The “Upfront Carbon” Cost of Google’s “Clean” AI Plan

Google’s “Project Suncatcher” is being positioned as a “clean” alternative to terrestrial datacenters, but it comes with a high “upfront carbon” cost. “Launching a single rocket into space emits hundreds of tonnes of CO2,” a fact that directly contradicts the project’s “green” credentials.
This is the central paradox of the orbital AI plan. To build its “compact constellations of about 80” satellites, Google will require numerous high-emission launches. This means the “clean” AI of the future will be built on a foundation of “dirty” CO2 emissions.
The industry’s defense, as articulated by Starcloud’s co-founder, is that this is a trade-off. He claims the “only cost on the environment will be on the launch” and that the system will yield a “10 times carbon dioxide savings” over its life, thanks to 8x-more-efficient solar power.
This “carbon trade-off” is a high-stakes gamble. It asks the public to accept a definite, immediate “carbon cost” in exchange for a projected, long-term “carbon saving.”
This plan also ignores the “other” environmental cost: the light pollution that has astronomers concerned about “bugs on a windshield.” Google’s 2027 prototypes will be the first test of this controversial “carbon math.”