But despite the science-fiction rhetoric, the SpaceX S-1 filing’s citing of complexity and “unproven technologies” is one of the more realistic renditions of what is involved in sighting data centres of any size in orbit. Speaking to npr.org, Olivier de Week, professor of astronautics at MIT, described the solar panels capable of powering a gigawatt data centre as “feasible, but not next year and certainly not in three years.” To place the size of a necessary solar array in context, the international space station’s (ISS) solar panels cover the area of half a football pitch, and provide 100 kilowatts of power, according to npr.org. A gigawatt of power would need panels 10,000 bigger – 5,000 football pitches.
There are also problems, as yet largely unsolved outside the marketing-driven pronouncements of those with skin in the game, with the effects of solar radiation on delicate electronics as a whole, and on computing chips and storage specifically. Even the tiniest amount of stray gamma radiation can disastrously bit-flip binary systems.
Perhaps the most obvious but often misinterpreted aspect of data centres sited in space concerns the physical laws around conduction, convection, and heat radiation. Most people know that space is cold (a fraction of a degree above zero degrees Kelvin), but assume that therefore, placing hot data centres in space means the cooling issue is automatically solved.
However, those assumptions are based on operating DCs in an atmosphere, where the movement of air or liquid cools equipment. In the vacuum of space, there’s no medium available to take excess heat away other than radiating it away in the infrared spectrum.
The ISS uses a number of extending radiator fins of around 75 feet in length to help it maintain its operating temperature using this method. The solar arrays, their associated infrastructure, and an orbital data centre itself would require huge numbers of similar devices, all of which would have to be positioned out of the sun’s rays to be effective.


