The company is starting production of Blackwell chips and is building factories in Texas with Foxconn and Wistron to enhance its AI infrastructure from U.S. soil.
NVIDIA has announced a historic step in its production strategy: for the first time, its Artificial Intelligence (AI) supercomputers will be designed and manufactured entirely in the United States. The company is collaborating with industrial partners such as TSMC, Foxconn, Wistron, Amkor, and SPIL, and has allocated over 92,000 square meters for new assembly, testing, and packaging facilities in Arizona and Texas.
The next-generation Blackwell chips have already begun production at TSMC’s facilities in Phoenix. Meanwhile, NVIDIA is constructing factories to assemble its supercomputers in Houston and Dallas, with the collaboration of Foxconn and Wistron, respectively. Mass production at these plants is expected to ramp up in the next 12 to 15 months.
A “made in USA” supply chain for an AI future
Jensen Huang, CEO and founder of NVIDIA, emphasized that “the engines of the global AI infrastructure are being built in the United States for the first time.” He explained that this move aims to respond to the explosive growth in demand for AI chips, strengthen the supply chain, and enhance the resilience of the tech ecosystem.
The company’s plan includes producing up to $500 billion in AI infrastructure within the country over the next four years, which could create hundreds of thousands of jobs and significantly contribute to the economic security of the U.S.
Smart factories with digital twins and robots
NVIDIA will employ its own digital twin technology and advanced robotics to design and operate these facilities. Key tools include NVIDIA Omniverse for modeling digital twins of the factories, and Isaac GR00T, a robotics platform that will enable the automation of production.
These new “AI factories” are not traditional data centers but infrastructures designed to carry out deep learning tasks, model training, and processing large volumes of information. Huang describes them as “a new category of data centers dedicated to AI-powered digital manufacturing.”
AI-Q and Agent Intelligence: the new digital workforce
The announcement is accompanied by the launch of NVIDIA AI-Q, a blueprint or reference model for building intelligent systems with multiple AI agents capable of reasoning, collaborating, and executing complex tasks. This framework is complemented by the Agent Intelligence Toolkit, an open-source library that allows for connecting, profiling, and optimizing teams of digital agents within an organization.
Companies like Visa are already using these tools to enhance their internal processes, such as detecting and responding to cybersecurity threats. NVIDIA’s approach with these technologies aims to transform AI agents into digital employees capable of interacting, learning, and making autonomous decisions more efficiently.
A boost to the U.S. tech ecosystem
By integrating strategic partners like Amkor and SPIL in Arizona, and involving major names like Salesforce, Atlassian, and ServiceNow, NVIDIA is reinforcing its commitment to a robust and localized supply chain. These alliances aim to accelerate AI adoption in key sectors such as finance, industry, energy, and utilities.
The commitment to producing supercomputers in the U.S. comes against a backdrop of increasing geopolitical tension and global competition for leadership in emerging technologies like AI. NVIDIA thus positions itself as one of the pillars of the U.S. technological industrial strategy for the next decade.
Interested developers can already access the AI-Q and Agent Intelligence tools on GitHub and participate in hackathons and training designed to foster the development of more advanced agent systems.
via: NVIDIA Blog