Analysis
Starcloud, a startup building AI data centers designed to operate in orbit, raised a Series A extension led by Manhattan West, TechCrunch reported Friday, with existing backers Benchmark, EQT, Soma, NFX and 776 participating alongside new investors including Nvidia, Cisco Investments, Cedar Capital, Goanna Capital and Standard Capital.
- New raise -- $250M Series A extension
- New valuation -- $2.3B post-money
- Total raised since 2024 founding -- $450M
What Starcloud is actually building
Starcloud's pitch is that data centers in orbit sidestep two of the hardest constraints facing terrestrial AI infrastructure: power and land. A satellite in the right orbit gets roughly continuous solar exposure without competing for grid capacity or real estate the way a ground-based hyperscale campus does, and cooling in vacuum -- while a genuinely different engineering problem, not simply an easier one -- doesn't require the water-intensive systems land-based data centers use. The company flew the first Nvidia H100 GPU to orbit aboard its Starcloud-1 satellite in November 2025, establishing proof that commodity AI accelerators can survive and operate in space before asking investors to fund a much larger buildout.
The new capital funds a 100,000-square-foot manufacturing facility in Woodinville, Washington and development of Starcloud-3, the company's flagship next-generation spacecraft, designed to launch aboard SpaceX's Starship rocket. That dependency is explicit in how TechCrunch framed the raise: launch options are "drying up" as demand for heavy-lift capacity outstrips supply, and Starship remains an unproven vehicle for routine commercial payloads at the cadence Starcloud's roadmap requires.
The scale of the actual vision
Starcloud's stated long-term goal is a constellation of 88,000 satellites delivering 20 gigawatts of orbital compute capacity -- for comparison, 20 gigawatts is roughly the combined generating capacity several large US metro areas draw on, if realized as ground-based compute instead. That's an enormous scale-up from a company that has flown a single test satellite, and it puts Starcloud in a category that barely existed eighteen months ago: Muon Space, which also raised a fresh $250 million round this week at a $1.5 billion valuation, is separately building toward its own orbital AI-compute capability alongside its core Earth-observation business, suggesting multiple well-funded players now see orbital compute as a real adjacent category to satellite manufacturing rather than a research curiosity.
What to watch
Nvidia's direct investment is the clearest signal that a major compute supplier sees strategic value in orbital deployment beyond a single GPU test flight, but the entire thesis remains gated by launch cadence and cost -- Starcloud's own roadmap depends on Starship reaching a reliability and launch frequency it has not yet demonstrated commercially. A $2.3 billion valuation on a company with one flown satellite and no disclosed revenue prices in a large amount of execution risk across manufacturing, launch access and in-orbit reliability all happening roughly on schedule, in a sector where each of those three things has historically slipped independently.
