WASHINGTON, Aug. 27 (DC Times Online) — NASA Administrator Jared Isaacman says one possible answer to the growing demand for artificial intelligence computing is to put data centers in Earth orbit and power them with sunlight.
The idea could reduce competition on the ground for land, water and electricity-grid capacity, Isaacman said in an interview with Nexstar DC. But orbital data centers remain largely a proposal. The supplied research does not establish that a full-scale facility has launched or begun commercial operations.
What is an orbital data center?
A data center is a facility filled with computers that store information and perform calculations. Artificial intelligence systems require large numbers of specialized computer chips, which use substantial amounts of electricity and produce heat.
An orbital data center would place some of that computing equipment on satellites. Solar panels would generate electricity, while lasers or other communications systems would connect the satellites with one another and with networks on Earth.
Isaacman described the concept as a way to use the sun’s energy without placing more large data centers in communities. He pointed to the sun as a continuing source of energy and said orbital facilities could help the United States compete in both space and artificial intelligence.
Why put computing equipment in space?
The main argument is access to energy and space. Solar panels in orbit can receive sunlight for much of the time, without nighttime interruptions and some of the weather conditions that affect solar generation on Earth.
Orbital facilities also would not need large tracts of land, cooling water or a direct connection to the local electric grid. Those needs have become central to debates over new data centers on the ground.
Space could offer another potential advantage: heat can be released through radiation rather than moved through cooling towers or water systems. A cooling system developed by Caltech and Sophia Space has been described as a possible way to support solar-powered space data centers, with the companies discussing potential use in the early 2030s, The New York Times reported.
These benefits would come with a major trade-off. Equipment, power systems and replacement parts would have to be launched from Earth, and communications would have to work across long distances and changing satellite positions.
What projects are being considered?
Google’s Project Suncatcher involves solar-powered satellites carrying the company’s artificial intelligence chips and linked by lasers, according to Google’s project description. Scientific American reported that Google planned a demonstration mission for 2027.
Starcloud has also pursued computing in orbit. Scientific American reported that the company launched a 60-kilogram satellite carrying an NVIDIA H100 graphics processing unit, a type of chip used for artificial intelligence workloads. That kind of mission is a demonstration of on-orbit computing, not proof that a commercial data center can operate at full scale.
Aerospace America reported that SpaceX and other companies have proposed large satellite networks for artificial intelligence processing. It also reported that companies in China have announced plans for on-orbit artificial intelligence and that several U.S. companies are seeking regulatory approval for orbital data centers.
The European Space Agency announced a contract with Edge Aerospace to study orbital data centers, according to Aerospace America. Satellite proposals would also involve U.S. regulators, including the Federal Communications Commission, which oversees many communications and spectrum matters involving satellites.
What could stop the idea?
The engineering problems are substantial. Orbital systems would need large radiators to shed heat, hardware that can withstand radiation, reliable high-speed communications and launches that are much less expensive than they are today.
Researchers have also raised questions about scale. An arXiv paper examining orbital data centers said generating 5 gigawatts of power in orbit would require solar arrays measuring about 4 by 4 kilometers. The paper warned that large satellite networks could affect views of the night sky and increase the risk of collisions and space debris.
Debris is a particular concern because a damaged satellite can remain in orbit and threaten other spacecraft. A large constellation would also require coordination over orbital positions, radio frequencies, safety and eventual disposal.
Cost is another barrier. Every satellite, solar array, computer chip and cooling component would have to be placed in orbit, maintained or replaced. The supplied research includes estimates and projections from companies and researchers, but it does not independently establish that orbital computing would be cheaper than building data centers on Earth.
Is this a near-term replacement for data centers on Earth?
No. The currently described systems are proposals, studies or small demonstrations rather than a proven commercial alternative to terrestrial data centers.
Google’s planned demonstration, proposed large constellations and research into orbital cooling could show whether the technology works in stages. But a useful commercial network would require reliable launches, durable equipment, fast communications, regulatory approval and a way to control heat and debris at very large scales.
For now, Isaacman’s comments place orbital computing among the options being discussed as artificial intelligence increases demand for electricity and infrastructure. The concept could shift some of that demand away from land, water and local grids, but it would move the problem into a much harsher and more regulated environment: Earth orbit.
