Taiwan-based Wistron opened its first U.S. manufacturing plant in Fort Worth on Tuesday, a $700 million, 324,000-acre facility that will assemble Nvidia’s GB300 Grace Blackwell Ultra Superchip and, soon, the Vera Rubin Superchip. The opening marks the most concrete step yet in Nvidia’s campaign to anchor its AI supply chain inside the United States, turning Texas into a production node for the systems that train and run the largest models.
The factory as a software product
Wistron describes the operation as “not typical manufacturing.” The plant runs on Nvidia accelerated computing and integrates the chipmaker’s Nemotron and Cosmos foundation models, Omniverse simulation libraries, and Metropolis vision tools. A digital twin of the factory optimizes design, workflows, and efficiency before physical production begins. In June, Wistron said it had already deployed Cosmos models for AI-driven quality inspection on certain lines and was building an intelligent factory platform with Nvidia. The Fort Worth site is the first full-scale deployment of that stack.
Jobs, chips, and the reindustrialization narrative
The plant has created 500 jobs so far, with a target of 1,000 by year-end. Nvidia CEO Jensen Huang framed the opening as evidence that the United States is “reindustrializing for the first time in a long time” through chip plants, packaging plants, and “AI factories.” Wistron Chairman Simon Lin said the location will become “one of the most important” in the next couple of years as AI infrastructure gets built domestically. The rhetoric is expansive; the headcount is modest for a $700 million outlay.
What to watch
Nvidia’s manufacturing partnerships now span Wistron, ABB Robotics for industrial automation, and General Motors for vehicle production lines. The Vera Rubin ramp at Fort Worth will test whether the digital-twin approach yields faster yield learning than traditional bring-up. The next signal comes when Wistron discloses throughput and whether the 1,000-job target holds without additional capital.