Meet the AI Centurions.

While reporting a story on the extraordinary comeback at GE led by CEO Larry Culp, I was struck by the high-voltage stock performance of GE Vernova. What from the outside looks like the ultimate old-economy stalwart was once a pillar of the world’s most famous conglomerate. But since it was spun off from GE in early 2024, it has surged from a starting market capitalization of $39 billion to $288 billion on July 21. In large part, the AI explosion powered the moonshot. GE Vernova is not alone. A whole cadre of old-line companies has mined the AI boom to multiply their valuations many times over and sprint into the $100 billion-plus club.

As a starting point, I went back three years to mid-July 2023, the approximate point when hyperscalers began planning major expansion in AI data centers, or later if the company was spun off or went public more recently. The two criteria were reaching the $100 billion market-cap benchmark after beginning the race far back, and achieving annualized returns of at least 100%. The search revealed five especially notable examples in addition to GE Vernova. Four are effectively old-timers that, like GE Vernova, sell equipment to hyperscalers: Seagate Technology, Vertiv, Western Digital, and Western’s recent spinoff, Sandisk. In the decade before the AI liftoff, they all sat alongside the GE power unit in the underachiever category, with tepid growth and weak profitability. Seagate and Western, the two that were publicly traded before the AI phenomenon, had fairly flat share prices, and the original Vertiv franchise was cut loose by its parent after chronically poor results.

Two of the selections are measured over a period shorter than three years. GE Vernova split from GE in April 2024, and Sandisk separated from Western Digital in February 2025.

The sextet suddenly reset onto a high-powered trajectory by developing products that helped AI data centers reach new frontiers in storage, cooling, and power. Seagate, Western Digital, and Sandisk benefited from rising demand for storage. Vertiv rode demand for cooling systems. GE Vernova gained from soaring demand for power and its giant turbines. The sixth qualifier is the only one that does not sell data-center gear. AppLovin is an AdTech platform that benefits from AI-driven gains in processing capacity. That data-crunching power, driven by Nvidia GPUs, helps AppLovin occupy a leading position in marketing and monetizing mobile apps and games.

GE Vernova started the contest with by far the highest market cap, just under $40 billion. It was also measured over a shorter period, since it has been publicly traded for only two years and four months. The others stood at just $7 billion to $14 billion and now boast valuations that, as of the market close on July 21, ranged from $114 billion for Vertiv to $288 billion for GE Vernova. At the median, their stocks advanced at a 145% annual pace over the past three years. During that period, their combined valuation jumped from $90 billion to almost exactly $1 trillion. The more than $900 billion increase accounts for roughly 3.5% of the rise in the S&P 500’s market capitalization since mid-July 2023. The six also posted share gains of more than 100% a year, more than quadruple the roughly 30% for the Magnificent Seven.

In terms of metrics, “100” is a common theme for these six names, since they all have valuations above $100 billion and have exceeded 100% annual share gains. As recently as spring 2024, none of these future stars would have made even the top 300 list of America’s most valuable companies. Seagate is officially headquartered in Ireland, but since it runs operations — and its CEO works — from Fremont, California, I included it among the six.

There is another “100” connection. Now, all but Vertiv would rank in the top 100 of U.S. companies by valuation as of July 21, and Vertiv stood on the cusp at 101st. GE Vernova, Western Digital, Seagate, and Sandisk are all currently worth more than Blackstone, Salesforce, Pfizer, and Uber.

To label the group, I chose the term “centurion,” rooted in centum for 100 in Latin. Centurion is used to describe Roman army officers who command 100 soldiers, cricket players who score 100 runs, and Navy pilots who notch 100 night landings. Hence, I’ll give these six beneficiaries of the 21st century’s signature technological leap the title of “The AI Centurions.”

These sturdy legionnaires are victorious for now, but they face uncertain prospects. They are highly dependent on a continued surge in spending on AI infrastructure. That total has risen from $35 billion in 2023 to a run-rate of $650 billion this year, and is expected to reach $1.0 trillion to $1.1 trillion in 2027. A reduction in those outlays, or even a slowdown in their pace, could send their shares sharply lower — especially since their P/E multiples, ranging from 29 for GE Vernova to 88 for Seagate, suggest the market expects strong profit growth from current levels to keep the stocks climbing.

Here is a look at how each of the Centurions has developed and exploited new technologies to ride the AI wave. (“Valuation gain” and “annualized stock gain” indicate the rise from mid-July 2023 to July 21, 2026, or shorter periods for GE Vernova and Sandisk because of their later public debuts.)

Vertiv (valuation gain: $10 billion to $114 billion; annualized stock gain: about 120%)

Vertiv is the retirement project for Dave Cote, the GE veteran who ran Honeywell from 2002 to 2017 and generated returns 2.5 times the trajectory of the S&P 500. Vertiv’s precursor was a division of Emerson Electric that developed the first precision cooling system for IBM mainframes. “Emerson hated the business and hated the industry,” Cote recalls, and sold it to a private equity firm. In partnership with Goldman Sachs, Cote bought Vertiv in 2020 for $4 billion.

