# Starcloud raises $250M as orbital AI runs short of rockets > Source: > Published: 2026-08-21 14:24:10+00:00 # Starcloud raises $250M as orbital AI runs short of rockets **Manhattan West led the extension at a $2.3B valuation, giving Philip Johnston capital for a factory, spacecraft and scarce launch slots.** By [RuntimeWire Staff](/author/runtimewire-staff) ยท Published Primary source: [TechCrunch](https://techcrunch.com/2026/08/21/starcloud-raises-200-million-for-orbital-data-centers-as-launch-options-dry-up/) ## Why it matters Starcloud is turning venture capital into launch reservations before orbital-compute developers crowd the same rockets. Its lead depends on manufacturing quickly while SpaceX remains supplier, technical dependency and potential competitor. Philip Johnston's [Starcloud](https://www.starcloud.com/?ref=runtimewire) has added a $250 million extension to its March Series A, [TechCrunch reported on August 21st](https://techcrunch.com/2026/08/21/starcloud-raises-200-million-for-orbital-data-centers-as-launch-options-dry-up/?ref=runtimewire), lifting the orbital-compute developer's valuation to $2.3 billion. Manhattan West Ventures led the extension, with Nvidia, Cisco, Benchmark, EQT, Soma, NFX, 776, Cedar Capital, Goanna Capital and Standard Capital participating. TechCrunch, citing a person familiar with the deal, reported that Nvidia invested $25 million. Starcloud said its [$170 million Series A in March](https://techcrunch.com/2026/03/30/starcloud-raises-170-million-series-ato-build-data-centers-in-space/?ref=runtimewire) brought its total capital raised to $200 million. Adding the $250 million extension reported by TechCrunch would imply $450 million in cumulative funding, though Starcloud has not reported that total. That is substantial financing for a 25-person hardware developer founded in 2024. Johnston is spending it against a specific deadline: Starcloud needs rockets before the orbital-compute sector consumes the available launch capacity. "We can see what's coming - we're going to need to book an enormous amount of launch," Johnston told [TechCrunch](https://techcrunch.com/2026/08/21/starcloud-raises-200-million-for-orbital-data-centers-as-launch-options-dry-up/?ref=runtimewire). ### A founder buying time on two clocks Johnston arrived at orbital computing through an unusual route. Starcloud describes him as a second-time founder who advised national space agencies on satellite projects at McKinsey. He also holds graduate degrees spanning physics, business and national-security technology. Before Starcloud, he co-founded the e-commerce aggregator Opontia. Johnston has said the idea took shape after a visit to SpaceX's Starbase operation in Texas. The production capacity being built around Starship prompted him and co-founder Ezra Feilden to examine businesses that could use dramatically cheaper access to orbit. Their calculations led them away from transmitting solar power to Earth and toward consuming that energy in space by running compute hardware. The other founders supply the engineering depth required to test that thesis. Feilden, Starcloud's CTO, previously designed deployable satellite structures at Airbus Defence and Space and Oxford Space Systems. Chief engineer Adi Oltean spent two decades working on large production GPU clusters at Microsoft before joining SpaceX, where Starcloud says he worked on Starlink tracking beams for spacecraft including Starship. The financing gives that group room to work on two schedules. Starcloud must manufacture increasingly powerful satellites while reserving launches years before its largest spacecraft is ready. The round will support production lines at a 100,000-square-foot facility in Woodinville, Washington, and development of Starcloud-3, a planned three-ton spacecraft with 200 kW of compute capacity. Starcloud-3 is designed around Starship's satellite deployment system. Johnston expects commercial access to Starship around 2028 or 2029, according to TechCrunch. Until then, Starcloud is considering a dedicated Falcon 9 mission and agreements with additional launch providers. Launch capacity has become a balance-sheet item. Starcloud is raising ahead of the contracts it expects to need, rather than waiting for Starship to become a routine commercial vehicle. ### One satellite before 88,000 Johnston has already put hardware in orbit. Starcloud-1 launched aboard a Falcon 9 in November 2025 with an Nvidia H100 GPU. Starcloud says the spacecraft subsequently trained a small language model, ran a version of Google's Gemini and processed radar data from Capella Space. That mission gave Starcloud telemetry on how a terrestrial data-center GPU handles launch, radiation, power and thermal constraints. It also exposed the cost of experimentation: an Nvidia A6000 carried on the mission failed during launch, Johnston told TechCrunch in March. The next step is Starcloud-2. Starcloud plans to send two 8 kW spacecraft on rideshare missions in 2027, with orbital inference work planned for customers including US government agencies. These satellites are closer to edge-computing nodes than hyperscale data centers. They can process information near Earth-observation spacecraft, reducing the volume of raw data that must travel through ground-station links. Johnston's larger pitch remains data-center-scale infrastructure powered by near-continuous sunlight and cooled through large radiators. In a [May interview with McKinsey](https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/the-case-for-data-centers-in-space?ref=runtimewire), he said Starcloud's long-term economics require launch prices to fall toward $500 per kilogram. Current orbital systems remain more expensive than terrestrial alternatives, he said. That dependency explains the size and timing of the extension. Starcloud can prove useful orbital inference with smaller spacecraft. Competing with terrestrial compute requires Starship-class mass economics, frequent launches and hardware capable of operating for years without physical repairs. ### SpaceX is both the ride and the rival SpaceX sits on both sides of Johnston's plan. Starcloud depends on SpaceX rockets, Starlink communications hardware and the expected economics of Starship. SpaceX has also asked the Federal Communications Commission for authority to operate [an orbital data-center system of up to one million satellites](https://docs.fcc.gov/public/attachments/DA-26-113A1.pdf?ref=runtimewire). Johnston argues that SpaceX will initially build for internal Grok and Tesla workloads, leaving room for Starcloud to supply third-party infrastructure. The distinction could narrow if SpaceX decides to sell orbital cloud capacity to outside customers. SpaceX would then control the rockets, communications network and competing compute fleet. Other entrants are also making large capital commitments before routine heavy-lift launch exists. [Google's Project Suncatcher](https://blog.google/innovation-and-ai/technology/research/google-project-suncatcher/?ref=runtimewire) is developing prototype AI satellites. Cowboy Space [raised $275 million in May](https://techcrunch.com/2026/05/11/there-arent-enough-rockets-for-space-data-centers-cowboy-space-raised-275-million-to-build-them/?ref=runtimewire) and plans to build a launch vehicle whose upper stage doubles as an orbital compute platform. Starcloud has taken the opposite route: buy launch capacity rather than develop a rocket. That keeps Johnston focused on spacecraft and compute, while leaving a core input under another operator's control. Starcloud has requested FCC authority for as many as 88,000 satellites. The figure sets the ceiling in a regulatory application; Starcloud currently operates one demonstration spacecraft. The American Astronomical Society has [raised concerns about the proposal](https://compasse.aas.org/aas-starcloud-data-centers-comment/?ref=runtimewire), including the effect that large numbers of continuously illuminated satellites and heat-dissipating radiators could have on optical and infrared astronomy. Johnston now has capital to move from a successful demonstration toward repeatable production. The harder milestone lies beyond fundraising: securing enough launches, at low enough prices, to make an orbital data center behave like infrastructure instead of a collection of expensive spacecraft.