Nvidia will move a part of chips manufacturing and produce AI supercomputers for the first time in the United States. A strategic movement influenced by Trump’s chaotic tariffs and will display in the next four years with an economical amount calculated at 500,000 million dollars.
“The world infrastructure engines will be built in the United States for the first time”has declared in the announcement of the project, Jensen Huang, CEO of Nvidia. «Incorporating American manufacturing helps us better satisfy the incredible and growing demand for chips and IA supercomputers, strengthens our supply chain And our resilience increases ».
Few doubt that the change from the production of chips is bound by the announcement of the strong tariffs of the Trump administration to China. Nvidia so far manufactures most of its products in Taiwan and the transfer of production is a movement of protection against them.
The company anticipates that this program will not only create “Hundreds of thousands of jobs” For Americans but also “It will generate billions of dollars in economic security in the coming decades”says the Executive. As part of this initiative, the TSMC factory in Phoenix, Arizona, will be the production center of the latest generation of Blackwell AI chips for NVIDIA, while Foxconn plants in Houston and Wistron’s in Dallas will be responsible for building IA supercomputer, the first that the green giant will produce in the country.
It is expected that mass production in these plants increases in the next 12 to 15 months. Nvidia has increased its investment in American manufacturing through association with Amkor and Spil for packaging operations and tests and has an alliance with Taiwan Semiconductor Manufacturing CO, the first World Foundrie that has also announced important investments in the United States.
An example will be the production of Blackwell chips, which are based on packaging technology chip-on-wafer-on-susstrate (Cowos) of TSMC, which began earlier this year at the Phoenix, Arizona factories. Nvidia has already used TSMC Cowos technology to incorporate high -gample memory memory modules in its high -end accelerators, although for the moment only outside the US. UU. At the end of last year, TSMC announced its plans to bring both its cowos technology and its fan -integrated in the United States.
Nvidia is a key actor in world technology and a strategic movement of this draft will have its importance. The company is not only the world’s first producer of dedicated graphics for consumption and professionals, but it is the main accelerators supplier for ia. Although the efficiency of models such as Deepseek is to be seen, it is about to see if so much hardware investment is necessary, in the short term the company sells chips by billionaire uprights.
