Starcloud has added a $250 million extension to the $170 million Series A round it announced in March, giving the orbital-computing startup a reported valuation of $2.3 billion. The company plans to use the capital to expand manufacturing and develop satellites intended to run AI workloads in space.
The financing arrives as launch availability becomes a strategic constraint. Starcloud’s business ultimately depends on sending large amounts of computing hardware into orbit at prices low enough to compete with terrestrial data centres. Its preferred long-term vehicle, SpaceX’s Starship, remains unproven as a rapidly reusable transport system.
Starcloud funds a much larger orbital plan
Starcloud currently has 25 employees and is establishing production lines in a 100,000-square-foot facility in Woodinville, Washington. The new funding is expected to support that expansion and advance Starcloud-3, its largest planned orbital data-centre spacecraft. The company intends that vehicle to fly on Starship.
Its ambitions extend far beyond a single satellite. Starcloud has asked the US Federal Communications Commission for permission to operate 88,000 spacecraft. An application of that scale describes the company’s desired network, not an approved or financed deployment schedule, and TechCrunch does not say when such a constellation could be completed.
Before Starcloud-3, the company plans to launch two 8-kilowatt Starcloud-2 compute satellites on rideshare missions in 2027. They are intended to perform AI inference for customers that include US government agencies. Starcloud is also considering a dedicated Falcon 9 flight for additional spacecraft and contracts with other launch providers.
The funding extension was led by Manhattan West Ventures, with participation from Nvidia, Cisco and several venture firms. A person familiar with the transaction told TechCrunch that Nvidia invested $25 million. That amount was not independently confirmed in the article.
Rocket capacity could determine the economics
CEO Philip Johnston says Starcloud is raising and reserving capital partly to secure launch contracts before capacity tightens. SpaceX plans to phase out Falcon 9 in 2028 and replace it with the far larger Starship. Competing vehicles such as Blue Origin’s New Glenn and United Launch Alliance’s Vulcan are not yet flying regularly, while Rocket Lab’s Neutron has not reached the launchpad.
This transition leaves satellite operators planning around vehicles whose future cadence and pricing are uncertain. Starcloud’s economic model assumes that Starship can reduce the cost of carrying heavy computing systems to orbit through frequent reuse. If suitable SpaceX capacity is unavailable in 2029, Johnston acknowledged that the company would face a serious challenge.
SpaceX’s own schedule remains fluid. TechCrunch reports that Elon Musk said the company would delay an attempt to catch a returning Starship by several months and target the vehicle’s first reflight for late 2026 or early 2027. Those plans are projections from SpaceX, not completed technical milestones.
Launch is only one part of the cost equation. Orbital computers must survive rocket vibration and radiation, shed heat through radiators and operate without conventional data-centre maintenance. These requirements could offset some theoretical savings from abundant solar power or reduced terrestrial infrastructure, which the source does not quantify.
In-orbit experiments precede commercial scale
Starcloud says it already operates an Nvidia H100 data-centre GPU in orbit and has trained a model with it. Johnston describes the company as the only known operator to have achieved those steps with that class of hardware. TechCrunch presents the claim with that qualification rather than as a complete survey of every space-computing project.
The startup is sharing lessons from its Starcloud One mission with Nvidia as the chipmaker develops Vera Rubin Space-1, a GPU intended specifically for orbital use. That processor has not yet been built, although Starcloud hopes to fly it in late 2028.
Engineers are still evaluating core design trade-offs, including how operating temperature affects radiator size, where radiation shielding should sit and how much ruggedisation the chips need. These are practical constraints that must be resolved before orbital AI can progress from demonstrations to dependable infrastructure.
The new capital gives Starcloud resources to manufacture larger systems and reserve scarce launches. It does not remove the programme’s dependence on rocket schedules, unbuilt chips or unproven economics. The next two satellites should provide more concrete evidence about whether useful AI inference in orbit can become a repeatable service rather than an expensive technical showcase.
