Skip to main content
Chase Lochmiller

Chase Lochmiller

Nov 17, 2026

162:28
But what is sentiment like in the
162:30
country? Look, I think Iceland, the sentiments, you know, still reasonably positive.
162:36
It hasn't been overwhelmed in the same way, you know, I think the total grid in Iceland, something's like three gigawatts, something like that.
162:45
But, you know, you look at the Texas grid is something like 90 gigawatts and they had, you know, 400 gigawatts of large load interconnection requests.
162:53
And, you know, it was just sort of like...
162:55
A bit of overload there.
162:57
You know, I think energy intensive use cases have sought out Iceland as a home in the past.
166:19
buildings
24:25
That was a bold move.
24:27
Well, so the thing was is that I think externally a lot of people thought this was a pivot for us.
24:34
But internally, and if you talk to any of our investors or any of our employees that have been here for a very long time, they'll tell you that this was sort of the plan all along.
24:42
was this energy-first approach to computing.
24:46
I had been doing research in AI as a quant portfolio manager.
24:50
And then I was at Stanford for grad school and working in AI and doing a lot of work in computer vision and natural language processing.
24:58
And I was pretty AI-pilled early days.

26 MINS LATER

51:13
I've never met someone that climbed Mount Everest, and I'm really impressed because What is it like? What is the training like? What does it feel like when you get to the top of that mountain?
24:22
Hmm.
24:22
So those are all air-cooled chillers, which is basically these wound pieces of copper pipe where, where you have this giant chilled water loop in, in the data center that's recirculating water.
24:35
And it, it, you know, basically goes from cool water on the inlet side, it goes into the rack of GPUs, and there's a thermal transfer event from, you know, the GPU that's being energized and producing a lot of heat.
24:48
There's a thermal transfer event from the chip to the water, and then the water goes out through the, the, through these chillers.
24:55
You basically are blowing a bunch of air over these wound copper coils to exhaust the heat out of the system, so the water temperature steps back down, and you have cold water to then cool the GPUs again.
25:07
So again, that's another big investment.
25:10
All of the plumbing that goes into this is really, really substantial, so we actually, we have a ton of plumbers and pipe fitters on site that are, you know, welding these big plumbing systems together.

11 MINS LATER

36:11
[laughs]
12:01
I don't wanna kind of really actually get into that, maybe weirdly, but I want to ask your advice, which is when asked how do you advise someone when to endure, keep going, persist versus when to change tact and go down a different route?
12:15
route?It's a great question.
12:17
It's interesting investors that you spoke to validated this claim, but, you know, from the outside it does appear to be a pivot.
12:24
Internally, when we started the company, the goal was always to build an AI platform.
12:28
That was what I had sort of spent my career working on with AI machine learning, you know, algorithms.
12:34
And, you know, I've been a big user and early adopter of, of deep learning for, you know, certain tasks.
12:40
And I felt like it was this meta science that ultimately was gonna, you know, transform every, every industry.

32 MINS LATER

44:28
When you think about that, like, what's the unit? Is it like dollars per token, per training run? What is that unit for you?
40:03
Chase, w- with that, those kind of three buckets you just outlined, what's the kind of percentage of sales for each currently, and how is Crusoe's business kind of shifting of, of the m- in terms of the mainstay of what you do?
40:15
Well, you know, each of those different businesses, uh, each of those different ways of, of, uh, sort of serving customers, they, they typically have different durations.
40:24
So when, when folks are making data center commitments to us, um, we're typically working with very large tech companies that have in- investment-grade credit ratings, and they're typically committing to us but, you know, for somewhere between 15 and 20 years.
40:36
They're, they're multi-decade long commitments.
40:39
Um, you know, when you think about the life cycle of a GPU, uh, and the managed compute clusters, those can range anywhere from, you know, a few months to, you know, five to six years, but they're typically don't go beyond that just because that's sort of the, uh, the life cycle of, of the chips.
40:53
Um, and then when you think about managed inference and managed AI services, um, those are more, uh, you know, they're developer tools.
40:59
They're...
41:27
You've added three important names as board members, and looking at the round, you know, the size, the participants, a lot of people asking if this is kind of a precursor, uh, to you, Chase, taking Crusoe on a path to the public markets and an IPO.
103:39
Yeah.
103:39
Um, but you know, we, we sort of, even at that founding of the company, we had this ambition to build an AI cloud platform, which today is, you know, where, where, where Crusoe's at, is this vertically integrated energy first AI, uh, AI platform.
103:53
Um, so the early days it was roughly 250 kilowatts was the first-
106:27
Sure
106:27
... uh, which we're planning to go live with, uh, by mid-2027.
106:31
Um, but you know, what's unique about Spark is, you know, given the small modular form factor, there's so many different... you know, there's so many unique ways you can power it.
106:39
It can, you know, you can power it with like small gas generation things like RECiPs and, you know, smaller gas turbines like Aero derivatives.
106:47
Um, you can also power it just in, in u- with, with, uh, with grid power.
24:06
Can you give us a sense of the degree of the pause?
24:10
Uh, it is not a, uh, permitting issue.
24:13
You know, I'm, I'm actually not, uh...
24:15
I, I can't disclose, like, all the specific details, uh, around the, around the site, uh, and the commercial details and the commercial constructs that were in, in place between the customer and our energy partners at the site.
24:27
Um, but, uh, you know, the, uh, uh, you know, the, the, the work has basically paused at that site.
24:34
Now, we are seeing tremendous demand across, you know, other locations, uh, and, you know, more than we can possibly fulfill.
24:42
Um, and, and so we, we are continuing to see momentum in the development strategy.
26:08
Can you give us a sense of what your, uh, other pipelines are? 'Cause you, in addition to building out for certain customers, you have on-demand and reserve computing as well, and I'm just trying to get a sense of the different, you know, the different demand levels across your business.
58:04
Hmm.
58:04
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.
58:19
And, you know, the Ratepayer 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.
58:31
And, you know, this has been sort of the Crusoe philosophy all along.
58:35
Um, I think the trend is gonna be, you know, a new term that I've, I've been using is, is this term called across the meter.
58:44
Um-
58:44
Hmm
58:44
... and, you know, people maybe heard behind the meter power, where you have like, you know, power gen on site behind the meter.

