If asked to name the top-performing stocks driven by the AI boom, most investors would probably say Nvidia (correct), perhaps Meta (incorrect), and a handful of hyperscalers (also incorrect). Few would guess Vertiv, an 80-year-old air-conditioning maker. Yet since Vertiv went public in early 2021, its market capitalization has risen nearly 10-fold, from under $11 billion to $109 billion. Over that six-and-a-half-year span, its annualized return of 50.5% ranks third in the S&P 500, behind only Comfort Systems at 55.2% and Nvidia at 54.9%.

The man behind that unlikely transformation is retired industrial executive David Cote, who almost did not take the job. In 2019, after stepping down from a long career leading Honeywell, where he had overseen a major revival after GE’s failed acquisition left the industrial company directionless, Cote decided he still wanted another challenge. He arranged an unusual partnership with Goldman Sachs: the bank would bring him acquisition candidates and secure financing once he selected a target that Goldman also supported.

Cote had a long record of finding undervalued businesses. At Honeywell, he acquired more than 100 companies, using deals to accelerate growth and improve profitability. He was looking for a company that fit the same pattern: a business with a dominant or potentially dominant position in a fast-growing industry, substantial room to expand both organically and through acquisitions, widening margins, and, most importantly, the chance to build a sector-leading new technology.

“I had weekly phone calls with the Goldman group, and we looked at over 1,000 companies, and it was Goldman that found them,” Cote recalls. That company was Vertiv, an 80-year-old industrial AC provider that under his leadership became a supplier of technology central to the expansion of AI. Data centers running Nvidia GPUs and other intensive computing equipment require far more cooling than traditional air systems can provide. Vertiv’s direct-to-chip liquid cooling systems, he says, are helping meet that need.

“It was not like I was an AI savant, but I knew that if I positioned myself in the middle of the data industry there was a good chance something good would happen,” Cote says. His path stands apart from many executives who have spent years trying to bolt an AI strategy onto an existing business. Cote did not chase AI directly; he identified a major trend and moved to become indispensable to it.

Cote wanted a place at the center of the digital age, but Vertiv got off to a rocky start

Cote’s informal, down-home style turns out to fit the more freewheeling tone of the AI era. The hip-hop-loving, Harley-Davidson-riding executive has long seemed closer to Silicon Valley than to traditional corporate America. In a 2012 Fortune profile on his Honeywell turnaround, Cote, pronounced Co-tee, described growing up in a small New Hampshire town where his father, who had an eighth-grade education, ran a garage. “I didn’t know what success was because it was hard to find anyone in town you’d describe as successful,” he said.

As a college student, Cote once took a year off to buy a 33-foot lobster boat with a friend and trawl for cod in Maine. That experience convinced him that “You can work very hard and accomplish absolutely nothing.” He began his career as a night-shift worker at a GE aircraft engine plant in New Hampshire, then moved up through the dishwasher and silicone businesses before eventually leading the appliance franchise.

His 25-year GE career ended abruptly when Jack Welch fired him in 1999. Cote says Welch approached him in the headquarters dining room in Fairfield, Connecticut, and immediately told him, “Dave, I want you out of the company by year end!” Cote asked what he had done wrong, and Welch repeated the demand more loudly. Cote says he kept asking for an explanation, but Welch only repeated the same words, each time louder than before.

In 2002, Cote took over Honeywell, a conglomerate whose businesses included avionics, specialty chemicals, and auto parts ranging from turbochargers to spark plugs. Over the next 15 years, Honeywell delivered shareholder returns that beat the S&P 500 by 150%. Cote often dressed informally, wearing a beat-up bomber jacket, baggy jeans, and work boots to work and even to board meetings. His office included a 210-gallon fish tank and a nonstop soundtrack from an iTunes library of 10,000 songs. “The music never stops!” he would say.

Every couple of months, Cote would spend a full day alone in his office, without calls or meetings, to think through a major strategic issue. With Billie Holiday’s “God Bless the Child” or Jay-Z’s “Hard Knock Life” playing in the background, he would work through decisions such as how to design a new generation of turbochargers.

