Wistron chief dismisses AI bubble fears as orders soar through 2027

The head of a major Taiwanese electronics company says artificial intelligence is here to stay and will keep growing through 2026 and beyond, pushing back against fears that the technology sector may be overheating.

Simon Lin runs Wistron, a company that makes components for Nvidia, the chip giant at the center of the AI rush. Speaking to reporters in Taipei on Friday, Lin said he believes the technology will change how every business operates. He called it the start of a new era rather than a temporary excitement that will fade away.

Production set to begin at American plants

Lin said his company expects to see bigger growth in AI-related orders this year compared to what they saw in 2025. Business looks strong all the way into 2027, he added. When asked about this year specifically, he described the expected growth as major.

The company’s new manufacturing plants in the United States are on track to open this year, according to plans announced previously. Jeff Lin, who serves as CEO of Wistron, said actual production at these American facilities will begin during the first six months of 2026.

Some of the space at these plants will support a massive project by Nvidia to manufacture AI servers on American soil. The chip company aims to build up to $500 billion worth of these specialized computers in the US over the next four years. Last April, Nvidia revealed plans to construct supercomputer factories in Texas, working with Foxconn in Houston and Wistron in Dallas.

Recent industry numbers back up this positive outlook. The worldwide semiconductor business got close to being worth $1 trillion in early 2026. Reports from the first week of February showed that computer chips used for logic operations and memory storage both grew by more than 30% compared to the same time last year.

On February 5, 2026, Foxconn, which works closely with Wistron in the region, announced its January earnings hit NT$730.04 billion. That represents a jump of 35.5% from the year before. The company said strong customer interest in AI server equipment drove most of this increase.

Next-generation chips enter mass production

The Texas production ramp-up comes as Nvidia moves to its newest chip design. In January 2026, the company said its “Rubin” platform had started full-scale manufacturing. This new system, which replaces the older Blackwell design, includes two main parts: the Vera processor and the Rubin graphics chip.

Engineers built it specifically for what they call “agentic AI,” and the company expects to ship large quantities starting in the second half of this year.

Making these chips requires advanced manufacturing methods. The new designs use a 3-nanometer production process, which puts extra demands on companies like Wistron to speed up their ability to assemble these products inside the United States.

On February 6, 2026, Tower Semiconductor said it would team up with Nvidia to create 1.6T silicon photonics technology. This system aims to solve connection problems in large groups of graphics processors used in AI data centers.

Around the same time, reports emerged that the US Department of Energy locked in $1 billion in funding to build two new supercomputers using cutting-edge AI hardware. This adds to the $500 billion in total orders Nvidia has reported for its current and upcoming chip designs.

The Dallas location fits into broader efforts to bring high-tech manufacturing back to American soil to make supply chains more reliable. With production starting this semester, the facility will handle building “AI factories“, special data centers made for training large AI models.

Experts note that customer needs are changing. Rather than just training brand-new AI systems, companies now need constant computing power to run AI applications, which they call “inference” work. This requires the kind of massive server infrastructure that Wistron and Foxconn are building across America.

As of early February, orders for these powerful computing systems stretch all the way through 2027, suggesting the current demand stems from real infrastructure needs rather than market speculation.