Space based data centers

 Elon Musk, the person who brought you Tesla and SpaceX, has another idea. This time it is about locating data centers not on some place on earth, but in space. On the surface of it, this quite visionary.

As most of us have learned, a major challenge in a world awash in AI is that the energy consumption will go up enormously. The current grid can not handle this major increase and a solution must be found. Building data centers on ground has significant environmental, infrastructure and quality-of-life concerns. In comes good old Elon and the idea of space-based data centers. 

 

Before you scoff at this idea, remember that Elon has already put a space based “Starlink Constellation” which is allowing internet connections that do not require massive infrastructure investment. His company, SpaceX has a powerful rocket, Starship, which will be put to good use in launching future data centers.

 

By the way, Musk is not the only one who thinks this way, Sam Altman of OpenAI has a similar vision. Except the two don’t want to deal with each other, so Altman is looking at other options to get space-based data centers. Google is also interested in this type of data centers and is planning to launch a prototype in near future, as soon as 2027. Their project has a catchy name, Suncatcher. So, even if you are a skeptic of what Elon Musk says or promises, there are a bunch of other folks who are pursuing the same dream. Hopefully, someone will succeed. 

 

The energy to power the data centers will come from the sun. That brought back memories of an inventor at one of the companies I worked for, Arthur D. Little. Peter Glaser, a Czechoslovakia born Vice President of the firm had come up with the idea of Solar Power Satellites to generate power and beam it to earth. This renewable energy source would have done wonders, but the cost was too high, and there was a lot of skepticism, especially around the safety of power beams. This idea is not dead, Japan and China are still pursuing this type of satellites.

 

With the advent of AI, the game has changed. The power generated does not need to be beamed back to the earth but used onboard the satellite to power data centers. However, there are technological and financial challenges posed by this solution. 

 

A major challenge would be to get rid of heat of the data center. Since there is no air to carry it away, as is the case in earth bound data centers, some other way needs to be explored.

 

That brought back another memory, again from Arthur D. Little. Peter Teagan, a senior engineer, and a Vice President had come up with a brilliant solution to a similar problem. That time, the satellite in the orbit was going to be involved with space base lasers, generating a lot of heat. Peter came up with an idea that used a liquid belt radiator. In this, a belt would rotate around the satellite, dipping inside to go through a bath of liquid metal. When if came out at the other end, heat was released by radiation and metal on the belt solidified until it went back again. The rotation could continue indefinitely with very low power since there would be very little friction.

 

One major problem was what would happen if the satellite moved, say because of changing orbit or docking with another satellite. This would cause instability in the rotation of the fragile belt. Would it ever get back to doing its work? 

 

I created a compute simulation of the belt and ran it through various scenarios. The answer was that stability will be gained in a short time. NASA wanted to test the simulation, by running a belt in an airplane that dives to simulate zero-g. I desperately wanted to take a ride but was bypassed by a technician who would cost a lot less to the client. 

 

Musk is not talking of a liquid belt radiator. Instead, he will deploy massive foldable radiators to dissipate heat. They will be taken up by the powerful Starship rockets.

 

All this is going to require a huge investment. Musk plans to use upcoming public offering of SpaceX to fund the venture. I suppose, given the desperate need to power the upcoming AI revolution, things will happen sooner than later. If they do happen in my lifetime, I will look up in the night sky, see a space-based datacenter and think of my former colleagues at Arthur D. Little. 

 

-------------------------------

PS: I fully acknowledge that Elon Musk is a controversial character. People hate him for what he has done politically, especially in recent times. I agree, but at the same time, I believe that he is a man of vision and at least two of the industries he has launched are a testament to that. We all have opinions about this man, so let me have mine.