“I was no AI savant,” says Cote, a long-time expert in manufacturing everything from appliances at GE to turbochargers at Honeywell, who now serves as Vertiv’s executive chairman. “But I knew if I positioned myself at the center of this rise in digital data, something good would happen.” Cote and his team connected with Nvidia and began work on a new cooling system for data-center platforms using Nvidia’s GPU chips. “Nvidia helped us understand what their customers, the hyperscalers, needed, and what would be needed for the technology,” Cote told Fortune.

To meet the huge new processing demands, data centers now deploy far greater rack, GPU, and other hardware density than in the past. If chips run too hot, their performance weakens and their lifespan shrinks. With Nvidia’s help, Vertiv developed a system that, instead of traditional air cooling, directs a water-and-glycol coolant directly over the chips to absorb and remove heat. The process keeps processors at ideal temperatures.

Vertiv’s customers are Amazon and Alphabet, which fill data centers with chips from Nvidia and other suppliers, as well as colocation companies such as Equinix that own facilities and provide power, hardware, and cooling to hyperscalers that rent from them. Vertiv is posting spectacular numbers that mark a sharp break from its sleepy past. In 2025, it grew sales 31% to $10.2 billion and operating profit sixfold to $2.1 billion. Cote thinks the good times will continue. “My career tells me that when something this wonderful happens, you wonder how long it will last,” he says. “But I’m convinced that this AI data center expansion is essential to the digital age, and will go on for a long time.”

Seagate Technology (valuation gain: $14 billion to $181 billion; annualized stock gain: about 130%)

Beyond advanced cooling, data centers need a huge increase in storage capacity to run applications as varied as large language models and roadway and traffic data used to guide driverless vehicles. Formerly struggling Seagate is enjoying a renaissance by offering solutions designed for that demand. Founded in 1978, Seagate first thrived by supplying hard disk drives, or HDDs, for early IBM PCs, but it made a wrong turn during COVID by building excess inventories. When post-pandemic orders fell far short of the volumes it anticipated, Seagate swung from strong profits in fiscal 2019 to a loss in fiscal 2023. By then, the long-time tech mainstay faced fading fortunes as a provider of price-competitive legacy hardware.

But as Vertiv did in thermal management, Seagate found a breakthrough tailored to the AI surge. In early 2024, it launched a new architecture on its Mozaic platform called HAMR that multiplies capacity on standard 3.5-inch disks. Using lasers to heat the disks during writing allows for super-dense data packing. The breakthrough overcame traditional physical recording limits. In the process, Seagate made parking an exabyte far cheaper and saved substantial data-center space.

A jump in revenue and profitability ignited the share price. In its third quarter ended April 3, Seagate shipped 199 exabytes of memory to data centers, a 39% increase from the same quarter a year earlier. Fueled by strong pricing, revenue rose even faster, up 44%, and the company has established a trend of collecting more profit per dollar of sales in each successive quarter. In Q3, it posted gross margins of 47%, an 11-point increase year over year.

Western Digital (valuation gain: $10 billion to $168 billion; annualized stock gain: about 160%)

Seagate and Western Digital are fierce competitors that dominate the global hard-drive memory market in a virtual duopoly, together commanding an estimated 80% of the business. They compete across enterprise solutions, cloud storage, and portable consumer devices, and differ mainly in product design: Seagate emphasizes maximum storage capacity, while Western offers a wider variety of specialized uses.

As the AI boom accelerated, Western re-engineered its hard drives to handle the massive new workloads at data centers. Like Seagate, it has also profited from a seismic shift. Before the AI takeoff, prices were declining by around 10% a year per terabyte. Now, orders are far outpacing the two rivals’ ability to supply them. Western is sold out through 2026 and is locking in long-term contracts starting in 2027 and 2028. The supply squeeze has lifted prices of its offerings by as much as 50% since the start of last year.

That combination of fast-rising volumes and stronger pricing power lifted Western from a $1.7 billion loss in fiscal 2023 to $6.4 billion in profit over the past four quarters. In the March quarter, gross margins reached 51%.

Sandisk (valuation gain: $8 billion to $235 billion; annualized stock gain: 1,600%)

In February 2025, Western Digital spun off its solid-state drive business into Sandisk, a fully separate publicly traded company. AI campuses typically use both the HDDs that are Western Digital’s specialty and the SSDs that Sandisk supplies for different purposes. HDDs are used to store “warm” and “cold” data that require massive capacity and low cost per terabyte, such as long-term archives. SSDs provide “flash,” or “hot” storage, using NAND technology that offers lightning-fast retrieval for real-time user profiles and active databases.

The spin proved to be a brilliant move. Although HDD demand is rising quickly, SSD sales are growing even faster. Data centers are dealing with enormous inference and training workloads that require vast flash memory, and they cannot secure enough storage quickly enough to remove bottlenecks. Supply shortages are pushing prices higher and speeding Sandisk’s ascent. It is also looking ahead through the development of High Bandwidth Flash, or HBF, a technology for much denser NAND storage that it created in partnership with SK Hynix of South Korea. The consortium expects to launch HBF in the second half of 2026.

