SPACEX’S proposal to put as many as 1 million satellites carrying artificial intelligence (AI) computing systems into orbit is raising questions about an environmental cost of terrestrial data centers. A report by Tereza Pultarovaw in Space.com asks what happens when the satellites and the rockets that carry them interact with Earth’s atmosphere.
The proposal is part of a broader effort by technology companies to develop orbital data centers. SpaceX, Amazon and Blue Origin are exploring systems that could use the nearly continuous availability of solar energy in space while avoiding the large water requirements of terrestrial data centers.
Rocket launches release gases and particles into the atmosphere, while satellites eventually reenter and burn up. Scientists are studying how those materials interact with atmospheric gases at high altitudes, where conditions differ substantially from those near Earth’s surface.
Eloise Marais, a professor of atmospheric chemistry at University College London, told Space.com that pollutants released at high altitude can remain in the atmosphere much longer than pollution near the ground.
“Close to the ground, we have rain and wind that removes pollution from the atmosphere within weeks. If we put it into the higher layers of the atmosphere, we rely on very slow atmospheric processes to bring this pollution back down,” Marais said.
Satellite reentries are of particular interest because spacecraft contain large amounts of aluminum. When satellites burn up, the metal can form aluminum oxide particles that enter the upper atmosphere. Researchers are investigating how those particles could affect atmospheric chemistry, including processes linked to ozone and climate.
The scale envisioned by SpaceX makes the question more significant. A constellation of 1 million satellites would be far larger than today’s commercial satellite networks and would require a continuing stream of launches, replacements and eventual reentries.
The effects would depend on factors including the mass and composition of the spacecraft, their orbital lifetimes, the number of launches required and the rate at which satellites return to Earth.
There is another environmental concern already visible from the ground. Large satellite constellations can interfere with astronomical observations by leaving bright trails across telescope images. More satellites mean more observations affected and more data that astronomers must remove or correct.
For the companies pursuing orbital computing, the potential benefits remain significant. Solar power can provide a large and relatively consistent energy source in orbit, while heat generated by computing equipment can be radiated into space rather than removed through conventional cooling systems.
Whether those advantages outweigh the environmental costs is not yet established.
A million-satellite computing network would represent a scale of space activity far beyond anything previously deployed. Scientists therefore face a new problem: determining how much material the atmosphere can absorb from an expanding space industry before its chemistry begins to change in measurable ways.
The orbital AI concept remains under development, and no company has demonstrated a computing constellation remotely approaching the proposed scale. The environmental effects will depend largely on how quickly the industry expands and how spacecraft are designed, launched and retired.