Elon Musk mocked critics who say artificial intelligence data centers cannot work in orbit because of cooling limits, arguing that some self-described experts dismiss the idea without understanding basic thermal constraints.
SpaceX Builds Around Orbital Thermal Constraints
"When someone who claims they’re super smart says that orbital AI is impossible due to cooling or whatever, but doesn’t even know things as basic as heat rejection of the radiator per m^2, coolant temp or max operating temp of the GPU," Musk wrote on X Monday.
Musk was responding to investor Gavin Baker, who said physicists spending only a few hours studying orbital computing should be cautious about declaring it impossible when Space Exploration Technologies Corp. (NASDAQ:SPCX) engineers have spent far longer on the problem. Baker cited a Keanu Reeves quote, which said, "Even if you say 1+1=5, you’re right. Have fun."
Cooling remains a major engineering challenge. Spacecraft cannot dump heat into surrounding air, so waste heat must radiate into space. Reuters reported in April that experts still question whether SpaceX can manage heat, radiation exposure, launch costs and hardware obsolescence economically.
SpaceX Builds Around Orbital Thermal Constraints
SpaceX says its AI1 satellite is designed around those constraints. Its Starmind page lists up to 250 kilowatts of peak compute and 175 kilowatts average, alongside solar power and radiators. The company has also asked the FCC to authorize up to 1 million orbital data-center satellites between 500 and 2,000 kilometers above Earth.
Musk, Isaacman Make Case For Orbital Compute
Musk has grown increasingly bullish. He previously called solar-powered orbital data centers a "no-brainer," predicted space would become the "lowest cost place to put AI" within two or three years and later said, "Space is the only way to scale at scale".
NASA Administrator Jared Isaacman has also backed orbital data centers, arguing recently that putting them in Earth orbit could tap abundant solar energy and help the U.S. compete in the space and AI races.
Lawrence Berkeley National Laboratory estimates data centers could consume 11.8% of U.S. electricity by 2030, with scenarios ranging from 9.5% to 15.3%.
Photo courtesy: Shutterstock
Login to comment