Sandisk has posted extraordinary results. In the nine months ended April 3, revenue jumped to $11.3 billion, up almost 110% from the comparable period spanning 2024 and 2025, while earnings swung from a large loss to a gain of $5.4 billion. Incredibly, while its former parent is up a vertiginous 17-fold over three years, Sandisk has risen ten times as fast, at 170-fold in just 18 months as an independent company, and as of July 19 its valuation was more than $60 billion higher.

GE Vernova (valuation gain: $39 billion to $288 billion; annualized stock gain: 120%)

By taking the CEO chair in 2018, Larry Culp accepted what was likely the biggest rescue job in U.S. corporate history. The conglomerate’s problem child was power. The business had built and acquired vast new capacity to serve what leaders expected to be a surge in global demand that never materialized, while also underestimating the market share soon taken by renewables. Culp spent his first year mainly trying to fix the power business. His move to kaizen-style lean manufacturing and his decision to split a highly centralized management structure into roughly eight profit centers, each responsible for its own P&L, put the business on a stronger recovery path. Culp also identified promising young manager Scott Strazik, now 48, as a rising star and chose him to run the newly independent business when it became GE Vernova in April 2024.

The revival in global energy demand to levels well above GDP has been a powerful tailwind. But AI has provided the decisive push. The rapid spread of data centers is straining utilities’ capacity to deliver enough power. As a result, hyperscalers and colocation providers are installing both large and smaller aeroderivative natural gas turbines on-site to keep server racks running and liquid moving through cooling pipes. The biggest units are about 30 feet long and cost roughly $250 million each, and GE Vernova has top share in both the super-sized machines and the aeroderivatives used by hyperscalers.

AI data centers typically use two to eight of the behemoths, or 40 to 60 aeroderivative models. GE Vernova turbines are now powering xAI’s Stargate project in Texas and OpenAI’s Colossus One campus in Tennessee. A venture combining Chevron and American Engine No. 1, an investment firm that funds projects driving American re-industrialization, recently bought seven large turbines to lease to Microsoft for a massive facility in Texas. Together, those machines provide enough electricity to power three million homes.

Last year, about 10% of GE Vernova’s $59 billion in orders came from data centers. For 2026, it is projecting one-fourth from AI on a much larger total. Orders are overwhelming production, a common theme among the Centurions, and that is helping push prices higher. It is also important to note that although AI is providing the biggest boost to growth, GE Vernova still receives 80% of its orders elsewhere. It does strong business with private and state-owned utilities around the world that are using its electrification segment to upgrade outdated power grids. The data-center boom is only one part of the growing global appetite for gigawatts, which is forcing the world’s electricity generators both to provide more power and to move it faster. Both trends are benefiting GE Vernova.

In its recently released first-quarter report, GE Vernova guided to 2026 revenue of $46 billion at the midpoint, a 21% increase from 2025, and projected $12 billion in free cash flow, more than triple the prior level. Simply put, AI has transformed this old-line franchise.

AppLovin (valuation gain: $10 billion to $143 billion; annualized stock gain: about 130%)

AppLovin does not supply AI infrastructure like its five fellow Centurions, but its expansion is driven by a new product powered by the immense computing capacity flowing from the AI boom. The company is now among the world’s largest pure-play AdTech providers. AppLovin was founded in 2012 by a team led by Adam Foroughi, whose family fled Iran during the late-1970s revolution and settled in California. Foroughi has said his father, a major real-estate developer in Iran, was likely a billionaire who lost everything when the family fled the overthrow of the mullahs in their homeland. The CEO says watching what his parents lost fueled his determination to succeed in his adopted country. For example, he insists on signing off on every new hire himself.

AppLovin helps app developers grow their audiences in social media and gaming, and it broke new ground by placing full-screen video and interactive ads inside popular mobile apps and games. Its rapid ascent began in 2023 with the launch of Axon, an AI “brain” that analyzes billions of consumer data points to target users most likely to buy products or engage with an app. Axon parses buying and travel patterns to identify the best candidates, for example, for Clash Royale or Block Blast. Customers value the platform because it is skilled at calculating the precise return on ad spending. That data allows AppLovin to run dynamically priced auctions in which advertisers compete on price to win customers.

AppLovin depends on AI data centers to accelerate Axon’s constant learning and to meet its growing capacity needs as it processes rising volumes of behavioral data on potential customers. That ability to scale quickly at relatively low cost is fueling one of the fastest ramps in sales and profitability among the Centurions. Last year, AppLovin lifted revenue 70% to $5.5 billion, and profits 111% to $3.3 billion.

The Centurions stand in the shadow of the hyperscalers as overlooked heroes of the AI upheaval. Put simply, they are selling the picks and shovels essential to the entire gold rush. Their future will depend on the big unknown: whether the hyperscalers will find the gold in large language models and enterprise solutions and keep buying the tools these companies have proven so resourceful in providing.

This story was originally featured on Fortune.com