Cote had never heard of Vertiv when Goldman presented it. Founded in 1946, Vertiv’s predecessor became the “network power” business of Emerson Electric and developed the first precision cooling system for IBM mainframes. By the time Cote found it, Emerson had sold Vertiv to California private equity firm Platinum Equity. “Emerson said it was a horrible business, they hated it,” Cote says. “They couldn’t wait to get out of it.” Under Platinum, Vertiv was struggling in its traditional business, providing air cooling for standard data-center functions such as primary storage, business software, and cloud computing for website hosting.

In early 2020, Cote and an investor group assembled by Goldman bought Vertiv for $4 billion through a special-purpose acquisition company and simultaneously took it public on the NYSE. Cote acknowledges that SPACs already carried a “tawdry reputation” at the time and have since largely disappeared. But this one worked: Vertiv became one of the few SPACs to launch a durable enterprise.

Cote says he was drawn to Vertiv for a straightforward reason. It had a meaningful position in what he saw as one of the great industries of the future: digital data. With the right products, he believed it could ride that growth. “The two biggest twentieth-century trends are biotech and the digital age, and I thought the latter had 50 or 60 years to go,” he says.

Vertiv was also lagging because it had focused on the wrong customers. “Their largest customers were the major banks, while their competitors were working with Google and Microsoft,” says Scott Davis, an analyst at Melius Research. “That’s why they were poorly positioned and losing market share.” One of Cote’s first moves was to reposition the company toward hyperscalers.

At the time, the AI boom was still several years away. But Cote believed the industry was entering a major expansion anyway. “When I got to Vertiv, pre-AI data generation was growing at 20% a year, but data centers were growing at 4%,” he says. “It didn’t make sense to me when all that data is being created and increasing everywhere. Machine learning, the predecessor to AI, was already requiring far more data. Eventually, the capacity of the existing centers was going to fill up, and we’d need a lot more of them. Data center growth has to approximate digital data growth. It was a tailwind no one was paying attention to.”

He also saw parallels with Honeywell. Vertiv had weak sales growth and operating margins of just 8% to 9%. Cote says those margins could have been 25% to 30%. For an experienced operator, he saw substantial room for improvement.

At first, the company appeared to be heading toward disaster rather than success. Three weeks after Vertiv’s NYSE debut, COVID hit. “The stock dropped, and Wall Street thought we were going bankrupt,” Cote says. But business rebounded quickly as the stay-at-home economy drove heavier use of cellphones and PCs. Even so, Vertiv did not immediately benefit. “We were getting all this business from competitors, but the reason was that we were underpricing our equipment by 20% to 30%,” Cote says. Profits fell sharply. By early 2023, the stock was around $13, roughly 55% below its high of nearly $29 from 18 months earlier.

To fix the problem, Cote became deeply involved in day-to-day management. In January 2023, the board replaced the CEO with Giordano Albertazzi, a mechanical engineer trained in Milan and at Stanford, and a veteran of Emerson who had risen from plant manager. Cote valued that kind of shop-floor experience, believing that the best solutions come from people who understand how products are actually made.

“It’s one thing to have the wind at your back as Dave did in the first couple of years. But companies almost always hit a big speed bump early on, and the really good ones are those that can fix it, and that’s what Dave did,” says Ethan Brown, portfolio manager for Omega Advisors, the family office of investor Leon Cooperman and a top 25 holder of Vertiv stock with a position exceeding $500 million. Brown adds that his firm’s original 2020 investment was a bet on Cote, and that after the stock fell, he and Cooperman tripled their position.

Vertiv adopted an all-new cooling technology from the outside—via acquisition

Even during difficult periods, Cote and Albertazzi were building the strategy that would transform Vertiv: new products. Cote had spent years at Honeywell expanding the role of software and engineering, and he wanted to repeat that model. When he arrived at Honeywell, 20% of engineers worked in software; by the time he left, the engineering workforce had grown fourfold and 50% were focused on software.

At the time of Vertiv’s public offering, the company was spending only 3% of sales on R&D. Cote pushed that figure toward 6%, where it stands today on much higher revenue. But the most important move came through acquisition, another of Cote’s specialties. A small buyout would provide a new technology that proved decisive.

Around 2021, Vertiv began working with a British startup called CoolTera on a new process known as liquid direct-to-chip, or DTC, cooling. Around the same time, Vertiv also began discussions with Nvidia about the cooling technology hyperscale customers would need to deploy Nvidia GPUs in new data centers. “Nvidia helped us understand the technology that would be needed,” Cote says. Albertazzi adds, “Nvidia knew about the CoolTera products through us. It was clear that Nvidia’s comfort level with their products was quite strong.”

