“I’m sorry, Dave. I’m afraid I can’t do that.” - HAL 9000, 2001: A Space Odyssey (1968)

Anyone who has ever seen a space movie knows the risks in space are plentiful and severe. From aliens, to blackholes, and rogue artificial intelligence there is no shortage of dangers in the great beyond.

While these risks specifically won’t factor into the design requirements for an orbital data center (although the jury is still out on AI turning evil), there are very real risks associated with space beyond the obvious engineering and economic hurdles we discussed earlier this month.

There is no singular regulatory body for space. Most countries have their own agency and while efforts are made to collaborate on a global scale no one agency has final say in this process.

There are roughly 45,000 tracked objects in orbit around the planet and a couple million more that aren’t tracked (check out this map). Radar is used to track objects which limits the size of detection to be about 10cm in diameter, anything less than this the sensors cannot pick up (the vast majority of untracked objects are less than 10cm). Additionally, the agencies that run these tracking operations have limited budgets and having enough sensors to keep track of everything is infeasible.

The other difficulty in preventing collisions/accurate tracking data is speed. At 600km above the earths surface (the altitude of our proposed data centers) objects are traveling around 7,600 ^m/_s or roughly 10x the speed of a bullet. This is comparable to standing on the Utah-Colorado border while tracking a flying bagel at Mach 20 on the other side of the state (sounds hard).

The other issue here is kinetic energy. At these speeds, a sphere of aluminum with a 10cm diameter contains the same amount of destructive energy as semi truck going over 100mph. More than enough energy to make sure your satellite has a very, very bad day.

In 2009 Iridium 33 (previously active US satellite) collided with Cosmos 2251 (decommissioned Russian satellite). These two satellites created 1,800 pieces of new, tracked debris and many more untracked pieces.

A satellites useful lifetime is often governed by the amount of fuel it has left for orbit adjustments and correcting it’s orientation. Given the high cost to get things to space, operators don’t like to use this fuel for any other purpose (i.e. avoidance) unless absolutely necessary in order to maximize the useful lifetime of the satellite.

Prior to the 2009 collision, Iridium 33 was given a warning of a conjunction (close pass) with Cosmos 2251 but Iridium 33 declined to adjust their orbit. The estimated conjunction distance they were given was 584m and they decided that the risk of collision wasn’t worth wasting the fuel. Hindsight is always 20-20.

This chart outlines the probability over time of a satellite colliding with any object over 3mm. It is broken down by cross sectional area (CSA) and we can see that for a 50 m² satellite the collision odds exceed 80% in just the first 5 years. The mega-scale satellites being proposed would have a CSA of over 2.5 million square meters. At this size, the odds of collision would jump to 100% nearly instantly.

If a mega scale data center was to collide with another object in space, like Iridium 33 did, the amount of new debris created would be astronomical and could potentially kickstart Kessler Syndrome. The debris from this first collision causes another collision and the amount of debris grows exponentially until all satellites are destroyed (this process would happen in plus or minus 90 min).

In higher orbits, the friction due to air is so low that debris could remain in orbit for hundreds or even thousands of years without intervention. This would make earths orbit extraordinarily hazardous, effectively trapping humans on the surface of the planet without GPS or any form of satellite communication and no feasible way to launch more satellites.

Honestly, for this one singular reason, I don’t think we will be looking at space-based computing anytime soon, at least not with the proposed architecture. The risks are just too high. That is not to say it is impossible, but it does require some tall asks: object tracking will need to be significantly improved, agencies must avoid collisions even at the cost of their own project lifetimes, and all nations will need to work together to manage and safely operate what gets sent into space. We can only hope that everyone has the good of the world in mind.