Power Player

Chase Lochmiller is obsessed with power.

More specifically: the cofounder and CEO of Crusoe is obsessed with how to deliver the massive amounts of power AI needs to reach superintelligence and work its way into every facet of the economy.

Chase saw AI’s energy problem coming more than a decade ago when he was a quant trader using machine learning models to predict stock returns. When workloads required a GPU, Chase experienced firsthand how much more energy they consumed than a CPU.

“I realized that if AI really took off, energy would become the bottleneck,” he says.

That insight inspired Chase to found Crusoe (opens in new tab) in 2018 with an energy-first mission. Crusoe builds and powers AI data centers — the company calls them AI factories — by vertically integrating energy, design, construction, engineering, and electrical equipment manufacturing. It also builds the software to operate the factories. This approach has allowed Crusoe to build faster and with greater flexibility than competitors, accessing energy that’s been stranded due to geographic isolation (much like the company’s fictional namesake).

“We view ourselves as the supporting infrastructure layer that empowers this era’s innovators to do their life’s work,” says Chase. “We want to be the best in the world at activating megawatts for compute.”

Between building massive data centers for hyperscalers and operating its own AI cloud platform (opens in new tab), Crusoe faces no shortage of demand. But Chase is looking ahead to the next phase of the AI revolution, which he expects to involve new form factors and still unclear uses.

“The entire infrastructure category in AI is evolving daily,” says Chase. “We're always pushing things at the frontier to support new use cases, and we're taking all of those new data points and feeding them back into how we think about the next generation of compute infrastructure.”

Ultimately, AI’s economic value will depend on the success of the industrial revolution occurring alongside it, and the AI industrialists like Chase driving it. By accelerating an abundance of energy, Crusoe accelerates an abundance of intelligence.

Energy as the Bottleneck to Compute

To understand Chase’s approach to business, you need to understand Crusoe’s first corporate value: Think like a mountaineer.

For Chase, who has summited five of the world’s seven highest peaks, the motto is meant to embed a sense of cultural resilience.

“It’s this notion that everybody should be prepared for things to go wrong,” says Chase. “When we think about these big ambitions that we take on, there's sort of our Plan A, which is everything's going to go right. Hopefully that happens, but it's not always going to happen.”

Chase experienced such a moment when his first attempt to climb Mt. Everest was aborted while he was still at base camp after an avalanche tragically killed 16 guides in 2014. He returned four years later with a plan for random risks, and successfully reached the summit.

“I did a tremendous amount of planning this second time around,” he says. “I felt like I fully understood the problem and the risks I was comfortable taking.”

Chase has always had an aptitude for finding creative solutions once he deeply understands a problem, no matter how complex it is. That studious curiosity led him to initially assume he’d spend his career unraveling the mysteries of the universe as a theoretical physicist. But the monastic, low-pay life of a researcher wasn’t for him.

Instead, Chase went into finance, despite knowing little about it at the time. “A recruiter was like, ‘Hey, you're really good at math, and there's this thing called quant trading where if you solve these math problems, people pay you a bunch of money,’” Chase says. “I was like, ‘How do I sign up for that job?’”

A decade of quant training rewarded Chase both financially and intellectually, but working with emerging technologies fueled him.

“We were using a lot of classical machine learning techniques to forecast stock returns, and then, as deep learning became popularized, I got to work on a lot of new techniques and apply them to stock trading,” Chase explains. “All of that was very, very exciting.”

While still working as a quant, Chase earned his Master’s in Computer Science with a specialization in AI at Stanford (opens in new tab). He had the fortune of studying under AI luminary Andrew Ng (opens in new tab) shortly after the AlexNet convolutional neural network (opens in new tab) ignited the ongoing AI boom. Andrew convinced Chase that AI is a meta-science akin to electricity. It wasn’t about the potential to help in some areas; it about unlocking improvement in everything.

“Being at Stanford then was just a really special chapter,” Chase reflects. “I was AI-pilled very early.”

During this time a long-form blog called Wait But Why (opens in new tab) sharpened his thinking on the path to superintelligence.

“The bottlenecks are data and compute,” says Chase. “And what’s the bottleneck to compute? Energy. That was pretty much the thesis of Crusoe.” But it took a trip to the mountains (naturally) with his high school friend and eventual cofounder, Cully Cavness (opens in new tab), for the next chapter to coalesce.

Bringing Data Centers to Energy Sources

When drilling for oil, three things will come out: oil, water, and natural gas. The oil and water can be separated and hauled away, while the gas is essentially a waste product without a nearby pipeline. The most economical way to deal with the gas is to simply burn it off through flaring.

As they hiked through the Rockies, Cully – who worked in the energy sector and came from an oil and gas family– explained how the valuable resource was literally being set on fire 24 hours a day. Going to an oil field with Cully to see the fiery bursts filled Chase with urgency.

“It’s a very visual experience,” he recalls. “There’s just so much gas being burned off with no use. It’s totally insane.”

At this point, Chase had taken a job at crypo fund Polychain (opens in new tab). Chase knew the most important profitability metric for Bitcoin — and compute in general — was energy cost. What if instead of bringing energy to a data center, a data center was brought to the energy source (opens in new tab)?

