Back to Bursting the AI Bubble with Ed Zitron, with Ed Zitron.
Why Do AI Data Centers Take So Long to Build?
Ed Zitron says an AI data center takes far longer to build than the race for more compute assumes. He calls each one the single most ambitious infrastructure project in history.
Oracle has been building OpenAI's site in Abilene, Texas for two and a half years. He puts the building at about 998,000 square feet, drawing the power of a small city, and he uses Bristol in England as the picture: the power of that city, packed into several large warehouses. Hyperscalers are adding the capacity because they see demand, but he says most of the demand is OpenAI and Anthropic. He estimates that 80 percent or more of AI compute spend comes from those two companies. Microsoft alone, he says, has spent over a quarter of a trillion dollars on capital expenditures since 2022.
Bursting the AI Bubble with Ed Zitron ·
From the transcript
Oh, God, thank you. No, no, really. No, because– no, because, John, you're being self -effacing about not knowing this stuff. We actually– a lot of this stuff is very obfuscated. Okay. Purposefully so, I would imagine. Yes. On purpose. Okay. So it's not so much the largeness of the outputs. It's that it takes a long time to build AI data centers, way longer.
Each one of these things is the single most ambitious infrastructure project in history. Each data center. Pretty much. So OpenAI's data center, the Oracle has been building it for two and a half years in Abilene, Texas. It is, I think, about 998 ,000 square feet. Holy shit.
What's a power, which is basically the power of a small city, but in a thousandth of a space. So I use the city of Bristol in England as an example. It's basically the power of the city of Bristol in a one– in a– basically several large warehouses. We've got Americans right now scrambling to find out where Bristol is and what it looks like.