NVIDIA announced the establishment of a US-based AI server industry chain, aiming for a production value of $500 billion in four years.

US AI chip giant Nvidia announced on Monday that it is collaborating with a group of suppliers and manufacturing plants to achieve the localization of AI supercomputer production in the US in the coming years.

(Source: Nvidia)

Nvidia disclosed that it has activated "over a million square feet" of production space in Arizona for Blackwell chip manufacturing and testing, and is establishing an AI supercomputer production line in Texas with partners.

As the most critical part of the entire industry chain, the Blackwell chip itself has already started production at TSMC's wafer fab in Phoenix. Meanwhile, Nvidia is also setting up supercomputer manufacturing bases in Houston (in collaboration with Foxconn) and Dallas (in collaboration with Wistron), with both facilities expected to gradually increase production over the next 12-15 months.

In the field of advanced packaging and testing, Nvidia will collaborate with Amkor Technology and SPIL to conduct packaging and testing cooperation in Arizona.

As a "promise" to US authorities, Nvidia also emphasized in its statement that it plans to produce "up to $500 billion" worth of AI infrastructure in the US over the next four years with the aforementioned partners, which is expected to create "hundreds of thousands" of jobs.

Nvidia founder Jensen Huang also stated in the press release that manufacturing the core engine of AI infrastructure in the US for the first time will help the company better meet the surging market demand for AI chips, while strengthening the supply chain and enhancing risk resilience. Undoubtedly, the "risk" here refers to the tariff measures haphazardly introduced by the Trump administration.

Facing the prospect of a collapsing industry chain and US capital market, the US government quietly announced last Friday that it would temporarily exempt "reciprocal tariffs" on imports of goods such as smartphones, chips, and servers. However, Trump and Commerce Secretary Lutnick also stated over the weekend that different forms of industry tariffs on the semiconductor supply chain may still be imposed in the future. Dan Ives, an analyst at Wedbush Securities, accused in a report last Sunday that

the continuous stream of news from the White House has created immense confusion, leaving the industry and investors dizzy, and bringing significant uncertainty and chaos to companies trying to plan their supply chains, inventory, and demand.

For Nvidia, the "US domestic industry chain" itself could be an unfavorable factor for stock performance. Some investors may recall that after releasing a largely better-than-expected earnings report at the end of February, Nvidia's stock price plummeted by 8% in a single day. The market interpretation at the time pointed to a 3% YoY decline in the company's gross margin, which was as high as 73%. Nvidia also explained that the decline in gross margin was due to the increased complexity and cost of new AI products.

However, compared to the mature industry chain in Asia, "producing advanced chips in the US" also means that costs will spiral out of control. Not to mention the more demanding AI chips, even Apple's iPhones would find it economically unfeasible to transition to "Made in the USA."

Dan Ives once warned that if Apple forcibly moves its iPhone production line back to the US, the $1,000 iPhones made in China and India might have to be priced at $3,000.

Therefore, at the earnings call at the end of May, global investors will closely monitor the impact of "Made in the USA" on Nvidia's business profits.

Of course, Nvidia's statement itself also has the potential to prevent the US government from further disrupting the global electronics industry chain. According to previous reports by Cailian Press, in early April, Jensen Huang visited Mar-a-Lago and successfully persuaded Trump to suspend plans to expand restrictions on the export of H20 chips to China. It is said that Huang's condition was to invest in building an AI data center industry chain in the US.