The technology is complicated to build but simple to understand. Air cooling is not enough to keep dense racks of GPUs from overheating, which can sharply reduce efficiency. DTC uses a water and glycol mixture that flows through tubes roughly the width of straws into a cold plate placed directly above the semiconductors. The liquid absorbs heat and transfers it through a heat-exchange process to a cooling tower system on the roof or outside the building, similar to a larger air-conditioning condenser. DTC keeps GPUs between 65 and 75 degrees. It does not use additional water; the same liquid circulates continuously. Vertiv says the approach reduces the space required for equivalent computing by 50% to 70%. As Albertazzi puts it, “It’s like moving from a Toyota to a racing car.”

In December 2023, Vertiv bought CoolTera. Even before the acquisition, Cote and Albertazzi were betting heavily on DTC and planned to use the CoolTera system. “Starting in 2023, orders were really starting to take off, and I thought it would continue,” Cote says. “It was the sunburst we’d been waiting for.”

In early 2023, Vertiv launched a plan to triple production over three years, largely to support new DTC products. The company added production lines and extra shifts at several plants, then built a 215,000-square-foot greenfield facility in South Carolina that opened in October 2024. This year, Vertiv reached its production goal and has now started phase two, aiming to triple capacity again by 2029.

Vertiv effectively commercialized CoolTera’s DTC product, which had been little more than a prototype before the acquisition, and did so almost from scratch. “It was the beginning of the re-architecture of data center infrastructure to accelerate higher and higher density compute,” Albertazzi says. Vertiv now competes with Schneider Electric and Eaton in DTC and reportedly holds the largest market share in that technology as well as the broader data-center cooling market, which is worth more than $30 billion and is projected by Grand View Research to reach $128 billion by 2033, a 22% annual growth rate.

Cote’s flexibility impressed Brown. “Dave switched from fixing operations to seizing on an unbelievable growth opportunity,” Brown says. “He saw it clearly early on, and made sure Vertiv invested in the supply chain, distribution, production and R&D to take full advantage of probably the most important secular change we’ll see in my lifetime.”

For Vertiv’s leaders, the best way to make the products powering AI is the old-fashioned way: by nurturing a culture of listening

Cote and Albertazzi have also worked to build a distinct culture that they say served them well in manufacturing and now helps in AI-related markets. Their approach has two main elements: encourage people at every level to speak openly about problems without fear of retaliation, and rely on shop-floor workers to identify the fastest, most reliable, and highest-quality way to make products.

“Before, no one complained about anything. I want them bitching about crummy processes,” Cote says. “The sales people, the hourly people on the floor. How do I make sure I have better processes? By having a thinking company. We have 20,000 people. If they all think every day about making the company better, and not just doing what they’re told, the better you’ll be.”

The company also uses a lean operating model called the Vertiv Production System, modeled on the system Cote used at Honeywell and inspired by Toyota’s kaizen, or continuous improvement, principles. It is similar to the approach Larry Culp used to revive GE and now applies at GE Aerospace. The goal is to get managers and engineers onto the assembly lines, where they work with mechanics and machine operators to design better workflows.

“The idea is that the best ideas come from the people who actually do the job,” Albertazzi says. “That’s Dave’s mantra, and I’m a firm believer.”

Cote says he is not especially worried about Chinese AI models such as DeepSeek, which may be cheaper to use than U.S. alternatives and could affect what hyperscalers earn from enterprise token processing. He argues that lower costs often increase usage. “It’s all about how much you’re billing customers for the AI they’re using,” he says. “Go back 24 months to when China announced DeepSeek. It was supposed to undermine the hyperscalers. But if you take something that’s valuable to people and make it less expensive, they’ll use more of it. If you find a way to process data more cheaply you’ll process a lot more data.”

Looking back on the company’s transformation, Cote says he can hardly believe what has happened. “When something this wonderful happens you wonder how long it will last. But I did my research, and I found that what we do is fundamental to the digital age, and will go on for a long time,” he says.

From GPUs to memory disk drives to Vertiv’s DTC equipment, AI is in large part a manufacturing business. In Vertiv’s case, Cote’s story suggests that older industrial disciplines still matter in the new era.