Space firms eye orbital AI data centers as potential solution to ground power and land limits

Leading aerospace and tech companies such as SpaceX, Starcloud and Google are developing AI data centers that would operate from Earth orbit to bypass the massive power, land and water demands of ground facilities. According to a Boston Consulting Group analysis, the market could reach $240-$320 billion under a basic outlook and as much as $800 billion if technology and costs improve faster than expected, representing 10-15% of the global AI data-center market by 2040.

Elon Musk called the approach the "only logical solution" and predicts the first orbital units could appear within two to three years, while Segro chief David Slis dismissed the idea as improbable. Technical obstacles include the lack of atmospheric cooling, the need for large radiators, and the impossibility of on-orbit repairs, which together keep projected costs about 1.5 times higher than comparable ground sites even if launch prices fall. BCG also notes that only non-real-time AI workloads, such as batch processing of documents or climate data analysis, are suitable for space-based platforms.