NVIDIA and Taiwan Semiconductor Manufacturing Co. (TSMC) have begun producing advanced Blackwell AI computing wafers in the United States, marking a significant step toward bolstering American technological independence and reindustrializing its semiconductor sector.
The inaugural NVIDIA Blackwell wafer was manufactured at TSMC’s new facility in Phoenix, Arizona. This event signifies that the Blackwell platform has entered volume production.
Jensen Huang, founder and CEO of NVIDIA, visited the Phoenix plant for a ceremony where he signed the wafer alongside Y.L. Wang, TSMC’s Vice President of Operations. The signing symbolized this key milestone.
“This is a historic moment for several reasons,” Huang stated. “It is the first time in recent U.S. history that the most important chip is manufactured here in the United States by the most advanced factory, by TSMC.”
Ray Chuang, CEO of TSMC Arizona, emphasized the rapid progress. He noted, “From arriving in Arizona to delivering the first U.S.-made NVIDIA Blackwell chip in just a few years represents the best of TSMC.” This achievement also underscores a three-decade partnership between the two companies.
TSMC Arizona is set to produce advanced technologies, including two, three, and four-nanometer nodes. It will also manufacture A16 chips, which are crucial for various industries. These technologies are vital for high-performance computing, telecommunications, and AI applications.
High-performance Graphics Processing Units (GPUs) are essential for training and running artificial intelligence models. Producing these critical components within the U.S. carries significant economic and national security implications.
The manufacturing process for a wafer is complex. After production, it undergoes stages such as stacking, patterning, etching, and thinning before becoming a high-performance AI chip ready for data centers and AI applications.
Locating manufacturing closer to research and development centers and customers helps reduce supply chain risks and delivery times. It also facilitates collaboration between chip designers and final assembly operations.
This domestic production aims to nationalize the AI technology stack. The companies state this will help convert data into intelligence and secure U.S. leadership in the AI era, creating jobs and revitalizing industrial output.
Beyond the technical aspects, the announcement holds economic and geopolitical weight. It confirms TSMC Arizona’s capability for advanced node production, which demands precise processes and strict quality control.
Economically, it supports local job creation and the development of a domestic microelectronics industrial ecosystem. Geopolitically, manufacturing the “most important” chip in the U.S. addresses discussions around technological autonomy and supply chain resilience.
NVIDIA also plans to utilize its AI, robotics, and digital twin technologies to design and operate new manufacturing facilities across the U.S. The company links these efforts to optimizing plant operations and accelerating industrial performance.
Further information on these technologies will be showcased at NVIDIA’s GTC Washington D.C. event. The conference is scheduled from October 27 to 29.