Nvidia and TSMC Produce First U.S.-Made Blackwell Wafer at Arizona Facility
 |
| Photo Credits: Nvidia |
The announcement states the wafer signals that Blackwell production has reached “volume production” in the United States.
Nvidia CEO Jensen Huang visited the Phoenix site and signed a commemorative wafer alongside TSMC operations leadership. He is quoted as saying, “You’ve created something incredible, but you’ll eventually realize that you’re part of something historic.”
According to Nvidia’s blog post, the move “bolsters the U.S. supply chain and onshores the AI technology stack that will turn data into intelligence and secure America’s leadership for the AI era.”
The Phoenix plant is slated to produce advanced nodes including two-, three- and four-nanometer chips, along with A16 chips designed for AI, telecom and high-performance computing applications.
TSMC itself has announced plans to expand its U.S. investment up to US$165 billion, with several new fabrication and packaging facilities under development as the wafer milestone comes as demand for AI infrastructure continues its rapid growth.
The same day TSMC raised its full-year revenue forecast after posting record profit, a move the companies attribute in part to AI-related chip demand.
A research site referring to HPC sources called the event “Blackwell rises from Phoenix fab.”
However, the wafer milestone does not mean full-scale U.S. production of finished AI accelerators yet. Commentators caution that while wafer fabrication has landed on U.S. soil, subsequent steps – packaging, testing and system integration, still rely on global supply chains.
TSMC’s Arizona facility currently has capacity for front-end process technology, while certain packaging steps remain overseas.
The timing of this announcement aligns with U.S. industrial policy goals. It supports directives to strengthen domestic manufacturing of advanced semiconductors, and Nvidia’s statement flags alignment with efforts by the Donald Trump administration to advance U.S. technological leadership.
From a company-strategy standpoint, Nvidia has previously committed to a U.S. investment of up to US$500 billion in AI infrastructure over the next four years, with manufacturing and assembly operations in locations including Texas and Arizona in partnership with TSMC, Foxconn and others.
The wafer produced in Phoenix marks a key execution step in that plan. It is a tangible item representing the transition from announcement to factory floor production.
Observation of the wafer’s existence allows stakeholders (supply-chain partners, investors, and policy-makers) to gauge progress beyond slide decks.
At the level of global supply chains, having one of the most advanced process nodes (2nm / 3nm / 4nm) produce wafers in the United States carries implications for export controls, geopolitical risk mitigation and regional manufacturing diversification.
That said, one wafer does not transform the supply-chain topology overnight.
Scaling volume, yield, downstream assembly, logistics and global integration remain work in progress.
In the semiconductor market, timing and cost efficiency are critical. Producing on U.S. soil has higher cost bases than Taiwan, according to TSMC commentary, which means competitive dynamics must factor in productivity, yield and end-market pricing.
This development will be monitored closely by data-centre operators, cloud providers, AI hyperscalers and policymakers, with attention to lead-times, cost curves and the ability of U.S. manufacturing to meet the high-growth demand for AI hardware.
