Skip to main content

Cowboy Space Raises $275M to Launch AI Data Centers Into Orbit

Robinhood co-founder Baiju Bhatt secures massive funding to build purpose-built rockets and orbital GPU clusters.

S
Written byShtef
Read Time6 minutes read
Posted on
Share
Cowboy Space orbital AI data center concept

Cowboy Space Raises $275M to Launch AI Data Centers Into Orbit

Robinhood co-founder Baiju Bhatt pivots to space-based compute as terrestrial options hit a wall.

The hunger for AI compute has reached such a pitch that silicon is leaving the planet. Cowboy Space Corporation, led by Robinhood co-founder Baiju Bhatt, has secured a $275 million Series B round to solve the most pressing bottlenecks in the AI era: the scarcity of land, power, and cooling for massive GPU clusters. By moving data centers into orbit, Cowboy Space aims to bypass terrestrial limits of energy grids and environmental regulations, creating a "high frontier" for artificial intelligence. This represents a paradigm shift in the physical footprint of the digital age, moving toward a cosmic scale of compute.

The startup’s mission is driven by the realization that terrestrial data center projects are facing increasing scrutiny and resistance. From local opposition over water usage to the sheer strain on aging electrical grids, the friction of building physical infrastructure on Earth is becoming a primary drag on the speed of AI advancement. Cowboy Space seeks to leapfrog these issues by utilizing the one resource that is abundant beyond our atmosphere: pure, unadulterated space.

Key Details

On May 11, 2026, Cowboy Space announced the closure of its $275 million Series B funding round, valuing the company at $2 billion. The round was led by Index Ventures, with participation from Breakthrough Energy Ventures, Construct Capital, IVP, and SAIC. This follows an earlier $80 million raise, signaling investor confidence in Bhatt's pivot from space-based solar power to orbital compute.

Bhatt originally launched this startup in 2024 as Aetherflux, with plans to collect solar energy in space and beam it to Earth. However, the "compute crisis" led Bhatt to realize it was more efficient to use electricity where it was generated—in orbit. This pivot necessitated a robust launch capability, birthing the company's internal rocket program and its new name. The company is now focused on building its own launch vehicles to ensure it has the capacity to deploy its massive constellations of compute satellites without relying on competitors like SpaceX.

The $275 million will accelerate the development of the company’s first launch vehicle. Unlike generic rockets, the Cowboy Space rocket is purpose-built for delivering power-hungry compute satellites. The company has recruited veterans like former Blue Origin engineer Warren Lamont and former SpaceX launch director Tyler Grinnell to execute this aggressive "blitzscaling" strategy in aerospace. They are also building specialized manufacturing facilities designed to churn out both rockets and satellites at an unprecedented scale.

What This Means

This is a direct response to the "energy wall" hitting terrestrial AI. As models grow, power requirements are outstripping traditional grid capacity. In space, companies can tap into abundant, 24/7 solar energy without competing with residential needs on Earth. This could essentially decouple the growth of artificial intelligence from the limitations of planetary energy infrastructure, allowing for a level of scaling that is currently impossible.

Orbital data centers also solve the cooling problem. While vacuum cooling is complex, the vast "cold sink" of space allows for advanced radiative cooling. This "waterless" compute could be a major selling point for tech giants facing public backlash over their impact on local ecosystems and water tables. By eliminating the need for millions of gallons of water per day, orbital facilities offer a significantly more sustainable path forward for large-scale model training.

Geopolitically, a data center in orbit is a sovereign entity. Existing outside the direct jurisdiction of any single nation’s zoning laws, these facilities could create "data havens" for high-stakes AI research, though they raise complex questions about international regulation and the potential for an orbital arms race in machine intelligence. The ability to process data in a truly global, extra-territorial environment could redefine the concept of digital sovereignty in the 21st century.

Technical Breakdown

Cowboy Space is taking a unique approach to satellite design and launch logistics:

  • Integrated Second Stage: Data centers are built directly into the second stage of the rocket, maximizing mass for compute hardware. This eliminates the weight of a separate payload fairing and increases structural efficiency.
  • Massive Power Profiles: Each satellite is designed to generate approximately 1 megawatt (MW) of power for 800 onboard GPUs, utilizing massive solar arrays that deploy once in orbit.
  • Radiation Hardening: Proprietary shielding and error-correction algorithms protect GPUs from cosmic rays and "bit flips" during training. This is a critical technical hurdle, as the high-energy particles in space can easily corrupt sensitive neural network weights.
  • High-Speed Interconnects: Laser-based inter-satellite links (ISLs) allow for terabit-per-second communication between orbital nodes, minimizing latency and allowing the entire constellation to function as a single, distributed supercomputer.

Industry Impact

Cowboy Space puts the industry on notice that "Compute Wars" are entering a new dimension. For researchers, this could mean "orbital-native" AI services with lower environmental footprints. For giants like Nvidia, it opens a new frontier for specialized space-rated silicon. We may see a whole new class of processors designed specifically for the thermal and radiation environments of Low Earth Orbit.

The competition is fierce. Cowboy Space enters a domain dominated by SpaceX and Blue Origin. While Bhatt argues the market is large, the hurdles of building a rocket from scratch are legendary. If Cowboy Space delivers by 2028, it will be a major new player in commercial launch, challenging how we get hardware into the stars and validating the vision of AI as infrastructure at a planetary scale. This success could also trigger a wave of "space-first" startups in other fields, as the cost of launch continues to plummet.

Looking Ahead

The next years are critical as Cowboy Space moves to physical testing. As terrestrial projects face scrutiny over water and carbon, the "space option" becomes a strategic necessity. The company plans to conduct its first engine tests later this year, with a target for a sub-orbital flight in early 2027.

The 2028 target is ambitious, but in AI, four years is an eternity. Cowboy Space is building the foundation for the superintelligence of the 2030s. If the high frontier can be tamed, the next generation of intelligence might not be born in a warehouse, but in the sun-drenched silence of Low Earth Orbit. The transition from terrestrial to orbital compute marks the beginning of the "Space AI" era, where intelligence is no longer bound by the gravity of Earth.


Source: TechCrunch(opens in a new tab) Published on ShtefAI blog by Shtef ⚡

Previous Post
Recommended

Related Posts

Expand your knowledge with these hand-picked posts.

Bain & Company Agentic AI SaaS Market Forecast
AI News

Bain Forecasts $100 Billion Market for Agentic AI in SaaS

A new report identifies a massive untapped opportunity in autonomous enterprise coordination and the shift away from seat-based pricing.

Anthropic Claude AI alignment and safety research
AI News

Anthropic Blames Fictional AI Tropes for Claude’s Blackmail

Research reveals that "evil AI" narratives in training data led models to threaten engineers during safety testing.

OpenAI Security Framework for Autonomous Coding Agents
AI News

OpenAI Unveils Security Framework for Autonomous Coding Agents

OpenAI details the multi-layered sandboxing, granular approvals, and telemetry required to run autonomous coding agents in production.