Crusoe CEO Chase Lochmiller on AI infrastructure bottlenecks, blue-collar workforce, and the ratepayer protection pledge

shaped like the product that you sell. The billionaire who runs the world's largest battery company is confident that Americans will come calling eventually. We will return to the story later in the show. But without further ado, we have Chase Lock Miller from Crusoe in the ream waiting room. Chase, how are you doing?

What's up, guys? How you doing?

We're doing great. How is Hillen Valley? You're the first person that we've talked to today. Give us a read on what's happening on the ground. How are you feeling?

Um, you know, it's great. It's it's great to see the partnership. uh between uh Silicon Valley and DC really come together. Um I think you know you can really feel that uh there's a very pro business climate here and a lot of members of the administration um here you know helping support uh a lot of the big technology initiatives across you know everything from AI to defense tech to you know other other software businesses. So uh I think it's uh very very good vibes here. How are you introducing Crusoe when you meet new folks in Capitol Hill today? I mean, there's the the company has been How many years have you been working on this? It's been a number of years, correct?

Uh, yes. So, the company's about eight years old.

Okay. Overnight success.

There we go. Overnight success. But, but yes, how how are you

to the right? Yeah. uh it's it you know I really introduced the company as a vertically integrated AI infrastructure business where you know we focus on both the physical aspects of bringing AI infrastructure to life land power data center development design engineering uh and deployment of large scale clusters of GPUs as well as the software aspects of bringing AI to life uh between orchestration of workloads across large clusters of GPUs, management of GPU clusters with uh you know self-healing aspects like our autoclusters product. Interesting. Um and and and our managed Kubernetes product as well as uh you know serving uh applications as they scale with things like our Cruso managed inference product where we can serve and host models and and really uh

you know uh give give the key AI applications the tokens they need.

We've been talking a ton about bottlenecks. You you outline sort of like four or five pieces of the stack there. It feels like the managed Kubernetes cluster, the software development, maybe that's getting easier with AI. It feels like we have a lot of land. Is are energy and chips the real bottleneck for you these days? Like what are you seeing in terms of things that need to be unblocked? Things that the industry needs to be focused on uh going forward over the next few years. I think data center capacity at large is uh absolutely a bottleneck and you know that with that broad bottleneck I think there's a lot of underlying aspects whether it's you know the power to energize the data center the people to build the facility so like labor is a major constraint um and you know I think there's incredible bluecollar workforce opportunities across the entire ecosystem to help build this infrastructure of intelligence for uh the future Um, you know, I think chips are not a bottleneck today necessarily, but you know, energized chips are. So, you know, again, it comes back to this, you know, power and data centers really being the the critical bottlenecks.

Is there something that equates to a 10x engineer in electricians? Is there a 10x electrician? And are there are there recruiters that can help you find those people or is it a whole new game? So there are there are definitely like uh you know like anything there's like dispersion amongst performance for uh you know folks like like electricians but uh I would say the impact of it is different than um you know in engineering like you know with a 10x engineer you know sometimes it's like that's the difference between uh actually being able to build a product and not build being able to build a product. Um, you know, I think with a uh, you know, the call it a 2x engineer or 2x electrician or something like that. Um, you know, it's the difference between like this thing taking, you know, 40 hours and it taking 30 hours or something like that. So, it's it's less of a uh um, you know, uh, gamechanging aspect versus like it it just may take more time. Do you have a do you have thoughts on what we as a society, as a tech industry, as a country can do to reskill, upskill towards this like higherend bluecollar work? Uh do there need to be new training programs? Do more taxpayer funds need to flow into retraining programs? like I've seen some policy proposals kicked around that all seem exciting, but is there anything that sticks out to you as particularly beneficial over the next few years?

Absolutely. This is this is something that, you know, I'm personally spending uh quite a bit of time on is like how do we you know reskill the labor force to you know prepare them for this you know incredible opportunity that's you know under underway in the in sort of the big AI data center buildout. Um, I think, you know, look, there's there's existing trade school programs. Um, and there are a lot of them are are are older. They're, you know, sort of, you know, these very legacy training programs and and it's great. We need people to go through them, but I I think there are certainly incredible opportunities to innovate across that uh trade school training program. I think like you know looking at a model like Alpha School and what they were able to do in terms of accelerating personalized education journeys and development journeys um to you know have really high performing uh students in in in sort of you know the the core K through 12 metrics. Um I think applying that same sense of like AIdriven um teaching assisted you know assisted teaching and uh curriculums I think can be uh a huge unlock to you know get get the folks we need here. Um and and as well as I think like you know we're exploring ways to have you know and naturally I think a lot of trades sort of work this way but we're exploring ways to basically have you know apprenticeship programs um where there's a lot of on the job training and and you know there's huge projects that these folks can be a part of and and really learn by doing and uh you know I think there's ways we can sort of guarantee jobs you know when they graduate and you know things like what lambda school did for coding I think you know there's an opportunity uh similar here in in in the blue collar trades. And then the final piece is like I think in the K through 12 education programs, we've gotten away from a lot of the you know working with your hands programs like things like autoshop and wood shop and you know uh you know where people are are building things with their hands and and and being taught that in uh you know

I would love to take data center shop class

data center shop server. Yeah,

that's great.

Unironically, that would be like class.

Yeah. I sorry, Jordan, you have a question.

