Wistron opens $700 million Fort Worth plant to build NVIDIA superchips in the US

Wistron opened its first U.S. manufacturing facility on July 21, 2026, a 324,000-square-foot plant in Fort Worth, Texas built to produce NVIDIA's most advanced AI superchips on American soil — part of a roughly $700 million commitment to expanding domestic AI hardware manufacturing.

What's new

The new facility, designated Wistron's D1 plant, opened with two manufacturing cells already running: one producing the NVIDIA GB300 Grace Blackwell Ultra Superchip and a second that will produce the upcoming NVIDIA Vera Rubin Superchip. NVIDIA says D1 is scaling up this year to output tens of thousands of boards per month. The plant has already created more than 500 jobs, with plans to reach 1,000 by year-end.

NVIDIA founder and CEO Jensen Huang appeared alongside Wistron Chairman Simon Lin at the opening ceremony, in front of Wistron executives, Taiwan government officials, and local Fort Worth leaders. "Manufacturing is an essential pillar for every economy and every country," Huang said, adding that "building chip plants, packaging plants, computer system plants like this, and AI factories all over the United States, has allowed the United States to really reindustrialize for the first time in a long time."

Context

The Fort Worth plant is the latest piece of NVIDIA's push, alongside partners like Wistron, Foxconn, and TSMC, to move more of the AI hardware supply chain — not just chip design, but board assembly and system integration — onto U.S. soil. NVIDIA has framed this as building "AI factories" across the country, a phrase it has used repeatedly over the past two years to describe end-to-end domestic production of the servers and superchips that power large-scale AI training and inference.

It also lands as demand for Grace Blackwell and the next-generation Vera Rubin platform continues to outstrip available supply, with cloud providers and AI labs racing to secure capacity. Standing up a dedicated volume-production line for both the current Grace Blackwell Ultra chip and the still-unreleased Vera Rubin chip signals NVIDIA is trying to get ahead of that same bottleneck for its next architecture rather than repeating the supply crunches that have accompanied past chip generations.

Why it matters

A dedicated U.S. production line for NVIDIA's superchips — reportedly backed by roughly $700 million in combined investment — is a concrete data point in the broader "reshoring" push around AI hardware, which has been driven as much by geopolitics and supply-chain risk as by economics. Every unit that moves through a Fort Worth facility rather than an overseas one shortens NVIDIA's exposure to cross-border shipping, tariffs, and export-control uncertainty for its most in-demand hardware.

It's also a signal of how central superchip assembly capacity has become to the AI buildout generally: a manufacturing plant opening now carries the same kind of front-page significance that used to be reserved for chip design announcements. As AI infrastructure spending increasingly gets measured by how many superchips can actually reach customers, not just how good the chip's specs are on paper, wins in physical production capacity are becoming as consequential to NVIDIA's roadmap as the chip architecture itself.