At Marcellus, we love to invest in businesses that dominate their industries. Companies that build and sustain deep competitive moats, thus staying a step ahead of the competition. And such moats are often found in boring, unglamorous companies. Ask an investor about the big winners in the AI boom and you’ll hear the usual names, Nvidia, the hyperscalers, memory chip makers or even transformers. But one name that we have found the most interesting is an 80-year-old air-conditioning company. This Fortune piece by Shawn Tully profiles Vertiv, which dominates the one thing every AI data centre now depends on, cooling. Since it went public in early 2021 its market value has climbed from under $11 billion to $109 billion. Its annualised return of 50.5% ranks third in the whole S&P 500, behind only Comfort Systems and Nvidia itself.
Even more interesting is the man behind Vertiv’s growing dominance. Dave Cote had just retired from Honeywell, where he’d beaten the S&P 500 by 150% over fifteen years, and he was looking for one more project. So, he cut an unusual deal with Goldman Sachs. They’d “present Cote with lots of acquisition candidates, and raise the funding once he found a choice that the investment bank also favored”. “We looked at over 1,000 companies….” he recalls. Cote knew what he was after, because he’d bought more than 100 companies running Honeywell. A dominant position in a fast-growing industry. Room to expand, at home and by acquisition. Fat margins waiting to be unlocked. And, above all, a shot at owning a new technology. As Tully puts it, Cote “didn’t chase a hot technology, he noticed a huge trend and quietly became indispensable to it.”
He’s honest that the timing was lucky. “It was not like I was an AI savant,” Cote says, “but I knew that if I positioned myself in the middle of the data industry there was a good chance something good would happen.” His read on the industry was simple. “The two biggest twentieth-century trends are biotech and the digital age,” he says, “and I thought the latter had 50 or 60 years to go.” Vertiv fit right in. Founded in 1946, it had built the first precision cooling for IBM mainframes, then drifted. Emerson Electric had dumped it to a private equity firm. “Emerson said it was a horrible business, they hated it,” says Cote. “They couldn’t wait to get out of it.” In early 2020 his group bought it for an “ultra-bargain” $4 billion through a SPAC, a structure Cote admits carried a “tawdry reputation,” and listed it on the NYSE.
What he saw that others missed was a gap. “When I got to Vertiv, pre-AI data generation was growing at 20% a year, but data centers were growing at 4%,” he says. “Eventually, the capacity of the existing centers was going to fill up, and we’d need a lot more of them….It was a tailwind no one was paying attention to.” The company itself was a classic Cote fixer-upper, with limp sales growth and operating margins of just 8% to 9%. “They could have been 25% to 30%,” he says.
Then it nearly went wrong. Three weeks after the listing, COVID hit. “The stock dropped, and Wall Street thought we were going bankrupt,” Cote says. Orders later came flooding back, but Vertiv wasn’t making money on them. “We were getting all this business from competitors,” he says, “but the reason was that we were underpricing our equipment by 20% to 30%.” By early 2023 the stock had sunk to around $13, roughly 55% below its high.
Cote got hands-on. The Board installed a shop-floor engineer named Giordano Albertazzi as CEO, and they went at pricing and costs. He also pointed the company at the right customers. “Their largest customers were the major banks, while their competitors were working with Google and Microsoft,” says Scott Davis of Melius Research. That left Vertiv “poorly positioned and losing market share.” Cote repositioned it to serve the hyperscalers. One big holder saw what it meant. “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 of Omega Advisors, Leon Cooperman’s family office. Their first investment, he adds, “was a bet on Dave.” When the stock cratered, they tripled it.
The fix bought time. The moat came from a deal. Cote pushed R&D from 3% of sales toward 6%, the same instinct that had quadrupled Honeywell’s engineers and turned half of them into software people. But the move that mattered was a small, snap-on acquisition, long a Cote specialty. From 2021 Vertiv worked with a British startup, CoolTera, on liquid cooling that runs coolant through straw-width tubes to a plate sitting right on top of the chips. Air can’t keep today’s dense racks of Nvidia GPUs cool. This can, in a fraction of the floor space. “It’s like moving from a Toyota to a racing car,” says Albertazzi. Nvidia itself helped guide the design. Vertiv bought CoolTera outright in December 2023, just as orders became, in Cote’s words, “the sunburst we’d been waiting for.” Vertiv tripled its production in three years, is now moving to triple it again by 2029, and leads rivals Schneider Electric and Eaton in the direct-to-chip market and the wider data-center cooling business, one that Grand View Research expects to grow from over $30 billion today to $128 billion by 2033.
How a 1940s manufacturer ends up making the gear of the AI age is, in Cote’s telling, gloriously old-fashioned. Get people to complain. Trust the shop floor. “Before, no one complained about anything. I want them bitching about crummy processes,” he says. “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.” It’s a lean, kaizen culture borrowed from Toyota, close to the one Larry Culp uses to revive GE. “The best ideas come from the people who actually do the job,” says Albertazzi. “That’s Dave’s mantra.”
And the obvious worry, that cheaper Chinese models like DeepSeek could choke off the orders? Cote waves it away with the oldest logic in economics. “If you take something that’s valuable to people and make it less expensive, they’ll use more of it,” he says. “If you find a way to process data more cheaply you’ll process a lot more data.” He knows a run this good invites doubt. “When something this wonderful happens you wonder how long it will last,” he told Tully, “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.” The lesson for anyone hunting the next AI winner isn’t really about chips. “AI is in large part a manufacturing business,” Tully writes, and “yesterday’s best practices still point the way forward.” ”
Note: Microsoft Corporation is part of the holdings within the Marcellus Global Compounders Portfolio, a strategy offered by the IFSC branch of Marcellus Investment Managers Private Limited and regulated by the IFSCA. Accordingly, Marcellus, its employees, their immediate relatives, and clients may maintain interests or positions in this security. Any reference to this company is intended strictly for informational and educational purposes within the context of this discussion and should not be construed as investment advice.
If you want to read our other published material, please visit https://marcellus.in/blog/
Note: The above material is neither investment research, nor financial advice. Marcellus does not seek payment for or business from this publication in any shape or form. The information provided is intended for educational purposes only. Marcellus Investment Managers is regulated by the Securities and Exchange Board of India (SEBI) and is also an FME (Non-Retail) with the International Financial Services Centres Authority (IFSCA) as a provider of Portfolio Management Services. Additionally, Marcellus is also registered with US Securities and Exchange Commission (“US SEC”) as an Investment Advisor.
