
Aetherflux, a California-based space energy startup, says it will bring its first orbital data center into commercial service by the first quarter of 2027, targeting space-based infrastructure to bypass the growing bottlenecks limiting AI development on Earth.
The project, branded Galactic Brain, plans on using solar-powered computing nodes operating in low Earth orbit and connected through optical links. The goal is to colocate compute hardware directly with abundant solar energy, which avoids the land acquisition, permitting, grid interconnection, and cooling challenges that increasingly slow Earth-based data center construction.
Aetherflux founder and chief executive Baiju Bhatt, who previously co-founded Robinhood, has framed the strategy as a move toward more capable AI systems with plentiful compute and energy. From the company’s perspective, space offers a way to sidestep Earth’s limits rather than compete within them.
Planning on 2026 Demonstration
The company’s first orbital node is expected to deliver multi-gigabit bandwidth and availability comparable to ground-based servers, using optical inter-satellite links and emerging relay networks. Aetherflux says initial deployments will rely on teraflop-class systems, with a roadmap to scale toward petaflop-level capacity as more satellites are added to the constellation. Enterprises would interact with orbital workloads much as they do with cloud infrastructure today. The company has not disclosed pricing.
Before launching its data center hardware, Aetherflux plans a 2026 demonstration mission to test wireless power transmission from orbit to Earth. The satellite will attempt to beam roughly one kilowatt of energy to ground stations using infrared lasers, a key step in validating the company’s broader space-based ambitions.
Earlier this year, Aetherflux raised $50 million in a Series A round led by firms including Index, Andreessen Horowitz, NEA, and Breakthrough Energy Ventures, bringing total funding to $60 million. The company has also received U.S. Department of Defense support for proof-of-concept work on power transmission, reflecting potential military interest in supplying energy to remote locations.
Aetherflux has plenty of competitors in the data center in space race. Industry figures like Elon Musk and Jeff Bezos have strategized about gigawatt-scale space-based data centers, and startups are also exploring orbital processing as AI workloads strain existing infrastructure.
Big Challenges Remain
Analysts note that while satellites can support edge computing and specialized workloads, the economics of moving large-scale data center operations into orbit are far from proven. Launch costs, hardware replacement cycles, and latency considerations all pose challenges that operators on Earth do not face.
There is also debate over terminology. Some industry observers argue that labeling processor-equipped satellites as “orbital data centers” risks overstating current capabilities, particularly when compared with the vast scale of modern hyperscale facilities on Earth.
Aetherflux counters that its approach is evolutionary rather than monolithic. The company plans to continuously launch updated hardware, allowing newer systems to handle priority workloads while older nodes are repurposed for less demanding tasks, extending their useful life.
Whether orbital computing becomes a core pillar of AI infrastructure or remains a niche solution will depend on execution as much as vision. For now, Aetherflux’s 2027 target places it squarely in the early wave of companies testing whether space can offer a practical answer to AI’s escalating appetite for power and compute.

