FROM THE DEPARTMENT OF: IS MUSK A LIAR OR A FOOL OR BOTH? PART 99.
And if there’s a sucker born every minute, is the U.S. a population of 340 million suckers given the apparent acceptance there of Musk’s latest scientifically and technologically illiterate claims about datacenters in orbit? Because Musk has made two contradictory claims about why that would be a good idea.
[1] Datacenters in space are a good idea because space is cold and the cost of cooling would be greatly reduced.
No.
Space is cold, as it has a background temperature of ~2.7K, or a few kelvin above absolute zero. But that’s the temperature of empty space as a radiation sink. It says f*** all about thermal management when there’s no atmosphere and the only heat dissipation mechanism is radiation, with no conduction and convection.
To radiate meaningful amounts of waste heat in space you need either very high temperatures or very large surface areas (or both). Look at a picture of the ISS: its thermal radiator system covers roughly 2,500 square metres and is one of its most complex subsystems, and it manages orders of magnitude less heat than a serious datacenter would generate. Datacenters in space would need enormous — like, maybe, kilometers wide — radiator panels. Any object in LEO would also be directly subjected to the Sun’s light every 90 minutes. That’s 120–150°C on Earth’s sunlit side, and can become higher. That’s a variance of 400-500°C, which adds an engineering problem of violent thermal cycling to that basic dissipation problem.
And just to beat this to death, no geostationary orbit will keep an object in shadow. The Earth-Sun L2 Lagrange point,does exist about 1.5 million km in the anti-solar direction, and gravitational and centrifugal forces balance there, so objects can maintain position relative to the Sun-Earth line. The James Webb Space Telescope is located there to keep its sunshield permanently facing the Sun and its instruments in cold, stable shadow. But the Sun is only partially occluded at that distance and the Webb telescope still needs its own sunshield rather than relying on Earth’s shadow.
[2] Datacenters in space are a good idea because sunlight in space is all free energy (with an additional helping of Musk nonsense about the Kardashev scale, with which all longtime SF readers are familiar and the rest of you can look up.)
So this has more substance, but is still moronically oversimplified by Musk. Yes, solar irradiance in space (~1,361 W/m² at 1 AU, the solar constant) is higher than what reaches Earth’s surface after atmospheric absorption. And in particularly high orbits like GEO or L2, you’re in sunlight for a high fraction of your orbital period, unlike LEO where you cycle in and out of shadow roughly every 90 minutes. But: –
[1] Space radiation degrades photovoltaic efficiency and meaningfully so over a satellite’s operational life.
[2] You still need to transmit the data from your orbital datacenter to Earth, which requires either laser or microwave links with distictly non-trivial bandwidth constraints and latency.*
[3] Everything needs to be built to survive launch and operate in vacuum with zero servicing. How does that work?
[4] Solar panels in space still have mass. Getting mass to orbit costs roughly $1,000–$10,000/kg with SpaceX’s Falcon 9. SpaceX’s Starship aims lower, but that hasn’t moved anywhere beyond rapid unscheduled disassembly yet.
* Not incidentally, there’s also the small geopolitical problem that this might be potentially weaponizable.
