Eight years ago, the Taipei-based electronics manufacturer hitched its star to Nvidia. As the chipmaker’s fortunes have soared in the AI age, so too have those of the far lesser-known company that got its start as a spinoff of laptop maker Acer and now builds the servers that hold Nvidia’s chips. 

Wistron reported $70.2 billion in revenue in 2025, more than double the year before. Servers accounted for 70% of total sales. In the first half of 2026, revenue already hit $55 billion. (Wistron is No. 198 on this year’s Fortune Global 500, a rise of 298 spots from the year before; it’s the largest jump on this year’s ranking.)

Wistron’s fate illustrates how the AI wave is reaching beyond trillion-dollar chipmakers and LLM developers to lift the firms manufacturing the hardware behind the revolution. But it’s a classic corporate comeback, too; Wistron teamed up with Nvidia at the lowest point in Wistron’s history, then rode the AI boom into a new era of growth. 

“Even during dark times, you need to make yourself ready for any change in the future,” Lin says. 

Lin, 72, has been in the computer industry for decades. He joined Acer in 1979, a few years after its founding. He’s the man credited by Acer cofounder Stan Shih with coining the “smiling curve” economic model, which posits that the lowest margins are found in the middle of the supply chain—namely, in assembly and manufacturing—while higher margins accrue at either end: during design and at final retail. 

In 2001, Shih decided to spin off Acer’s manufacturing division into its own company, Wistron, and named Lin chair and CEO. Since then, Wistron has been trying to escape the bottom of the smiling curve by moving away from assembling computers toward conceiving entire AI systems. (Lin stepped down as CEO in 2022.) 

Early on, Wistron made PCs for other brands, a commoditized, low-margin product. A turning point arrived in 2007, when Apple unveiled the iPhone. Wistron was “stuck in the middle,” Lin recalls in July from Wistron’s office in a leafy suburb north of Taipei. “The PC business was no longer as strong, and smartphones were not our main focus.”

Wistron briefly dabbled in smartphones. It launched one of India’s first iPhone factories in 2017. But by 2023, it had exited the business amid low margins and following labor unrest at its Karnataka, India, plant.

The departure from smartphones, it turned out, would free Wistron to shift focus to something far more valuable. 

Nvidia has been a Wistron customer for almost a decade. Initially, it helped manufacture Nvidia’s graphics cards, the processors that power computer games; this proved to be a higher-margin business than ordinary PCs.

Then, in late 2022, OpenAI unveiled ChatGPT. Demand for Nvidia’s processors, wellsuited to train large language models, skyrocketed. Nvidia rushed to longtime partners like Wistron to build ever more powerful AI servers that stacked multiple processors in giant, refrigerator-size racks.

Lin had to find new factory space, hire new talent, and navigate new supply chains. “These aren’t strategic decisions but rather very practical challenges, day and night,” he says.

As soon as Wistron inaugurated a state-of-the-art 287,000-square-foot server plant in the northwest Taiwanese city of Zhubei last year, Nvidia booked all its capacity through 2026. Wistron immediately leased a nearby textile factory for a rapid-fire conversion to make even more AI servers. 

The speed of change has unsettled even Lin. “In the past, one server generation typically lasted about two and a half to three years. You’d spend roughly a year designing a product, then it would stay in the market for about two years,” he says. “With AI, not only are there new products every year, but each generation is a revolutionary leap.”

Outside of Taiwan and mainland China, Wistron operates facilities in Vietnam, Malaysia, Mexico, Brazil, the Philippines, and the Czech Republic. Yet its most consequential new bets may be two complexes near Fort Worth, covering 1.09 million square feet and dedicated to assembling Nvidia AI supercomputers. 

Wistron had been weighing a move to the U.S. for years but made the $761 million investment official last August, as President Donald Trump threatened a 32% tariff on imports from Taiwan.

Trump regularly accuses Taiwan of “stealing” from the U.S. chip industry, but Lin says the U.S. deficit in chip infrastructure is its own doing: “They threw it away,” he says. “Not stealing, okay?”

Wistron started from square one in Texas. “All the equipment we needed had to be shipped from Southeast Asia, because we couldn’t buy it in the U.S.,” he says. Wistron trained workers “from scratch,” he says. “It simply requires more time, but the foundation is good.”

Dodging Trump’s tariffs motivated Wistron to construct its Texas factories, which are expected to create over 800 jobs, but a U.S. footprint has other perks, too. For one, AI servers are heavy, sometimes upwards of six tons, which can be hard to transport via air or ship. 

“Even if this decision was slightly forced by tariffs, it became a reasonable one later on,” Lin says. 

Lin called himself a “somewhat lazy CEO” in an address to graduates in 2024. He grimaces when it comes up now. What he meant was that he chooses to focus on a few things rather than spreading his attention too thin. 

Lin would rather describe himself as a useful corrective to groupthink. “When the company is doing well, I act as a devil’s advocate. I try to remind the management team where we could be wrong,” he says. “When people feel discouraged or the company is going through a difficult time, I take on the role of a cheerleader. I always say, ‘Tomorrow will be better.’”

For an exec who owes his firm’s frenzied growth to AI, Lin admits he doesn’t rely on the technology all that much: He mainly uses it for research. 

“I’m not sure whether it’s so intelligent [that] it can replace me to make a decision,” he says.

This article appears in the August/September 2026 issue of Fortune with the headline “Simon Lin’s early bet made Wistron an unsung winner of the AI boom.”