AWS CEO Doubts Commercial Viability of Space Datacenters for Decades as Musk Disagrees

On November 12, 2026, cnBeta reported that Matt Garman, CEO of Amazon Web Services, said during an interview with the Sources podcast that space data centers are unlikely to become economically viable or practically useful within the next ten to twenty years. Garman acknowledged that putting data centers in space sounds attractive, but argued that from both an engineering and a commercial operations standpoint there remain many hard problems. He specifically pointed to launch costs, lift capacity, heat dissipation and equipment maintenance as the key constraints on the idea. The report also noted that Elon Musk's SpaceX plans to use Starship to deploy large-scale orbital computing satellites and hopes to power AI computation with solar energy in space. Musk takes the opposite view, arguing that heat in space can be handled effectively by aluminum radiators pointed toward the cosmic background. Garman said building a large space data center would require moving massive amounts of servers, GPUs, solar panels, thermal management equipment and communications systems into orbit, while current launch cadence remains too low and costs too high to support such infrastructure. He suggested that for a space data center to be genuinely commercially viable, launch cadence might need to reach at least once per day, and even if SpaceX keeps working to increase Starship's launch capacity, one company alone cannot quickly solve the industry-wide lift bottleneck. Weight is also a practical issue. Garman gave the example that a full rack of GPU servers can weigh as much as one ton, so even if Starship eventually delivers about 100 tons of payload to orbit per launch, deploying a large-scale computing facility would still require a very large number of launches.