Summary: Starcloud announced a $170 million Series A on March 30, 2026 at a $1.1 billion valuation to build orbital AI data centers. Its Starcloud-1 spacecraft flew an NVIDIA H100 in November 2025, while Starcloud-2 is planned for later in 2026 with AWS Outposts, an NVIDIA Blackwell chip, and bitcoin-mining ASICs. The 88,000-satellite network still awaits regulatory approval, and broader terrestrial workloads depend on Starship-class launch costs.
Starcloud, a Redmond, Washington company working to run compute in orbit, announced a $170 million Series A on March 30, 2026 at a $1.1 billion valuation — unicorn status for a business that got its start only about two years before. SpaceNews reports the company is still waiting on approval to deploy an 88,000-satellite orbital data center network; a Payload interview lays out the plan for the capital: a new manufacturing facility for spacecraft sized to fly on Starship.
The timing says a lot. One day earlier, OpenAI had closed its record $122 billion round (our coverage); that same week, Mistral lined up $830 million in debt financing for its Paris data center (our coverage). With power, land, and interconnection queues now acting as the hard ceiling on AI expansion, investors are starting to price “put the data center in low Earth orbit” seriously — and Starcloud is the first contender that shows up with flight heritage, a concrete schedule, and a unicorn valuation.
The Case for Orbiting Data Centers
The pitch comes down to three advantages: solar energy available nearly around the clock, heat rejected by radiators straight into vacuum with no cooling towers to build, and no grid interconnection or land permitting to clear. On that foundation Starcloud is underwriting three separate lines of business. One is near-real-time compute for data generated in space — imagery and telemetry from satellites face tight downlink bandwidth, so processing on orbit and returning results is frequently the cheaper move. Another hosts sensitive terrestrial workloads off Earth, where physical isolation buys security and sovereignty. The third takes on everyday inference, competing head-to-head with ground data centers.
CEO Philip Johnston did not sugarcoat it when speaking to Payload: the first business already makes sense at Falcon 9 launch prices, while the other two need Starship-class cheap mass to orbit. “The big constraint is getting them up cheaply.”
Who Is Behind the $170M Series A
The capital arrived in two stages: Benchmark led the initial close with EQT Ventures taking part, and the two firms then co-led the extension. Alongside funds administered by Macquarie Capital, the backer list runs through NFX, Nebular, Y Combinator, Adjacent, Seven Seven Six, Fuse Ventures, Manhattan West, and Monolith Power Systems. The angels lean toward aerospace and power backgrounds: Gen. Stephen Wilson, formerly vice chief of staff of the US Air Force; Dennis Muilenburg, who previously led Boeing as CEO; and Kevin Johnson, once CEO of Starbucks.
Johnston argued the round could have been much bigger — “at the current price, we could’ve easily taken three times the amount of capital” — and boiled the investment case down to one line: “you can’t back Google, you can’t back Amazon, and we’re like a mile ahead of any other startup.” Coverage has also billed Starcloud as one of the fastest Y Combinator companies ever to reach a unicorn valuation.
Three Spacecraft, Three Years: From a Single H100 to Starship
- Starcloud-1 (November 2025): launched on a Falcon 9 Bandwagon-4 rideshare from SpaceX carrying the first NVIDIA H100 GPU ever flown to space, showing that AI hardware can keep operating through radiation and thermal stress.
- Starcloud-2 (later in 2026): aims for 100 times the power of the craft before it and carries what is billed as the largest deployable radiator ever put in orbit. Its payload combines an NVIDIA Blackwell chip, AWS Outposts hardware, and bitcoin-mining ASICs, handling edge compute and cloud workloads for partners including Crusoe, AWS, Google Cloud, and NVIDIA.
- Starcloud-3 (Starship-class): the fresh capital funds a manufacturing facility for spacecraft built to the dimensions of commercial Starship flights, which Johnston expects to begin by the end of 2028. Headcount climbs from 13 to as many as 50 by year-end.
How the Business Works — and Where the Limits Sit
Bitcoin miners and AI inference riding on one satellite can look like a publicity stunt, yet both obey the same cost equation: what compute in orbit costs is dictated by the per-kilogram price of getting there, and whichever company first welds mass-produced spacecraft to deployable power and radiators is the one that turns workloads nobody else can serve into revenue. By flying AWS Outposts, Blackwell, and mining ASICs side by side, Starcloud-2 is in effect one platform testing whether three separate customer groups will pay.
The limits are equally plain. Until Starship brings cheap lift, orbital data centers can only take jobs that are flat-out impossible on the ground — not just cheaper. The 88,000-satellite network is still pending regulatory approval, and spectrum, orbital slots, and debris rules all remain undecided variables.
Takeaways for Developers and Infrastructure Teams
Three notes worth keeping. First, launch cadence is now a variable inside the compute supply chain: anyone planning inference capacity over multiple years should treat the orbital option as a supply line to track, not science fiction. Second, in-orbit processing of satellite data is the most credible near-term application — running inference and producing summaries of imagery and telemetry before downlink cuts bandwidth cost that is very real. Third, sovereignty and isolation form the second credible route: for financial and government data governed by strict compliance rules, “physically in orbit” works as a forceful isolation boundary. Until Starship commercializes around 2028, read Starcloud as a technology outpost to watch — ground-based, multi-cloud, multi-supplier setups remain the mainstream answer.