Starcloud Has $250 Million for Orbital AI. It Still Needs a Ride.

Starcloud’s pitch is an orbital layer for AI inference, but its $250 million funding extension cannot solve the constraint at the center of that plan: getting enough spacecraft to orbit. The company now has capital to build, test and reserve launches, while its largest proposed system remains tied to Starship, a rocket that has yet to become a routine commercial ride.

The extension, added to Starcloud’s March Series A, values the company at $2.3 billion. Manhattan West Ventures led the financing, with Nvidia, Cisco, Benchmark, EQT, Soma, NFX, 776, Cedar Capital, Goanna Capital and Standard Capital participating. A person familiar with the deal said Nvidia put in $25 million.

Starcloud says the money will support a larger manufacturing facility and its Starcloud-3 orbital data-center spacecraft. The 25-person company is developing production lines in a 100,000-square-foot facility in Woodinville, Washington. Starcloud-3 is intended to fly on SpaceX’s forthcoming Starship, making its long-range deployment schedule dependent on a vehicle still being brought online.

Starship’s progress is itself unsettled. Elon Musk said SpaceX would postpone an attempt to catch a returning Starship by a few months and would attempt the vehicle’s first re-flight at the end of the year or early 2027. Neither step amounts to routine commercial service, while Starcloud says it wants to contract for Starship capacity as soon as it can.

That dependency is not incidental. CEO Philip Johnston said launch capacity is now among the company’s biggest costs, and that Starcloud expects to need an enormous amount of it. He said Falcon 9 is scheduled to end in 2028. SpaceX is shifting toward the larger Starship, while Blue Origin’s New Glenn and ULA’s Vulcan are not flying regularly and Rocket Lab’s Neutron has not yet reached the pad.

$250 millionNew extension

Starcloud added a $250 million extension to its March Series A.

$2.3 billionCompany valuation

The financing valued Starcloud at $2.3 billion.

88,000 spacecraftFCC request

Starcloud has requested FCC permission to operate 88,000 spacecraft.

The nearer plan is more modest. Starcloud aims to launch two 8-kilowatt Starcloud-2 compute satellites on rideshare missions in 2027, performing inference work for customers that include U.S. government agencies.

Those smaller missions offer a nearer operational milestone, but do not remove Starcloud’s reliance on future Starship capacity. The company is considering a dedicated Falcon 9 launch for more spacecraft and contracts with other providers, rather than announcing either as secured capacity. Johnston said that being unable to book SpaceX capacity in 2029 would be challenging.

Three separate tests now sit between the round and the larger network

• Manufacturing: Starcloud must turn its Woodinville production lines into repeatable spacecraft output.
• Launch: the company must secure rideshare, dedicated or future Starship capacity before an increasingly constrained market closes off its options.
• Hardware: Starcloud is tracking chip temperature, radiator size, radiation shielding and launch ruggedization for space-based compute.

Starcloud has already put an Nvidia H100, a terrestrial data-center GPU, into orbit and says it used the hardware to train a model. Most other space GPUs are designed for edge processing, rather than data-center use. The company is sharing its findings with Nvidia as the chipmaker develops the Vera Rubin Space-1, a purpose-built GPU for space that has not yet been built. Starcloud hopes to fly that chip in late 2028.

The FCC request shows how far the company’s ambition extends beyond its initial satellites. But a regulatory ceiling is not a deployed fleet, and the commercial case still turns on launch frequency and cost. The new round buys Starcloud time to prepare; it does not establish whether Starship capacity will arrive on the schedule its orbital data-center model requires.