Unknown podcast

The Future of Everything: What CEOs of Circle, CrowdStrike & More See Coming in 2026

Jan 25 · 2h 14m

110:15
housing, I would assume.
110:16
Yeah, housing, you know, housing, you know, the market's actually fairly efficient.
110:19
There's a lot of people that won't be long-term permanent workers there that are actually, you know, they bring mobile homes, you know, they collect a stipend.
110:28
It's kind of a very, you know, efficient process in that regard.
110:32
But, you know, long-term, we're going to have about 2,000 permanent workers at the campus.
110:38
So there's both, you know, short-term job, you know, spike and boom in terms of demand for the construction and installation process.
110:44
But there's also great long-term permanent jobs to operate the facility.

12 MINS LATER

122:48
What are we doing with the minnows? You're specifically targeting up and coming startups to be their provider.
speaker_6UNKNOWN
122:48
"What are we doing with the minnows?" You're specifically targeting up-and-coming startups to be their provider.
122:54
Yeah, so you know, uh, I really think about Crusoe, you know, being a vertically integrated business is like we offer three core things, right? We can, we can build the data centers, and we can rent those to, to customers.
123:04
We're really only focused on a small subset, maybe five customers, these very big tech, uh, tech companies.
123:10
Um, and that's really their key bottleneck, right? That, that's, that's their key pain point.
123:13
They're not, you know, renting out capacity from clouds because, you know, Meta doesn't know how to run a GPU cluster.
123:19
Like, they...
123:20
Of course, they know how to run, you know, big clusters of GPUs.
speaker_6UNKNOWN
125:53
... your current offering, I should say.
19:38
And I'm interested as to what sets you apart from a core weave and other neoclouds out there.
19:44
I think there's a number of things.
19:46
I mean, we've taken this vertically integrated energy first approach kind of from the start and the founding of the business.
19:56
I think a lot of the value that was unlocked from this first Stargate campus in Abilene was the fact that we had access to 1.2 gigawatts of power that could energize this very large cluster of NVIDIA GPUs.
20:10
Building it very, very fast required us to have creative designs and flexibility with a lot of modular components that went into constructing this very large campus with a lot of those modular components being manufactured offsite.
20:25
And I think we've built a very high performance software stack to enable these AI factories to work very effectively and at scale, integrating a lot of great managed services like our managed file systems and managed Kubernetes that enable innovators to do their life's work on incredibly reliable and high-performance infrastructure.
21:47
Just your reaction to what the administration's doing about speed matching supply of power to infrastructure?
21:55
Yeah.
114:46
Mm-hmm.
114:46
So what that means is we're building, um, you know, these factories that produce intelligence.
114:51
And this is both a hardware problem and a software problem.
114:55
Uh, and we focus on all of the things ranging from, uh, cultivating the energy and helping develop the energy resources needed to, uh, power AI infrastructure.
115:05
Um, we're building out the data centers themselves and redesigning them to fit these large-scale accelerated computing clusters.
115:12
Um, and then we, and then we also built out this whole software platform that operates these large-scale AI factories so that they can actually be effective in producing intelligence, uh, and really enable our generation's, uh, greatest minds and entrepreneurs to do their life's work and help build, um, eh- and scale their AI applications to impact people around the world.

