Comment Doing the right thing for the wrong reason. (Score 4, Insightful) 115
Sometimes you can do the right thing for the wrong reason. Datacenters are misguided. Soon enough models and hardware will be good enough that everyone will be able to run their own local AI at home. It's just like computers back in the 60s. It's not going to stay on server farms forever. The future is personal AI just like the future was personal computers, and not even at the prices Nvidia is talking about. Normal computers will be able to run frontier models. When that happens, data centers are going to be worthless. They'll be trillion dollar investment failures, built for a future that doesn't exist.
And that's the optimistic take! What happens if the AI bubble bursts and this all goes south? Then all those data centers are worse than worthless. All that land, money and water they used will be totally wasted, and we're not talking an insignificant amount.
Having skepticism about the future of AI is a good thing, even if it has a future. It's not healthy to be building these trillion dollar projects based on wild speculation about a future that may never happen even in an optimistic scenario. If this all goes south, everyone who didn't build a datacenter is going to be better off than anyone who did.
So sure, let China go crazy poisoning their land and water with bad investments in obsolete data centers, while we proceed with a little more caution and sanity. I don't know about you, but that seems healthy to me. One advantage of living in a free and open society is you can express skepticism of bad ideas.
And that's the optimistic take! What happens if the AI bubble bursts and this all goes south? Then all those data centers are worse than worthless. All that land, money and water they used will be totally wasted, and we're not talking an insignificant amount.
Having skepticism about the future of AI is a good thing, even if it has a future. It's not healthy to be building these trillion dollar projects based on wild speculation about a future that may never happen even in an optimistic scenario. If this all goes south, everyone who didn't build a datacenter is going to be better off than anyone who did.
So sure, let China go crazy poisoning their land and water with bad investments in obsolete data centers, while we proceed with a little more caution and sanity. I don't know about you, but that seems healthy to me. One advantage of living in a free and open society is you can express skepticism of bad ideas.