So how long could a data centre take before it could cease relying on a river-cooled industrial park to supply its power and begin to subsist solely on sunlight? SpaceX has submitted an outrageous variant of such a notion to the US regulators, asking them to allow a constellation that could eventually grow to one million satellites in the low-Earth orbit.
The submission made by the company of the Federal Communications Commission describes the plan as a move to install “orbital data centres” above the atmosphere in the air where there is abundant and continuous supply of solar energy. It also employs an eye-catching point of reference an advancement to a “Kardashev II-level civilisation”, a designation that references a scale created in 1960s which categorises hypothetical civilisations in terms of how much energy they are able to use. In such a plan, the yield of its star is seized by a Type II civilisation, a standard that is more of a cultural shorthand than an engineering roadmap.
Practically the suggested satellites would fly at between 500 and 2000km altitude, the same extensive neighbourhood as that occupied by the present communications constellations. An argument on X by Elon Musk is that such a network would be dilute in space even in itself: “The satellites will actually be so far apart that it will be hard to see from one to another. Space is so vast as to be beyond comprehension”. Yet, orbital mechanics concentrate action into a few useful shells, and it is this piling up, rather than the vacuum in the outer space, that has become the constraint.
The attraction is clear. The requirement of computing is escalating and site areas of terrestrial server farms are further caught up in grid enhancements, water access and local approvals. Orbital computing Concepts Orbital computing is based on the same physics: photo arrays in orbits selected with enough full sun exposure may provide power as they are not interrupted by night. Even academic and industrial scheming has focused on sun-synchronous orbits due to that reason, however, the cost to engineering is high: radiation protection and fault tolerance in electronics, and heat rejection of large radiators, which increase mass and complexity.
That bulk has to be thrown off, refurnished and finally disposed. A common goal that has been mentioned as feasible is reaching launch costs of less than 200 kgl by 2035, the target quoted in the coverage of the Project Suncatcher proposal by Google. The filing by SpaceX itself is pegged on the prospect of a fully reusable Starship with the argument that rates of flight might result in on-orbit access to processing capacity at a rate that exceeds that of building-out on land. It is also alleged in the same filing that orbital data centres may be less polluting than terrestrial facilities, but life-cycle assessments are disputed when launch emissions and the impact of re-entry are also included.
There is also the problem of how it will happen to the low-Earth orbit when it is filled. Debris experts have put it at around 130 million objects of orbital debris already in motion, comprising of the debris of dead spacecrafts and fragmentation remnants. This process of collision cascading that we sometimes call the Kessler syndrome is more difficult to disregard when whole fleets fall within the same altitude bands, and when the tracking and coordination is not evenly distributed among operators.
There is regulation, but not fully complied. Studies that have been conducted on post mission disposal have established that out of all the requirements by the existing regimes to deorbited satellites, only half are effectively deorbited despite the guidelines requiring satellites to be cleared within 25 years of mission end life within the critical zone of less than 2,000km. Suggestions to use the proposals to make launches gatekeepers that is, to block launches unless satellites are to common standards are intended to turn voluntary good behaviour into enforceable engineering practice.
Astronomy introduces another type of constraint; lack of interference which could not be swept away by improved tracking. Rosilla et al. (2021) have characterized the presence of strong out-of-band emissions of second-generation Starlink spacecraft, with flux densities in the 500 Jansky range and stating that such emissions can complicate radio observations far beyond the bands that satellites actively occupy. They have further claimed that the existing regulations consider such leakage as accidental and hence more difficult to policing thus mitigation is a matter of negotiation and not compulsory.
Finally, the biggest question that a million-satellite filing raises is not whether any of them will ever fly, but whether it is possible to scale orbital infrastructure and leave low-Earth orbit as a fragile, contested utility layer one in which the ambition to compute, the stewardship of debris and the night sky are competing to the same limited operating room.