Cowboy Space Raises $275M Series B at $2B Valuation to Build Orbital AI Data Centers

Cowboy Space Corporation, the orbital infrastructure startup developing space-based AI compute systems, has raised a $275 million Series B round at a reported $2 billion valuation as the company advances plans to deploy large-scale AI infrastructure directly in orbit.

The financing round was led by Index Ventures, with participation from IVP, Blossom Capital, SAIC, and existing investors including Andreessen Horowitz and Breakthrough Energy Ventures. The company also formally unveiled its new branding, transitioning from its previous name, Aetherflux.

Building AI Data Centers in Space

Cowboy Space is developing what it describes as a vertically integrated orbital infrastructure stack for the AI era. Rather than treating rockets and satellites as separate systems, the company is engineering them together as a unified architecture optimized specifically for in-space compute workloads.

The company’s design centers around launch vehicles whose upper stages remain in orbit after deployment and transform into operational one-megawatt data centers. These orbital platforms are intended to run AI inference and high-performance computing workloads powered primarily by solar energy collected in Low Earth Orbit.

According to Cowboy Space, integrating the launch vehicle and compute payload into a single system reduces redundant mass and improves power efficiency compared to traditional satellite deployment models. The company argues that orbital compute infrastructure could eventually help address mounting pressure on terrestrial data centers as global AI demand accelerates.

NVIDIA Partnership Brings Advanced AI Hardware to Orbit

A major component of the company’s strategy involves collaboration with NVIDIA (NVDA ) to deploy the new NVIDIA Space-1 Vera Rubin Modules in orbit.

The recently announced hardware platform was specifically designed for space-based AI workloads operating in size-, weight-, and power-constrained environments. The technology is intended to support real-time onboard processing for applications such as geospatial intelligence, autonomous spacecraft operations, climate monitoring, and orbital analytics.

Instead of sending massive volumes of raw satellite data back to Earth for processing, future orbital systems could increasingly analyze data directly in space before transmitting only actionable insights to ground stations. Cowboy Space has also stated that its first satellite mission later this year will demonstrate space-to-Earth power beaming capabilities, another foundational component of its long-term infrastructure plans.

A Growing Race to Build Orbital Data Centers

Cowboy Space is entering an increasingly competitive field as aerospace and AI companies begin exploring whether portions of future computing infrastructure could eventually move into orbit. Companies including SpaceX (SPCX ), Starcloud, Axiom Space, and other emerging space-compute startups are now investigating orbital data center concepts, while NVIDIA has introduced dedicated hardware designed specifically for AI workloads in space.

The interest is being driven largely by the explosive growth in AI infrastructure demand. Modern AI systems require enormous amounts of electricity, cooling, and physical space, creating mounting pressure on terrestrial grids and data center construction. Orbital systems are being explored as a potential long-term alternative because satellites can access near-continuous solar energy while radiating heat directly into space rather than relying on traditional cooling infrastructure.

At the same time, orbital computing remains highly experimental. Launch costs, radiation exposure, hardware durability, communications bandwidth, and long-term maintenance all remain major unresolved challenges. Some researchers and industry analysts have questioned whether large-scale orbital AI infrastructure can become economically viable compared to rapidly improving Earth-based facilities.

Still, the broader concept could have important implications for the future of AI and satellite operations. Instead of transmitting massive volumes of raw satellite data back to Earth, future orbital systems may increasingly process information directly in space. That could reduce latency for applications such as climate monitoring, defense intelligence, disaster response, autonomous spacecraft operations, and global communications networks.

The sector is also beginning to attract serious infrastructure investment. NVIDIA’s rollout of hardware specifically designed for orbital AI compute, alongside multiple startup funding rounds across the space infrastructure sector, suggests that space-based computing is beginning to move from theoretical research toward early-stage deployment and testing.