Google launched four Tensor processing units into orbit to test the in-space compute concept, initiating a push by tech companies to explore orbital computing. Driven by terrestrial power grid constraints and massive energy demands, startups and industry giants are betting that space-based solar power and vacuum cooling can solve the looming compute bottleneck.
Why Are Tech Companies Racing to Put AI Data Centers in Space?
Terrestrial electrical grids are facing strain. Data centers will account for almost half of U.S. electricity demand growth between now and 2030, forcing companies to look skyward for power.
In space, solar panels collect about eight times as much energy as they can on Earth and operate with effectively continuous output in the right orbit, eliminating the need for battery storage or grid tie-ins. The orbital compute movement gained momentum when Jeff Bezos stated during an Italian fireside chat that data centers “will be better built in space, because we have solar power there, 24/7.” Elon Musk subsequently named space computing a leading ambition of SpaceX, declaring at Davos earlier this year that space will host the lowest-cost AI infrastructure within two years, three at the latest.
How Do Projects Like Google Suncatcher and Starcloud Test This Technology?
Testing is already underway in low Earth orbit. Google launched four of its homegrown Tensor processing units into space aboard a solar-powered Planet Labs satellite via an uncrewed SpaceX Falcon 9 rocket on October 1. Alphabet holds an $82 billion stake in SpaceX.
https://x.com/DowdEdward/status/1998516087030428075
The launch is part of Project Suncatcher, an initiative announced in November 2025 where Google aims to run AI workloads on clusters of dozens of satellites linked by free-space lasers. Meanwhile, Starcloud has already flown an Nvidia H100 into space. Philip Johnston, the CEO of Starcloud, argues that hosting servers in orbit bypasses power grid queues, eliminates water cooling debates, and avoids terrestrial land rights disputes. Gwynne Shotwell, SpaceX president and COO, announced in September that the company aims to launch the first purpose-built orbital data center satellites in late 2027, and SpaceX has filed with the Federal Communications Commission to launch up to a million satellites. Nvidia has also announced a space-grade Vera Rubin module, and Blue Origin has filed for a 51,600-satellite network.
What Are the Major Technical and Economic Hurdles Facing Orbital Compute?
Industry skeptics and engineers point out that space presents thermal management and economic roadblocks. Rocket payloads remain constrained and expensive.
While space is cold, a vacuum prevents air and water from flowing past chips to carry away heat. Instead, satellites must rely entirely on radiation and large surface radiators to shed heat slowly as infrared light. During Google’s “Suncatcher” tests, the Trillium TPU chips run only short Gemini queries for limited stretches before shutting down to cool off. A Saarland University paper titled “Dirty Bits in Low-Earth Orbit” argues rocket launch and reentry emissions alone would cancel out any gains from getting data centers off Earth.
Does Orbital Debris Threaten the Viability of Space-Based Data Centers?
Orbital clutter poses a collision risk. Adding tens or hundreds of thousands of new satellites threatens to trigger Kessler Syndrome, a chain reaction where a single collision creates thousands of fragments that hit other objects and create more fragments. European Space Agency debris reports already show crowded low Earth orbits, with new low-orbit communication satellites mostly to blame.
Worth a look
- FC Köln Loses Strikers Dallinga and Ache to Injury in Test Match
- Murphy vs. McDonald Rematch: How to Watch the Fight on October 3
- First WA Student Alaxandrea O’Brien Joins NASA Through National Indigenous Space Academy (world-today-news.com)
- The Space Station Is Always Falling. Here's What Keeps It Up. (daybreakwire.com)
