|
actually, I'm even more convinced of my idea now, and I'm apparently not the only one: https://wlockett.medium.com/the-spacex-xai-merger-is-a-giant... But to you're point here, I disagree with this take: > I think most people would allow a datacenter that isn't using electricity from the grid, isn't consuming water, and isn't polluting the air or with noise. Am I wrong to think that? This may be uncharitable of me, but I don't think people will. People in the states generally hate AI. Hate it. And, frankly, most people don't know how any of this stuff works. They don't know how the inference gets done and they couldn't care enough to try to learn it. It's kind of like the meme "keep your government hands off my medicaid" that was going around a few years back. People just... totally don't get it and will hate what they hate. Even the people I agree with politically often are absolutely unable to envision anything other than the status quo. People can't even begin to imagine a world where there is no obligation or requirement to work. We could build that world and free ourselves from the tyranny of scarcity, but we're going to need AI to help, and the transition (what we're going through now) is going to be rocky. Even now you can find people on here who say, "AI gets everything wrong" etc. People will fight musk building something simply because it is Musk. No other reason is needed. People don't actually give a damn about the environment. (this has been hard for me to swallow as an environmentalist at heart, even if I did work for an oil company for a bit). I learned this when I was doing some flying to some mines most directly. Which mines get opposed in the popular press? The mine that happened to be close to a place where a wealthy bank/family had property (a family lodge) that could be wrecked if the rivers got polluted. A hundred or two nautical miles away the same story was playing out with another piece of earth. That mine was unopposed because... well, nobody wealthy gave a damn and the people who inhabited the villages around it were poor. Why do people talk about ANWR like it's some sort of glorious beautiful location of resplendent natural beauty? Because someone said it once, and they're more stochastic parrot-like than the AI lol. I've been up there, lived up there, etc. I'd much rather we protected a few other places (especially some extraordinarily beautiful places that have been absolutely devastated by logging, but I digress), just saying, and ain't nobody going to go up there and spend time up there unless they need to for work. That is more or less the level of engagement and understanding I expect people to have about this stuff. They read a few Musk tweets, thought, "good lord what a train wreck" (which is largely the right takeaway, to be fair), and will now oppose everything he does, even if it is actually something is probably good for the world. Generally, people don't care about where things come from, how their food gets here, the logistics, etc. I'm being overly broad, obviously some people care, but the amount is far smaller than you'd think, and largely, people will oppose datacenters in America because it is popular to oppose datacenters in America presently. It wouldn't matter if it was a datacenter dedicated to curing cancer and ending world hunger that was entirely sourced from green and conflict-free minerals, you'd have a non-trivial amount of people opposing it. Now for this:
>The way I look at it, land for the actual datacenter isn't a delay. Start that permitting at an appropriate time in your process and it'll be done before you're actually ready to build. This is right in the "old world" way of thinking about it. Like, if this was the problem 5 years ago, this would be totally right. However, today these guys all view this as a race to AGI and more realistically ASI. If I had effectively infinite money (and Musk basically does), and I wanted to start building a datacenter in here in Alaska, if I built in town it could be 4-8 months to approve the permitting for construction. Double that timeline down south in more desirable parts of the country or with adequate infrastructure. If I'm in a race to build a datacenter as fast as possible, 4-8 months is billions of dollars on the back end when you're being your opposition. |
For the delay part, how many months do you think it takes to set up a substantial satellite production line? And go through enough test versions that you're ready for prime time? I think it's a lot more than 16 months. With a bunch of money you can get enough land and permits for a decade of datacenters long before you can get your first gigawatt of satellites launched.