Musk's Space AI Data Centre Plans Are Moronic
Musk has diarrhoea of the mouth and constipation of the brain.
The stupid spectrum has a fascinating flaw. You’d think the more stupid something is, the easier it would be to spot as stupid, but that is only true to a point. Past that point, when something is monstrously dense, we lose all context, and it can be hard to nail down not just how brainless the idea is, but what about it is even daft in the first place. So, sometimes, the more stupid a plan, claim, idea or alike is, the harder it is to recognise how neurologically challenged it is. I can think of no better example of this than Musk’s recent claims that SpaceX plans to deploy 100 GW of solar capacity in space annually to power orbiting AI data centres. On its face, it sounds like a mildly ambitious task that SpaceX, as the largest space launch company in the world, could hit. Yet, it is possibly the most laughably blockheaded and assinine thing to have ever left Musk’s lips. And that is saying something.
For starters, some context. The US national grid, the second-largest energy grid on the planet, has a total energy capacity of 460 GW. Musk wants to build an energy system that orbits the Earth that is larger than this immense energy grid that took decades to build down on terra firma, in just five years. That should give you an idea of the scale at which Musk is talking about here. It is beyond James Bond villain levels of preposterous.
But, the realities of building, launching and deploying these satellites are, to put it mildly, impossible.
Take the solar panels themselves. You can’t use the same ones you strap to your house in space, the vacuum pressure, the giant fluctuating heat gradients and constant intense radiation would rapidly destroy them. For space, you need far more robust, and therefore, far more expensive, solar cells.
Gallium arsenide is the current go-to technology for space solar cells. They are efficient, relatively light, and insanely durable. There are some experimental advanced lightweight solar technologies being developed, but they are far from ready for deployment, and there is no manufacturing capacity for them. So, if Musk wants space solar, he will have to turn to gallium arsenide.
A space-rated gallium arsenide solar array currently costs around $300 per watt of capacity (according to polytequnique insights) and has an energy density of 350 watts per kilogram (according to KeAi). This means that 100 GW of these solar cells will have a mass of 285,714,285 kg, and would cost $30 trillion to purchase. Even Musk’s deep pockets couldn’t afford that.
These are, admittedly, very rough estimates. After all, if the solar cells don’t need to last that long, then they can weigh less and cost a bit less, and ordering this many would change the economy of production. However, these changes will only marginally optimise things, not deliver radical reductions in cost and mass. So, these figures do give us a good sense of just how expensive this endeavour would be.
