Now, NVIDIA Corporation has announced new factory plans to build higher-level AI chips and supercomputing systems over here. It involves partnerships with some of the biggest names in the field and plans to build manufacturing facilities in Arizona and Texas. And that strategic move reflects the company’s desire to bolster its national supply chain for critical AI infrastructure.
Arizona to be the Site
However, Arizona will play a significant role in NVIDIA’s updated manufacturing roadmap. Additionally, we now know that the Taiwan Semiconductor Manufacturing Company (TSMC) is making NVIDIA Blackwell family of AI chips at its fab in Phoenix. NVIDIA is also working with Amkor and SPIL to provide three levels of chip testing and packaging, which will also take place in Arizona. If one state is a model, the focus of this multi-stage process has been on efficiency and cutting out points of friction.
Texas Features in the AI Supercomputer Mission
NVIDIA, for its part, already has plans to construct its world-class AI supercomputers in Texas. NVIDIA announced that it will manufacture its new DGX Station and DGX Spark in this state. Foxconn, which can be established in Houston to manufacture in and between, and Wistron should be established in Dallas. Over the next 12 to 15 months, NVIDIA will gradually ramp production at both Texas sites, which amounts to a sizable boost in domestic supercomputing.
TSMC and broader US semiconductor manufacturing
Moving beyond its Mt. Olympus manufacturing volume growth strategy, TSMC has made and is continuing moves to grow its manufacture volume in the US, in part to counter earlier Chinese price competition in chips, which may well include its newest Blackwell chips — directly related to the fact TSMC will be involved with their manufacture in Arizona. The semiconductor maker, meanwhile, is riding the wave of considerable support from the CHIPS Act of this administration — $6.6 billion, to be exact. Additionally, in March of this year, TSMC committed to a record $100 billion investment in the US. The expansion will enhance TSMC’s capacity to fulfill the domestic high-end semiconductor manufacturing demand from US technology companies, including Apple and Qualcomm, the sources said.
Nonetheless, even in itself, this turn to local manufacturing is already a strategic play by NVIDIA at this time of international trade wars and making chips abroad. While some duties on laptop components and chips were “delayed, building US manufacturing is the only proper, solid solution to head off disruptions that global trade policies could create.” As NVIDIA seeks to maintain its growth and innovation in the rapidly changing AI industry, it also aims to bring stability and resilience to its supply chain by planning to localize production for its key AI technologies.
Some key Insights
Blackwell architecture: NVIDIA’s next-generation compute GPU architecture—Designed for the fastest possible compute acceleration for the most demanding AI and high-performance systems. DGX Spark and DGX Station are AI supercomputers that enable dedicated use for data science, machine learning, and another AI research. Taiwan Semiconductor Manufacturing Company (TSMC) is the world’s largest independent semiconductor foundry, a critical node in the global technology supply chain
