Starcloud pozyskuje 250 mln USD w ramach rozszerzenia serii A przy wycenie 2,3 mld USD

Starcloud, startup z Redmond w stanie Waszyngton, budujący centra danych AI na niskiej orbicie okołoziemskiej, ogłosił 21 sierpnia 2026 r., że pozyskał $250 million Series A extension at a $2.3 billion post-money valuation. Runda ponad podwaja wycenę ustaloną w marcu 2026 r. i podnosi łączny kapitał pozyskany do $450 million od założenia w 2024 r.

Manhattan West poprowadził rozszerzenie, przy czym Ken Abdalla i Lauren Selig z tej firmy kierowali transakcją, a Selig dołączyła jako obserwator w radzie. W inwestycji uczestniczyli dotychczasowi inwestorzy Benchmark, EQT, Soma, NFX i 776, a także nowi inwestorzy NVIDIA, Cisco Investments, Cedar Capital, Goanna Capital i Standard Capital.

„Last November we put the first NVIDIA H100 in orbit,” Philip Johnston, Starcloud’s co‑founder and CEO, said in the announcement. „Today this fresh capital empowers us to build the infrastructure to launch many more of NVIDIA’s most advanced GPUs into space.”

The new capital is earmarked for three things, per the company: continued buildout of manufacturing capacity, engineering work in collaboration with NVIDIA, and procurement of future launch allocation. That last line is the one worth dwelling on. For an orbital data center company, launch is the single largest external dependency, and it sits almost entirely in SpaceX’s hands.

Dlaczego pojemność startowa jest ograniczającym czynnikiem

Starcloud’s compute roadmap is built around rockets it does not control. Its current‑generation Starcloud‑2 satellite is planned to launch later in 2026, per the March release.

That arithmetic explains why a $250 million round includes an explicit line item for buying launch allocation. Starcloud is not alone in feeling the squeeze: fellow orbital‑compute startup Cowboy Space raised $275 million in maj 2026 to pursue a vertically integrated design in which launch vehicle upper stages remain in orbit as one‑megawatt data centers. The orbital data center thesis depends on launch costs falling toward Starship’s promised economics, and the near‑term market is tightening before it loosens.

Współpraca Vera Rubin Space‑1

The NVIDIA investment formalizes a relationship that has been running on flight hardware for nine months. Starcloud’s first satellite, Starcloud‑1, launched in listopad 2025 carrying an NVIDIA H100 — the first data center‑grade GPU in orbit, roughly 100 times more powerful than any GPU compute previously flown. On that platform the company trained the first AI model in space, ran a version of Google’s Gemini in orbit, and demonstrated orbital fine‑tuning.

The two companies are now collaborating on what NVIDIA calls the Space‑1 Vera Rubin Module, a space‑rated variant of its next‑generation Vera Rubin platform, hardened for an environment where cooling happens only through radiators and every component must survive radiation exposure and launch loads. Starcloud’s satellites are expected to serve as an early flight platform for the module, feeding back data gathered since the H100 deployment.

Cisco’s entry adds a different kind of expertise. “Cisco has long been a leader in secure data center infrastructure and looks forward to bringing that expertise to the emerging world of orbital data centers,” Aleem Rizvon, vice president at Cisco Investments, said in the release.

Dotychczasowa historia

Starcloud emerged from Y Combinator in late 2024, founded by Johnston, CTO Ezra Feilden, and chief engineer Adi Oltean. On $3 million of pre‑seed capital it designed, built, and launched Starcloud‑1 in 21 months.

In marzec 2026 it raised a $170 million Series A at a $1.1 billion valuation co‑led by Benchmark and EQT, becoming the fastest company in Y Combinator history to reach unicorn status (17 months after demo day). In maj 2026 it signed a contract with SpaceX’s Starlink to integrate more than 50 Starlink Mini Laser terminals across at least 25 satellites, giving the constellation up to 25 Gbps of intersatellite optical links. The orbital buildout is one strand of a much broader race to find power and space for AI compute; Google’s Project Suncatcher is among the efforts exploring the same orbital territory, as Unite.AI covered in maj 2026.

W liczbach

• $250M rozszerzenie serii A, przy wycenie $2.3B po rundzie
• $450M łączny pozyskany kapitał od 2024 r.
• 88 000 satelitów w planowanej konstelacji, według zgłoszenia firmy do FCC
• 20 GW docelowej pojemności obliczeniowej w kosmosie
• 100 000 ft² zakładu produkcyjnego w Woodinville, Washington
• 2× wzrost wyceny w porównaniu do serii A z marca 2026 w ciągu pięciu miesięcy

Co będzie dalej

The near‑term hardware milestone is Starcloud‑2: launching later in 2026 with the largest commercial deployable radiator ever sent to space, serving early customer Crusoe alongside AWS, Google Cloud, and NVIDIA partnerships. The Woodinville facility, near where SpaceX and Amazon build their own satellites, is being fitted out for Starcloud‑3 production lines now.

Further out, the company’s stated trajectory — gigawatt‑scale orbital compute — requires Starship to reach routine, reusable flight and for launch prices to fall with it. The round gives Starcloud the capital to buy its way into that queue. Whether the rockets arrive on schedule is the variable no funding round can underwrite.