The Orbital Data Center Race: Who Is Building Computing Infrastructure in Space

Two years ago, the idea of operating data centers in orbit existed primarily in academic papers and speculative conference presentations. Today, it is a funded industry with demonstrated technology, multiple competitors, billion-dollar valuations, and a market projection that would have seemed absurd until the AI energy crisis made it plausible. The in-orbit data center market is projected to reach $1.78 billion by 2029 and $39.1 billion by 2035, growing at a compound annual rate of 67.4% — one of the fastest-growing segments in the entire space economy.

The speed of this market's emergence is driven by a simple economic force: AI compute demand is growing faster than terrestrial power infrastructure can supply it. When utilities are pausing new data center connections and projecting multi-gigawatt grid shortfalls, the economics of putting computing hardware in orbit — where solar power is abundant, cooling is free, and land is unlimited — shift from theoretical to compelling.

Market Size and Growth Trajectory

The 67.4% CAGR makes orbital data centers one of the fastest-growing market segments in the space industry — faster than satellite communications, Earth observation, or space tourism. The growth is driven by the convergence of declining launch costs (SpaceX's Starship promises $200–500 per kilogram to LEO), increasing GPU performance per watt, and the intensifying terrestrial energy crisis that makes alternative power sources economically viable. The primary customer base initially comprises AI training companies with massive compute needs, defense and intelligence organizations requiring secure processing, and earth observation companies needing on-orbit image analysis.

The Competitive Landscape

Starcloud: The Funded Leader

Starcloud occupies the pole position with $200 million in funding, a $1.1 billion valuation, and the only demonstrated high-performance AI computing in orbit. The company's roadmap — Starcloud-2 with Blackwell B200 GPUs in October 2026, Starcloud-3 with 200-kilowatt power in 2027, and 40-megawatt commercial clusters in the early 2030s — represents the most concrete and well-funded development path in the sector. The company's partnership with AWS (evidenced by the AWS server blade on Starcloud-2) suggests a go-to-market strategy that integrates orbital compute into existing cloud infrastructure rather than requiring customers to build new workflows.

Starcloud's advantage is execution speed. The 21-month timeline from founding to orbital demonstration, achieved on $3 million, established a pace that competitors have not matched. The founding team's combination of SpaceX satellite manufacturing experience (Oltean), Airbus spacecraft engineering (Feilden), and strategic vision (Johnston) creates a talent base that spans both the aerospace and computing domains required to succeed.

Aethirflux: The NVIDIA-Backed Contender

Aethirflux (formerly Aethero) has taken a different path by partnering directly with NVIDIA to use the Space-1 Vera Rubin Module — NVIDIA's purpose-built platform for autonomous space operations. Founded by Baiju Bhatt (co-founder of Robinhood), Aethirflux brings consumer technology credibility and fundraising capability to the orbital computing space. The company's vision extends beyond data centers to include space-based energy infrastructure — generating solar power in orbit and potentially beaming it to ground stations.

The NVIDIA partnership gives Aethirflux access to purpose-built space computing hardware that Starcloud lacks — the Vera Rubin Module is designed from the ground up for the space environment, potentially offering better radiation tolerance and power efficiency than commercial GPUs adapted for orbit. However, Aethirflux has not yet launched orbital hardware, putting it behind Starcloud in the demonstration timeline that investors and customers use to assess credibility.

The Hyperscaler Question

The most significant competitive dynamic may come not from other startups but from the cloud hyperscalers themselves. Microsoft's Azure Orbital program already provides ground station services that connect satellites to the Azure cloud. Amazon's Project Kuiper gives AWS a satellite manufacturing and operations capability. Google has invested in satellite imagery and space-based connectivity. If orbital computing proves economically viable, any of these companies could build or acquire their way into the market with resources that dwarf any startup.

The startup opportunity exists because hyperscalers move slowly on hardware infrastructure. Microsoft, Amazon, and Google are focused on building terrestrial data centers as fast as possible — they have neither the organizational bandwidth nor the risk appetite to simultaneously pursue orbital alternatives. This creates a window for companies like Starcloud to establish technology leadership, build customer relationships, and reach a scale where acquisition becomes more attractive than competition for the hyperscalers.

Challenges and Open Questions

Despite the momentum, orbital data centers face significant unresolved challenges. Bandwidth is the most immediate: getting data to and from orbit requires high-throughput ground station networks or optical inter-satellite links, and current satellite communication bandwidths are orders of magnitude below what terrestrial data centers achieve via fiber optics. Latency adds another constraint — LEO satellites at 500 km altitude introduce roughly 3–5 milliseconds of round-trip latency, acceptable for batch AI training but potentially problematic for real-time inference. Satellite lifespan, orbital debris risk, regulatory frameworks for orbital computing, and the logistics of launching and maintaining large constellations all represent challenges that the industry must solve at scale.

The market's 67.4% projected CAGR reflects both the genuine opportunity and the speculative nature of early-stage projections. If terrestrial power constraints intensify and launch costs continue declining, orbital data centers could become a mainstream computing infrastructure. If fusion power, advanced nuclear, or other terrestrial energy solutions scale faster than expected, the economic case for orbital computing weakens. The next three to five years — as Starcloud and its competitors move from demonstrations to commercial operations — will determine whether this market reaches its projected scale.

Frequently Asked Questions