Starcloud secures $250M for orbital AI data centers

An orbital data center startup has secured fresh capital to advance its plan of deploying high-performance computing infrastructure in space. The funding reflects growing investor interest in alternative AI infrastructure, even as technical and economic hurdles persist for off-planet operations.

Funding and strategic backing

Starcloud has raised $250 million in a Series A extension round, bringing its total funding to $450 million. The investment values the company at $2.3 billion post-money and was led by Manhattan West, with participation from existing backers such as Benchmark, EQT, and Soma. New investors include NVIDIA, Cisco Investments, and Cedar Capital. The capital will be used to expand spacecraft manufacturing, secure launch capacity, and deepen engineering collaboration with NVIDIA.

The involvement of NVIDIA and Cisco signals strategic interest in adapting computing and networking hardware for space environments. NVIDIA’s role focuses on developing radiation-tolerant accelerators, while Cisco’s expertise in networking could address data transfer challenges between orbital and terrestrial infrastructure. However, the funding does not resolve fundamental questions about the viability of space-based compute at scale.

Technical milestones and challenges

Starcloud has demonstrated AI workloads in orbit, including training models and running inference on an NVIDIA H100 GPU aboard its Starcloud-1 satellite, launched in late 2025. The company has also tested a version of Google’s Gemini AI in space. These achievements confirm that AI hardware can function in orbit, but they do not address whether such infrastructure can compete with terrestrial data centers on cost, reliability, or serviceability.

Operating in space introduces unique constraints. Hardware failures cannot be repaired on-site, radiation exposure requires specialized designs, and heat dissipation relies on radiators rather than traditional cooling systems. Launch availability and replacement cycles add further complexity to capacity planning. Starcloud’s long-term vision includes deploying 88,000 satellites and 20 gigawatts of orbital compute capacity, a scale that would rival major terrestrial data center buildouts but requires industrialized manufacturing and launch cadence.

Economic and regulatory hurdles

The economics of orbital compute remain unproven. While falling launch costs could improve feasibility, Starcloud has not disclosed details on when space-based AI workloads might achieve cost parity with terrestrial alternatives. The company’s manufacturing expansion—including a 100,000-square-foot facility in Washington—highlights the shift from demonstration hardware to industrial-scale production. However, producing thousands of satellites requires supply chain assurance and launch frequency that few computing companies have managed.

Regulatory and legal questions also loom. Spectrum rights, launch licensing, debris mitigation, and data sovereignty could complicate deployment. Workloads in orbit remain subject to the legal frameworks of the countries they serve, despite their physical location above national borders. Enterprises may hesitate to adopt orbital infrastructure without clear service guarantees, security controls, and dependable connectivity.

What to watch

Investors and industry observers will focus on several key indicators of progress. Manufacturing throughput, launch commitments, and spacecraft reliability will reveal whether Starcloud can transition from technical demonstrations to repeatable infrastructure. Early customer workloads and partnerships will signal market demand, while advancements in radiation-tolerant hardware and networking could address some of the current limitations. For now, orbital AI data centers remain an ambitious experiment rather than a proven alternative to terrestrial cloud regions.