Dylan Patel: prediction

Transcript context

…So all in all, what does that imply for space data centers? Space data centers effectively are not limited by their energy advantage. They are limited by the same contended resource. We can only make two hundred gigawatts of chips a year by the end of the decade. What are we going to do to get that capacity? It doesn’t matter if it’s on land or in space. It doesn’t really matter, because you can build that power. Human capabilities and capacity could get to the period where we’re adding a terawatt a year globally of various types of power. At some point, we do cross the chasm where space data centers make sense, but it’s not this decade. It is much further out, once energy constraints actually become a big bottleneck and land permitting becomes a much bigger bottleneck as it subsumes more of the economy. And crucially, once chips are no longer the bottleneck. Right now, chips are the biggest bottleneck. You want them deployed and working on AI the moment they’re manufactured. There are a lot of things people are doing to increase that speed faster and faster. They’re modularizing data centers, or even modularizing racks where you put the chip in at the data center, but only the chip and everything else is already wired up and ready to go. There are things like this people are doing to decrease that time that you cannot do in space. At the end of the day, all that matters in a chip-constrained world is getting these chips producing tokens ASAP. Maybe by 2035, the semiconductor industry, ASML, Zeiss, and suppliers like Lam Research and Applied Materials and other fab manufacturers will catch up once the pendulum swings and we are able to make enough chips. Then we will be optimizing every dial and it makes sense to optimize the 10-15% of energy costs. As we move to ASICs potentially, and if Nvidia’s margins aren’t +70%, maybe that energy cost becomes 30% of the cluster. These are the things to optimize. But Elon doesn’t win by doing 20% gains. He never wins that way. Elon wins when he swings for the fences and does 10X gains. That’s what SpaceX is about. That’s what Tesla is about. All of his success has been about that, not chasing the 20%. I think space data centers will eventually be a 10X gain as Earth’s resources get more and more contentious, but that’s not this decade. Just to drive some intuition about how much land there is on Earth… Obviously, for the chips themselves, especially if you move to a world where you have racks that have megawatts—…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.