Eric Saunner grew up in Buell, Idaho, picking rocks in fields and feeding cows before school. He didn’t go to college. He was dyslexic and struggled academically. His original career plan was taxidermy. Today, at 46, he runs Data Airflow, a company with 75 employees solving the most complex thermal and cooling challenges in AI and high-performance computing data centers across the United States. He was named Entrepreneur of the Year for the Mountain West in 2026 and is a national finalist for the award.

On the latest Ever Onward Podcast, Saunner joined host Tommy Ahlquist and co-host Travis Hawkes of Capital Eleven for a conversation that covered data center myths, the real story on water and power, the race with China, nuclear energy’s timeline, and what it looks like when an Idaho farm kid figures out he can build spaceships after all, just different ones.

The Origin: Bitcoin Mining in a Barn

Saunner’s path to data centers is worth understanding because it explains why he thinks differently than most people in the industry. He built his career in HVAC through the union apprenticeship program, eventually starting his own mechanical contracting company in 2008, perfect timing, right in the middle of the financial crisis.

In 2015, a neighbor came to him with a problem. He had a Bitcoin mining operation in central Washington that he couldn’t cool properly. Saunner didn’t know what Bitcoin was. He went to look at the facility and found what he described as Home Depot shelves with wires hanging everywhere. His boss at the time laughed when he saw the pictures. Saunner went back and figured out how to cool it anyway, essentially applying HVAC principles to a problem nobody in his industry had encountered yet.

That early work led to an opportunity in Sweden, where he quit his job, ordered a truckload of fans, and watched the client disappear when Bitcoin’s price crashed. The fans ended up in Florida. He eventually sold them to a Bitcoin mining operation in Colorado, accepted payment in Bitcoin, and kept it. The rest of the story from there involves a chance meeting in the back of a rideshare at a conference, a business card with the wrong phone number on it, and a contract that went from 2 megawatts to 20 to 50 to 200 to a current pipeline that includes a 500-megawatt site in Kentucky and a gigawatt facility still in development.

“It’s been very fast,” he said. That may be the understatement of the episode.

Myth 1: Data Centers Are Drinking the West Dry

The most persistent concern about data centers in western states is water consumption, and it deserves a direct answer.

Traditional data center cooling uses evaporative cooling towers, the same basic principle as a swamp cooler, that can consume up to 300 million gallons of water per year for a 100-megawatt facility. That’s the number people are responding to when they hear about data centers and water, and it’s a legitimate concern for the open-loop systems that still exist.

The technology has moved on. Closed-loop cooling systems circulate and reuse the same water continuously. A comparable 100-megawatt facility using a closed-loop system might consume around three million gallons per year, roughly one percent of the open-loop figure, through a small amount of misting that helps improve efficiency. The water stays in the system. It doesn’t evaporate into the atmosphere or draw from aquifers.

“Nobody’s gonna put billions of dollars worth of servers into something that’s going to suck the lake dry,” Saunner said. States like Oregon are already curtailing data centers that haven’t made the transition to closed-loop systems. The industry is moving in one direction on this, and new facilities being built today are built to the new standard.

In context, the closed-loop water consumption of a modern data center is significantly lower than a golf course, substantially lower than most agricultural operations, and almost certainly lower than the water lost annually to evaporation and leakage from the New Canal system running through the Treasure Valley.

The Real Problem: Power

Water is the easy part of the conversation. Power is harder.

The Treasure Valley uses roughly 3.3 gigawatts at peak capacity. The first Micron fab uses about 500 megawatts. The Meta data center outside Boise draws around 127 megawatts in its first phase, eventually scaling to approximately 500. A second Micron fab would add another 500 megawatts. The math adds up quickly in a region where power generation capacity is essentially fixed in the near term.

Idaho Power has built its generation mix around hydroelectric, which is clean, reliable, and not expandable. Solar and wind development, which had been adding capacity through the previous decade, has largely stalled following the rollback of federal incentives. No new dams are coming. That leaves natural gas as the near-term bridge, with all the emissions constraints that implies.

