failed during launch.
For business and infrastructure leaders tracking where AI compute capacity is heading, orbital data centers have moved from theoretical to operational. The timeline to cost competitiveness is still years away, but the technical proof of concept now exists. Starcloud put it there.
