In a bid to avoid massive import tariffs on expensive AI hardware expected to be imposed by the Trump administration, Nvidia has teamed up with its partners to build AI servers in the U.S.
The company and its manufacturing partners plan to build plants for AI servers in Texas. More broadly, Nvidia and its partners aim to create an AI server supply chain that will span from chip production and packaging to servers that are ready to deploy.
Nvidia's manufacturing partners Foxconn and Wistron plan to construct two server assembly plants in Houston and Dallas. The construction is slated to start shortly, and assembly of actual machines is expected in the next 12 – 15 months. This will mark the company's first attempt to assemble its AI infrastructure entirely within the United States, aiming to strengthen manufacturing resilience and meet growing global demand. The company expects to produce $500 billion worth of AI machines over the next four years.
Nvidia plans to deploy its own technologies to optimize the design and operation of these new facilities. The company will use platforms like Omniverse to simulate factory operations and Isaac GR00T to develop automated robotics systems, bringing AI-driven improvements to every step of the production process.
Nvidia also said that TSMC had already started production of Blackwell processors at its Fab 21 near Phoenix, Arizona. However, for now silicon is shipped back to Taiwan for packaging. This should change soon as Amkor and SPIL expect to establish advanced packaging and testing operations in Arizona on over one million square feet (92,900 meters^2) of production space.
"The engines of the world's AI infrastructure are being built in the United States for the first time," said Jensen Huang, chief executive of Nvidia. "Adding American manufacturing helps us better meet the incredible and growing demand for AI chips and supercomputers, strengthens our supply chain, and boosts our resiliency."
Building AI servers in the U.S. is a big deal for Nvidia, as it would localize the AI supply chain. Considering the high price of AI servers, it does not look like more expensive labor in the U.S. will affect their cost significantly. Furthermore, as Nvidia and its partners plan to increasingly use AI and robots at the plants in Dallas and Huston, expect the actual costs of assembled systems to lessen with further optimizations.