The longer-term answer, which Saunner visited in person just before the podcast, is small modular nuclear reactors. Deployable Power, a company developing transportable reactor units, recently took one of its reactors critical, and the federal government has mandated deployable energy as a priority. Test modules are expected to be in place by 2028, with commercial deployment following. Saunner believes this is when the power equation fundamentally changes.

“Once that gets to a place where it’s reproducible and you let the free market in,” Ahlquist said, “that’s going to be ultimately the solution.”

Until then, the constraint is real. Any major new data center in the Treasure Valley or similar markets has to answer the question of where its power is coming from, and the honest answer right now involves natural gas generation behind the meter, interconnection to existing transmission infrastructure, and careful site selection based on what capacity is actually available.

What Data Airflow Actually Does

Saunner’s company sits in a specific and valuable place in the data center construction process. Data Airflow is not the engineer of record. It comes in as a third party to help owners design systems that hit their target power utilization efficiency numbers, then manages procurement of major equipment like chillers and dry coolers, and finally provides commissioning services to validate that everything works before a facility goes live.

The commissioning piece is particularly significant. Data Airflow has 50 commissioning agents who travel to facilities and spend months running around-the-clock tests: simulating loads, deliberately trying to break systems, triggering emergency transitions, and documenting that everything performs as designed. Staff rotate on two weeks on, two weeks off.

The procurement role has also become strategically important as supply chains for critical equipment have extended dramatically. Major chillers, some weighing 100,000 pounds and requiring special highway permits, are being manufactured with coils made in Belgium and assembled domestically. Guaranteed production slots with manufacturers, secured through Data Airflow’s volume and relationships, are now a meaningful competitive differentiator when a data center operator needs certainty on equipment delivery timelines.

The National Security Dimension

Ahlquist and Saunner both landed on a point that doesn’t get enough attention in the public debate about data centers: this is a national security conversation as much as an energy and environment one.

The United States is in a competition with China for AI capability that is not moving slowly and does not have an obvious finish line. China is investing heavily in the compute infrastructure that underlies AI development. American data centers represent the physical layer of the country’s AI capacity. Slowing their deployment through regulatory friction or local opposition, while China continues building, has strategic implications that go well beyond any individual community’s concerns about water or power or traffic.

“It’s an arms race of data that needs to be in the United States, not in China,” Saunner said.

That framing doesn’t dismiss legitimate local concerns. But it does suggest that communities and policymakers who are evaluating data center proposals need to weigh the national interest in AI infrastructure alongside the local impacts. The two are not unrelated: the tax revenue, job creation, and economic activity generated by major data center facilities can be substantial, and the downstream employment effect, estimated at roughly 7.4 jobs per direct data center position, compounds over time.

From Buell to the National Stage

The personal arc of this conversation is worth naming directly. Eric Saunner is a farm kid who was told by the education system that he struggled, who taught himself through YouTube videos and goal-setting, who got lost in East LA with a hand-drawn map at 19, and who built a company that is now commissioning 500-megawatt AI infrastructure facilities and competing for national Entrepreneur of the Year.

He attributes almost everything to two things: taking care of the people around him, 80% of his hires come through internal referrals, and setting goals that seem impossibly large and then building toward them anyway.

“If you got a big ship and it’s in the harbor, trying to get to Hawaii, if you don’t have a plan to get there, it’s never gonna get out of the harbor,” he said. “You may have the baddest ship and the biggest yacht in the whole world, but if you don’t have a plan, it’s never gonna get there.”

Travis Hawkes put the larger point well: the same qualities that Saunner learned picking rocks in Buell, the work ethic, the lack of complaining, the willingness to go figure something out without anyone telling him how, are the same qualities that show up in every founder who builds something worth building. Idaho keeps producing them. The challenge is making sure enough of them stay.

Learn more about Data Airflow at dataairflow.com and follow the company on LinkedIn at Data Airflow.