5 MINS LATER

120:42
Hmm.
120:43
... um, you know, compared to, uh, other markets where, uh, you know, there was less da- digital infrastructure being built where, where prices actually went up substantially.
10:42
But I mean, how do you think about the level of risk in the AI ecosystem right now and the level of risk for your company, given that it's so capital intensive to build anything in this space?
10:54
Yeah, no, it's a great question, especially when you have a business that has a tremendous amount of CapEx.
11:00
The way we think about this is really, we are constantly doing demand forecasts in terms of how much compute demand is coming from our customers in the cloud business and how much infrastructure demand we're seeing from customers on the data center side.
11:20
And right now, I think the, It's very broad-based and the demand, no one has enough compute, no one has enough data center space, no one has enough energy.
11:34
The capacity for infrastructure is largely saturated right now and so we're really focused on building more to meet the demands of customers.
11:44
When we think about GPU purchases, uh you know there is kind of a you know uh some payback math that we go through in terms of you know making the upfront investment in the capex and working with our customers to uh you know ensure that they're able to get a great service and and high performance with high reliability and we're able to earn a great return for our investors
12:41
What keeps you up at night?
12:45
I think probably the biggest risk to the business is, or the biggest risk to the ecosystem, I think really is debt.
39:25
And whether the profitability is going to kind of be there at that layer, how fast you have to depreciate some of the technology that you install, and whether that's just kind of a treadmill that you're going to have to run indefinitely.
39:41
Yeah, I mean, there's a lot of different assets in this ecosystem, ranging from power plants to data centers to chips.
39:47
I think the most uncertain question is, what is the right depreciation schedule for the chips as new releases of new hardware platforms have accelerated? The way we view it is that there's certainly a lot of demand for the newest, latest, and greatest generations of chips to help train the next generation of frontier foundational models and scale inference workloads.
40:15
However, we still are seeing tremendous demand for historical and legacy generations of chips.
40:22
Blackwell is sort of all the talk on the street right now with NVIDIA releasing that platform recently.
40:28
But we're still seeing tremendous demand for Hopper, the previous generation, and even Ampere, the generation before that.
43:14
Is it Robinson Crusoe? Is it named after the book?
43:16
Yep.
JonHOST
45:01
Okay.
45:01
So, um, you know, there's a lot of investment being made today, uh, for these very large clusters, these a- large-scale AI factories, um, that are bringing, you know, intelligence to the masses and, and embedding intelligence in m- in the economy.
45:16
Um, but, uh, you know, the- a lot of that infrastructure is not used nearly as efficient- efficiently as it, as it could be.
45:24
A- and, uh, a lot of that has to do with, like, getting the data into the right place at the right time, um, so that your GPU utilization rate can go way, way up.
45:33
Um, and, you know, sort of our commitment in, in, uh, this Atero acquisition is, uh, you know, building both very, very large AI- large-scale AI factories, and then building the operating systems, um, and, uh, you know, the software systems to basically operate those AI factories with an incredible degree of efficiency, um, and, uh, scale.
45:53
So-
JonHOST
46:45
Yep.
46:46
And what we realized in addition, like, one of our, uh, uh, members of the founding team, Ben Chess, he came out of OpenAI.

Unknown podcast

Winning the AI Race Part 3: Jensen Huang, Lisa Su, James Litinsky, Chase Lochmiller

Jul 23 · 1h 4m

29:47
I'm Chase Lochmiller, the co-founder and CEO of Crusoe, and I'm here to talk to you about the AI Industrial Revolution.
29:55
I'm going to start with a quote, and it's from Warren Buffett in his 2020 shareholder letter to investors.
30:03
And he said, in its brief 232 years of existence, there has been no incubator for unleashing human potential like America.
30:12
Despite some severe interruptions, our country's economic progress has been breathtaking.
30:17
our unwavering conclusion, never bet against America.
30:23
Buffett's words were true then, and as we enter this global race for technological dominance of artificial intelligence, they ring even truer today.