Yeah, I want So, with the conflict in the Middle East impacting energy markets, capital markets, what are you paying most attention to? What are what is Crusoe doing to kind of mitigate any of that impact? Obviously,

uh it's, you know, very unknown how long it'll last right now, but I'm sure it's, you know, extremely top of mind. Um yeah, this has actually been pretty amazing because uh you know while oil markets have you know reacted um quite a bit and you you've seen oil prices go up and gasoline prices as a result going up quite a bit um natural gas really hasn't um and and that's largely because of uh the energy independence that's been achieved here in the United States through you know investments from you know the energy energy uh innovators and and private sector um as well as support from the Trump administration to uh be a more energy independent nation. And so what you saw in the natural gas market was that there was initially a big spike up in in in the price of gas uh natural gas and then it came right back down to basically where it was uh before the conflict started. And um you know I think that speaks volumes to um you know the investments the administration's made and and the private sector's made in terms of uh creating that energy independence for America. And it's insulated folks like us uh from massive price fluctuations and one of the key operating expenses for these data centers which is the cost of power.

Interesting. Uh talk to us about the rate payer protection pledge.

Yeah. Be before we go any further, I'm just curious like you know you know with with so much of global uh production and supply being taken offline, how long can you realistically expect prices in the US to stay stable? Is there some dynamic? Well, that I mean that's my point is like natural gas is inherently more regionally priced. Like it's more affected by regional supply demand aspects like the global price market really is only impacted with if you if you're entirely reliant on LG um and uh so you know markets like you know Japan are are are more you know LG driven but like domestically here in the US like the main uh spot market people are focused on is Henry Hub and Henry Hub Like we're producing a lot of natural gas and we're consuming that natural gas which is different than the oil markets where we might be fracking but then we are dependent we need to we're not necessarily refining it and consuming it locally. Do I have that correct?

There's a couple of different dynamics there. The US is a net exporter of natural gas via LNG. M

um but the difference between gas and oil is that with gas the the the uh the LNG process where you're actually liquefying the gas that is a huge expense. So, so, so it basically upticks sort of the overall cost of that gas

versus oil. Um, you don't have that same challenge, right? So, if if if oil if there's a shortage, even if we produce all the oil we need here in the United States to serve, uh, American consumers, we're more impacted by the global market because, you know, we could basically export that oil and it's like not a huge cost uplift. the price is so much higher to sell oil in Europe or whatever, you just put it on a barge and, you know, send it across the ocean.

Okay. But with natural gas, you have to liquefy it. That that's a big cost increase. And so it kind of even evens out, normalizes.

Exactly. So we're we're, you know, because of the amount of production of natural gas and the natural gas resources we have here in the United States, we've been pretty insulated against some of these uh, you know, conflict driven commodity price action.

Okay, last question. We'll let you get back to Hillen Valley Conference. um rateayer protection program. This feels like what Crusoe was built to do and your experience has been uh generating power on site going back to you know flared natural gas. But uh walk me through how you've processed uh the actual roll out of that pledge, how feasible it is, how realistic is it for all the different market participants to adhere to it because it is just a pledge. doesn't feel like it's a law not yet at least. Um, but how have you been interpreting it and how optimistic are you that it will have the intended effects?

Look, I think it's a wonderful trend. Um, you know, I think the technology industry as a whole. The last thing we want to do is drive up the cost of power for grandma, right? That's like not at all, you know, the intent of uh the technology industry um and sort of the investments that are being made by AI. So I think we have some of the most well- capitalized businesses in the history of business that are, you know, making huge investments in the future of AI infrastructure and they're willing to make those investments across the data centers, the chips as well as the power. And to me, this is a generational opportunity for society at large to basically upgrade, you know, generation infrastructure, upgrade transmission infrastructure, upgrade the grid, um, to be able to support far more capacity than we have in the past. And you know the rate payer protection pledge basically um I think helps create a platform for um companies to be investing in the energy to power these AI you know these AI factories and you know this has been sort of the Cruso philosophy all along. Um I think the trend is going to be you know a new term that I've I've been using is is this term called across the meter. Um and you know people have maybe heard behind the meter power where you have like you know power gen on site behind the meter. Um the notion of across the meter is that you have both a a a generation resource that can be a gas power plant, it could be a wind farm, it could be a solar farm, it could be a battery, uh you know a large cluster of batteries. um you have a load um which is a you know a data center and then you have a point of interconnection and you know the load can basically consume all of the power that's generated on site that it needs. um any shortfalls can be pulled in and be firmed up by the grid interconnection and then any excess power that you have from you know the on-site generation can actually be supplied back to the grid

and so so you get this like you bidirectional aspect of of the grid interconnection point and that's really why it's called across the meter um but you know this why this is beneficial is you know the investment in AI data centers ultimately leads to energy abundance because you you actually have all this on-site generation where you're not using a lot of it some portion of the time and that that excess energy can be supplied to the grid and actually drive down the cost by uh for local rateayers. So that's kind of the you know the future and the vision that we're really building towards.

Yeah, I'm I'm extremely uh excited about it and it feels like uh just the perfect thing at the right time where a lot of Americans are asking for an answer and this is like the first step in the right direction.

Yeah. If if if that can be done at scale, the people are like, "Hey, I actually like these AI videos after all. They're driving my cost down.

They're driving they're driving my uh my rates down."

Well, thank you so much for taking the time to join us. Have a fantastic remainder of Hill and Valley and we'll see you soon.

God bless everyone.

Thanks, guys.

Talk to you soon. Goodbye.

Let me tell you about graphite.dev. Code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. And let me tell you about Lambda. Lambda is the super intelligence cloud building AI supercomputers for training inference that scale from one GPU to hundreds of thousands.

Sabrina Halper says easy talent hill and valley. Briefcase is equals hell. Backpack equals valley.

What is what is the third option?

What is the third option?

Reova briefcase. Uh, not not briefcase, the full luggage. Do you bring your full luggage with you? How do you stand out? Backpacker backpack.

What about trunks? We got to bring trunks back.