Starcloud Raises $250 Million at $2.3 Billion Valuation to Build AI Data Centers in Orbit

Starcloud, a startup building data centers in orbit, announced on August 21 that it has raised $250 million in a Series A extension led by investment firm Manhattan West, with Nvidia among the investors backing the round.

The funding doubles the Redmond, Washington-based company's valuation to $2.3 billion and brings its total capital raised since its founding in 2024 to $450 million, according to SpaceNews. The raise is one of the largest yet for the emerging orbital computing sector, and it arrives as the industry's push to move AI workloads into space shifts from concept to hardware. For the latest on the infrastructure race powering AI, bookmark AI Buzz Wire.

From an H100 in Orbit to a 200-Kilowatt Spacecraft

Starcloud's roadmap is built on progressively larger satellites designed to run cloud computing workloads in low Earth orbit.

Its first satellite, the 60-kilogram Starcloud-1, was launched by SpaceX in November and became the first satellite to run Nvidia's H100 graphics processing unit in orbit. The company said it used the satellite to train an AI model and to run a version of Google's Gemini in space.

Next up is Starcloud-2, a 450-kilogram satellite slated to launch in January on a SpaceX Falcon 9 rideshare mission. The spacecraft is designed to generate about eight kilowatts of power — roughly one hundred times more than its predecessor.

The long-term centerpiece is Starcloud-3, a three-ton, 200-kilowatt-class spacecraft intended to support AI inference and training workloads uplinked from Earth. Starcloud has filed plans with U.S. regulators for a constellation of as many as 88,000 satellites, ambitions that rely heavily on SpaceX's much larger reusable Starship rocket, which is being developed to eventually replace Falcon 9.

The Nvidia Connection

Nvidia's participation in the round extends a partnership that has already produced space-rated computing hardware. In March, Nvidia unveiled the Space-1 Vera Rubin Module, a computing system designed to bring more powerful AI processing to satellites that the chipmaker says delivers up to 25 times more AI compute than the H100.

Starcloud plans to use the more powerful module on future spacecraft to run cloud workloads in space for customers including Crusoe, an AI infrastructure provider. The new capital is intended to build out manufacturing capacity, procure rocket launches, and support engineering work with Nvidia, the company said.

Extending a March Milestone

The new capital builds on a March Series A that raised $170 million, according to TechCrunch, and the speed of the follow-on reflects how quickly sentiment around orbital computing has shifted. A sector that was widely dismissed as science fiction two years ago now counts some of the most consequential names in AI among its backers — and its engineering partners.

Nvidia's infrastructure bets now span both ends of the market. On the same day the Starcloud round was announced, Reuters reported that the chipmaker had also made a minority investment in Cloverleaf, a company that works with utilities and energy providers to secure terrestrial infrastructure for data centers. The parallel moves underscore the same underlying calculation: compute demand is growing faster than conventional infrastructure can accommodate it.

A Race With SpaceX

Starcloud is not alone in betting that AI computing will eventually migrate off the planet. SpaceX, which acquired AI company xAI in February, has outlined plans for up to one million orbital data center satellites of its own — a vision that also depends on Starship's capacity and cost profile.

SpaceX is partnering with Nvidia to design the Starmind AI1 satellite compute payload, with each satellite slated to carry Nvidia Rubin GPUs and Vera CPUs for data-center-class computing in space, according to details shared by the venture and highlighted by SpaceNews.

TechCrunch, which first reported the funding, noted that the raise comes as launch options dry up — rideshare capacity on trusted rockets has become a gating factor for startups whose business plans require mass in orbit. Securing capital now, ahead of Starcloud-2's January launch, positions the company to buy its way to the front of the queue.

The broader appeal of orbital data centers is straightforward: AI training and inference consume enormous amounts of energy, and advocates of space-based computing argue that orbit offers resources that are increasingly constrained on the ground. Whether the economics — launch costs, radiation hardening, thermal management, and laser links back to Earth — can work at scale remains the industry's central open question.

Who Is Behind Starcloud

The company was founded by Philip Johnston (CEO), Ezra Feilden (CTO), and Adi Oltean (Chief Engineer). Under their plan, each successive satellite generation scales power and compute by orders of magnitude rather than increments — a bet that the trajectory of AI demand will outpace what terrestrial grids and data center construction can absorb.

With $450 million raised, Nvidia's engineering partnership behind it, and a launch slot on the calendar for January, Starcloud has moved from the fringes of the AI infrastructure debate toward its center. The next proof point arrives with Starcloud-2 — and with it, the first real test of whether purpose-built orbital data centers can deliver on one of the industry's most audacious promises.

Explore the AI Infrastructure Boom Read more AI news →