Using $600,000 of his own money, Chase funded the design and build of Crusoe’s first modular data center that could operate alongside a remote energy resource. Oil companies were happy to have an alternative to flaring and the significant carbon dioxide emissions it produced.

Mining Bitcoin was the fastest path to monetization for Crusoe, but Chase felt conviction that AI infrastructure would eventually be the bigger opportunity. Creating an AI cloud product with an energy-first approach became his pet project.

Over time, Crusoe became one of the largest off-grid Bitcoin miners. But being tied to both crypto and oil meant exposure to their inherent volatility, especially during COVID when prices for both plummeted.

In 2022, Crusoe launched Crusoe Cloud (opens in new tab) in alpha. The timing was auspicious. Just a few months later, OpenAI would release ChatGPT to the public. Chase’s side project was suddenly the center of attention.

Winning Project Stargate

A fellow entrepreneur once told Chase the secret to success is surviving long enough for a miracle to happen. ChatGPT wasn’t quite a miracle for Crusoe, but it was an inflection point. Another happened in 2024 over drinks with an Oracle executive, when Chase learned about an AI data center project that would need upwards of a gigawatt of power.

Crusoe hadn’t built anything at that scale, but it had strong relationships with suppliers and its own electrical manufacturing facilities. Chase and his team scrutinized their supply chain to see if they could meet the aggressive timeline sought by the project’s leads, Oracle and OpenAI. Crusoe hit a bottleneck trying to acquire the medium voltage switchgearthat convert medium voltage from a power distribution center into low voltage that GPUs can use. The fastest they could get them from suppliers was 100 weeks, nearly twice the timeframe they wanted. But what if Crusoe simply built them? That, the team determined, would take about 28 weeks. Chase also knew where to find the power: a relationship he had from the Bitcoin world already had a 200 megawatt substation waiting for use in Abilene, Texas, and approval for another gigawatt.

Crusoe won what would become OpenAI and Oracle’s flagship Stargate project by committing to a timeline of a year when others balked at doing it in under two.

“Being able to build things ourselves was ultimately what led us to come up with a unique, creative solution when the market didn't have one,” Chase says.

Securing the first piece of the $500 billion Stargate Project vaulted Crusoe into becoming one of the industry’s power players (opens in new tab). In 2025, the company sold its Bitcoin mining business to focus on AI infrastructure.

“Chase is among the rare founders who combine domain expertise with unbridled creativity, giving him the ability to find solutions while others are still trying to identify the problem,” says Aydin Senkut (opens in new tab), founder and managing partner of Felicis. “When we met ahead of their Series C, we were blown away by his vision of where sustainable compute was going. And his instincts have proven to be absolutely right.”

Building the Birthplace of AGI

Crusoe has a lot of building to do. Crusoe's 2026 Series F valued the company at more than $30 billion (opens in new tab), propelled by more than $140 billion in total contract value and more than 20X year-to-date growth in Crusoe Cloud bookings. It’s already contracted 6 gigawatts of AI infrastructure for its data center projects and capacity for Crusoe Cloud, while its development pipeline exceeded 40 GW.

Being vertically integrated and a cloud provider itself, Chase argues, gives Crusoe a more holistic view of the needs of each project. “We feel we operate a better cloud platform and deliver better services because we know how to build the underlying data centers to serve them. We feel we build better data centers because we run our own cloud. So it's these complementary, circularly reinforcing muscles that get built out by building for hyperscalers. We're learning at the cutting edge. We invent new processes; we manufacture new components.”

Now, Crusoe’s focused on more modular builds. Giant data centers are necessary for training powerful models, but most uses, like inference, won’t require so much power. Enter Crusoe Spark (opens in new tab), its small modular AI data centers that can be dispatched alone or in fleets.

For giant data centers, construction sites are often remote, which requires temporarily moving labor there. But by manufacturing the “building blocks of compute” — the pieces that have to be built over and over again for each data center — as modular pieces that can be assembled on site, Crusoe shifts “the people bottleneck away from the site and into factories.”

And while a region like West Texas is a dream for the power needs of a hyperscaler-sized data center, thanks to cheap, abundant renewable energy, other projects will require more creative solutions. Crusoe is already building microgrids from end-of-life EV batteries with Redwood Materials (opens in new tab) to support Crusoe Spark deployments, and has partnered with Aalo Atomics (opens in new tab) to develop the first nuclear-powered AI data center, among other initiatives. Chase believes if the AI infrastructure buildout is done with an energy-first approach like Crusoe, data centers can support their own needs and send even more power to the grid.

As AI and the chips that power it continue to develop, Chase says “the data center is probably going to look a lot different to support them.” Changes may come sooner than anyone thinks.

In early September 2026, OpenAI introduced Astra (opens in new tab). It was trained at the Abilene facility Crusoe built. OpenAI president Greg Brockman said it’s so capable it could be seen as the dawn of the AGI era.

“That’s very fulfilling to me because that would mean we built the birthplace of AGI,” says Chase. “And it feels like we’re just getting started.”