Robinhood co-founder Bhatt raises $275M for Cowboy Space to build data centres in orbit

What's the deal? Aetherflux, a space startup founded in 2024 by Baiju BhattDealroom has a profile for this one. Try Dealroom → — co-founder of Robinhood — has raised $275M in a Series B round at a $2B valuation, rebranding as Cowboy Space Corporation.

The round was led by Index Ventures, with participation from Breakthrough Energy VenturesDealroom has a profile for this one. Try Dealroom →, Andreessen Horowitz, and others. Bhatt also invested personally.

Cowboy Space is building what it calls a power grid in outer space for AI — a constellation of satellites called Stampede that harness solar power to run GPU data centres in low Earth orbit, transmitting compute and optical data to Earth. Its architecture integrates the rocket's upper stage and the satellite into a single unified vehicle, eliminating redundant structure and maximising compute capacity per launch.

The company plans to launch its first satellite later in 2026 to demonstrate space-to-Earth power beaming, operate a cluster of GPUs in early 2027, and launch its own rocket by the end of 2028.

Why now? AI infrastructure is running into a fundamental constraint: Earth's power grid cannot keep pace with demand. In major US markets, average grid connection lead times for new data centres run five to seven years or more — a timeline incompatible with the pace of AI development.

Cowboy Space's thesis is that low Earth orbit offers abundant solar power, no permitting delays, and a physics-driven solution to the energy bottleneck that is increasingly limiting AI compute expansion on the ground.

The company originally focused on space-based solar power delivery to Earth. The pivot to orbital data centres reflects a sharper commercial thesis: rather than selling energy, sell compute — a product with proven demand and pricing that AI companies already understand.

What could go wrong? Cowboy Space is attempting to do two extraordinarily difficult things simultaneously: build its own rockets and build orbital data centres. Each of these individually has defeated well-funded, experienced teams.

The first rocket launch is not planned until 2028, and the competitive landscape for launch is dominated by SpaceX, which has more than a decade of operational experience and is preparing for an IPO at a reported $2 trillion valuation.

The orbital data centre concept, while compelling in theory, has never been demonstrated at commercial scale. Thermal management, radiation hardening, and the cost of launching and replacing GPU hardware in orbit are all unsolved engineering challenges at the scale Cowboy Space is envisioning.

The signal: Cowboy Space is the most ambitious expression yet of a thesis gaining traction in the AI infrastructure space: that the constraints on AI compute are ultimately physical — power, land, cooling — and that solving them requires looking beyond the planet's surface.

The involvement of Breakthrough Energy Ventures, Bill GatesDealroom has a profile for this one. Try Dealroom →'s climate-focused fund, alongside AI-native investors like a16z and NEA signals that the company is being taken seriously across multiple investment communities.

Whether Bhatt can build rockets, satellites, and orbital data centres from scratch in four years is a different question. But the fact that serious investors are funding the attempt at a $2B valuation tells something about how acute the AI infrastructure bottleneck has become.

Sources:
Business Wire
Bloomberg
TechCrunch
Payload Space
Cowboy Space, LinkedIn
Finsmes
Zamin

Image credits:
Cowboy Space Corporation

J.V.