SpaceX’s orbital AI plan could multiply launch emissions more than 150-fold, study estimates
The preprint projects large increases from launches and returning spacecraft, but missing technical details prevent a full accounting of the environmental effects.
A University College London preprint estimates that SpaceX’s proposed million-satellite AI constellation could push rocket-launch emissions above 150 times their 2025 level, requiring at least 5,000 launches a year versus about 330 worldwide in 2025. Its projections also flag pollution from spacecraft reentries and methane venting, but depend on assumptions because SpaceX has not disclosed key technical data. With no public assessment of the full proposal, the estimates underline a gap between launch-by-launch reviews and the cumulative environmental effects of operating and disposing of a constellation.
01
Using Starship Block 4 capacity of at least 200 metric tons per flight, the researchers estimate the network would need about 14 launches daily.
02
Spacecraft returns could produce an estimated 233,000 to 507,000 metric tons of nitrogen oxides annually, though reentry-speed data are scarce.
03
If methane venting continues as in test flights, the researchers estimate about 60,000 metric tons could escape each year.
SpaceX’s proposed million-satellite AI data-center network could push rocket-launch emissions above 150 times their 2025 level, according to a University College London preprint described by Scientific American on November 18. The analysis, awaiting peer review, also projects a sharp increase in pollution from returning spacecraft—but missing technical details leave important parts of the environmental cost unquantified.
Researchers Eloise Marais and Connor Barker examined the constellation SpaceX sought permission to launch in a January 2026 application to the U.S. Federal Communications Commission. Their work estimates emissions from launching and disposing of that network, rather than measuring pollution from a system already operating.
The researchers say SpaceX has not disclosed basic information needed for the calculations. They filled gaps with assumptions and third-party material, including space-enthusiast forums, Wikipedia and astrophysicist Jonathan McDowell’s spaceflight catalog. That limits the precision of the estimates and leaves some pollutants beyond their calculation.
A launch schedule far beyond today’s total
The model assumes SpaceX would use its planned Starship Block 4, carrying at least 200 metric tons into orbit per flight. Even with that capacity, the researchers calculate that the proposal would require about 14 launches a day. The annual demand dwarfs the number of orbital launches conducted worldwide in 2025.
The modeled launch demand
At least 5,000 launches annuallyProposed constellation
The researchers’ estimate assumes Starship Block 4 carries at least 200 metric tons per launch.
About 330Worldwide orbital launches in 2025
About 330 orbital rocket launches took place worldwide in 2025.
Returning spacecraft create a different problem from rocket exhaust. At hypersonic speeds, they heat surrounding air enough for nitrogen and oxygen to react, forming nitrogen oxides. These gases can drive reactions that destroy ozone, weakening the atmosphere’s protection against harmful ultraviolet radiation.
The team estimates spacecraft returns could produce 70 to 150 times more nitrogen oxides. Reentry speed strongly affects the amount formed, and experimental measurements are scarce. Using estimates derived from NASA space-shuttle studies, the researchers calculated an annual range of 233,000 to 507,000 metric tons.
Less soot does not settle the pollution question
Starship burns methane and produces less soot than SpaceX’s kerosene-burning Falcon 9. But if SpaceX continues venting unused methane during fueling and after landing, as it has in test flights, the researchers estimate roughly 60,000 metric tons could escape annually. Methane is a more potent greenhouse gas than carbon dioxide.
Satellite disposal adds another source. SpaceX plans to move 80 percent of the satellites into higher “graveyard” orbits. If the remaining fifth burns up on reentry, the team estimates about 200,000 metric tons of spacecraft material would vaporize in the upper atmosphere annually—more than 200 times the amount from satellites and other objects in 2024.
Burning spacecraft release metals that can form particles, accelerate ozone-destroying chemistry and potentially create high-altitude hazes that change how much sunlight reaches Earth. Yet the researchers could not calculate the constellation’s metal pollution because SpaceX has not disclosed the satellites’ material composition.
An assessment needs more than launch permits
No environmental assessment of the proposal has been made public. In September, the FCC exempted satellite megaconstellation filings from environmental review. The Federal Aviation Administration can still require assessments of rocket launches, but those reviews do not necessarily capture the cumulative effects of operating and disposing of entire constellations.
The disclosures the researchers want
Rocket fuel types and quantities, plus the mass returning to Earth during each launch.
Satellite lifetimes and material composition, along with spacecraft reentry speeds.
Marais and Barker say broad material percentages could improve estimates without exposing detailed engineering. Aaron Boley, a University of British Columbia researcher uninvolved in the study, suggests trusted intermediaries could handle sensitive information while enabling independent environmental assessments.
Reader comments
Newest comments first. Replies stay oldest first.