11 MINS LATER

41:39
And then in data centers like Texas, what's the limiting factor? Is it workforce to actually go build these things? Is it materials? Is it the cooling towers? Is it the chips? Is it the hyperscalers giving you the contracts? What's the limiting reagent?
Lisa Su
Lisa SuGUEST
29:46
Yeah.
29:47
I'm Chase Lochmiller, the, uh, co-founder and CEO of Crusoe, and I'm here to talk to you about the AI industrial revolution.
29:55
Um, I'm gonna start with a quote, and it's from Warren Buffett in his 2020 shareholder letter, shareholder letter to investors.
30:03
And he said, "In its brief 232 years of existence, there has been no incubator for unleashing human potential like America.
30:12
Despite some severe interruptions, our country's economic progress has been breathtaking.
30:17
Our unwavering conclusion: never bet against America."Buffett's words were true then, and as we enter this global race for technological dominance of a- artificial intelligence, uh, they ring even truer today.
30:33
American dynamism has always prevailed, and it will continue to do so.

11 MINS LATER

speaker_6HOST
41:38
And then in like the, you know, data centers like Texas, like what's the limiting factor? Like is it like workforce to actually go build these things? Is it like materials? Is it the cooling towers? Is it the chips? Is it the hyperscalers giving you the like, you know, you know, sort of contracts? What's the, you know, sort of limiting reagent?
22:48
As a means to provide power to scale data center, this is a really significant step.
22:56
uh absolutely thanks for hosting us ed uh you know uh thrilled to be joining you here from uh from from the field where uh redwood and crusoe have partnered on uh building this uh very cool off-grid uh uh ai data center that's powered by on-site solar and end-of-life ev batteries you can see the the array of batteries behind me you can see this uh uh this mobile and modular ai data center that we've dubbed crusoe spark Each of these data centers houses about 500 GPUs, and there's going to be four of them here, all powered with this off-grid microgrid and batteries.
25:29
Are you able to kind of quantify or give our audience a sense of the challenge of how much a bottleneck the energy component is when you're building out these data centers around this country?
25:41
Yeah, absolutely.
25:42
You know, one of the major aspects to the bottleneck for energy really is the speed.
25:47
You know, people are applying for grid interconnections and they're finding themselves in multi-year, you know, five, ten-year queues to basically get that load approval.
25:59
What we've done here is rapidly deployed these modular and mobile data centers alongside this solar and battery installment.
26:09
And this all came together in about four months.
speaker_0HOST
86:03
Mm-hmm.
86:03
Uh, so, you know-A lot of the, you know, government labs and high-performance computing communities have been experimenting with things like immersion cooling, you know, single-phase and two-phase, as well as, you know, water cooling, uh, DLC, uh, for decades.
86:17
Uh, but no one's ever done it at the scale that's unfolding right now.
86:20
Uh, and the reason it's happening is because you just have so much energy density, uh, so much heat being produced by these new NVIDIA chips, um, as we move on to, you know, uh, more advanced architectures, uh, that there's, you know, simply just not enough, uh, heat capacity to basically move that heat off the chip, uh, from a t- traditional, uh, uh, aluminum heat sink.
86:43
Um-
speaker_1HOST
89:23
So I'm sure you have to be pretty...
89:26
Yeah, I mean, we're trying.
89:27
Uh, you know, we are, we have a couple other projects that are underway, um, that, you know, uh, hopefully we'll be able to talk about, you know, more soon, but, you know, uh, similar scale or bigger, uh, is kinda like, you know, what we're seeing.
32:13
Which part are you more responsible for?
32:17
So Crusoe as a business will handle the entire stack.
32:21
For this specific project, we're focused on the design, build, construction of this large-scale data center.
32:32
I think it's worth appreciating that data centers as a product have changed very dramatically in line with the changes in computer architectures over the last 20 years.
32:44
So if you look 20 years ago, a standard data center rack might have been 2 kilowatts, maybe 4 kilowatts.
32:49
Today, for the GB200 and VL72 configuration, we're budgeting 130 kilowatts per rack, all with direct-to-chip liquid-cooled water cooling.
33:01
And if you look forward to the next generation, what Jensen announced recently at GTC with the Vero Rubin Ultra, those are 600 kilowatts per rack.
35:20
to
2:21
So what's the elevator pitch for Crusoe?
2:24
Sure.
2:26
You know, Crusoe's a vertically integrated AI infrastructure business that focuses on unblocking some of the critical blockers that are preventing the proliferation and scaling of AI.
2:39
You know, this is both sort of down the stack from the actual chips.
2:42
So, you know, you talked a lot about the data center itself.
2:46
You know, we've built out a vertically integrated data center business that does everything from land, you know, and power acquisition and procurement and production.
2:55
all the way to data center construction, operation, engineering, and management.

8 MINS LATER

10:46
So just curious why Abilene outside of the power aspect of it is such a hotbed as well.

We value your privacy

We use cookies to understand how you use our platform and to improve your experience. Click “Accept All” to consent, or “Decline non-essential” to opt out of non-essential cookies. Read our Privacy Policy.