Energy Hungry Intelligence: Chase Lochmiller (Crusoe) at CVAI London
📅 Jun 27, 2025Newcomer13 MIN422 VIEWS31 SEGMENTS · 2 SPEAKERS
Crusoe is in talks to build multiple 5 gigawatts AI data centers, each far larger than the current 1.2 gigawatt planned for its site in Abilene, Texas, which will power OpenAI’s models.
“You’re talking about building all of the data center capacity that we’ve built over the last decades—you’re talking about building all that infrastructure in a single location for a single customer in a short period of time,” Lochmiller said.
Crusoe, which started as a cryptocurrency mining business, is looking to differentiate from Microsoft, Google, and AWS with facilities built specifically for AI model t...
2:20So what's the difference between an AI factory and a data center in your mind?
3:28Have you thought a bit to the extent to which this infrastructure build is going to remake the labor force or what resources we're spending this on?
5:01So where do you fit in with as a neocloud and where is the role for a neocloud in this moment?
7:48How have you been thinking about kind of building up energy capacity?
11:00... play a big part in the conversations between maybe east-west, the DeepSeek moment, how China powers their AI movement, how do you think about where energy is going to be going within maybe also siloing different parts of AI development?
Excited to be here, too. Well, I want to jump in because we have only a little bit of time and a lot to cover, but one thing that I'm sure we would love to clarify a bit as much as we can on stage without lawyers in the background. I'd love to talk to you a little bit about Stargate right now. It's been mentioned that you've been involved in this project. Where do you see Stargate as it is right now? Is it a dream, an idea for powering this AI moment? How tangible is it? Where are things going?
Yeah. There's a lot of aspects to Stargate, but what I can comment on is what we're doing in Abilene, Texas, and a very large data center project that we've been working on there. That's a 1.2 gigawatt-scale AI factory that will house 400,000 GB200s. It's going to be a fully interconnected coherent cluster for a single customer. So it's a very, very large project. We think it's kind of the largest in the world. It's been sort of a massive investment in capital but also resources to really bring this project to life. So today we have about 3,500 people on site every day kind of working to make this happen around the clock and we're sort of bringing it up in record time.
Totally. And it's, I think, the fastest a 100-megawatt center had ever been built was a little over two years and basically we started construction about a year ago and we sort of started delivering the first capacity recently. So it's like less than a year, a little over 300 days for our first buildings and then the next six buildings I think our average time is like 215 days. Really, really focused on speed and really helping accelerate AI infrastructure and helping bring AI to life.
Definitely. I mean that's going to be a large undertaking but also massive opportunity. I mean you mentioned you called it an AI factory at one point. So what's the difference between an AI factory and a data center in your mind?
Yeah. So, you know, AI factory is sort of this concept that's really been coined by Jensen. And I think when people think about, oftentimes historically people have thought about data centers as these cost centers, right? They're these places where you're spending opex to kind of run your business. But when people think about factories, they're producing something of really meaningful material value. And what we're seeing with this new landscape of AI infrastructure is that what we're building is these massive factories that are steel, copper, large-scale infrastructure projects and then of course silicon and high-performance networking and what they manufacture is intelligence, right? So the input is power, steel, and silicon and the output is material goods which is driving productivity for the economy.
Sure. I mean it's definitely something that can be, you know, we talk about is AI creating or replacing jobs. Building AI factories sounds like it's providing a lot of jobs. Have you thought a bit to the extent to which this infrastructure build is going to remake the labor force or what resources we're spending this on?
Yeah, I mean we're sort of in a position where labor is actually one of our major bottlenecks. It's like we can't hire enough trade workers to build infrastructure fast enough and workers for the physical AI.
There's a lot of other things that we're working on, but labor certainly becomes a bottleneck when you're trying to hire many thousands of people to collectively work together to bring this infrastructure online to really scale AI.
Lochmiller, C. (2025, June 27). Energy Hungry Intelligence: Chase Lochmiller (Crusoe) at CVAI London [Interview transcript]. Newcomer. CEOInterviews.AI. https://ceointerviews.ai/interview/1352280/
