Nvidia CEO Jensen Huang recently announced that the company plans to invest nearly $500 billion in purchasing chips and other electronics made in the United States over the next four years. This bold move reflects Nvidia's commitment to reshaping its supply chain and reducing reliance on overseas manufacturing.

Speaking at a recent event, Huang stated, "Overall, we will procure, over the course of the next four years, probably half a trillion dollars' worth of electronics in total. And I think we can easily see ourselves manufacturing several hundred billion of it here in the U.S."

Nvidia is already producing its latest data center chips and Blackwell systems at manufacturing sites run by TSMC and Foxconn in the U.S. Huang confirmed that production is underway at TSMC's Fab 21 in Arizona, stating, "We are now running production silicon in Arizona." This means Nvidia has moved beyond prototypes and is manufacturing fully functional chips domestically, though the exact production volume remains undisclosed.

The decision comes amid growing geopolitical tensions and potential trade tariffs. By shifting production to the U.S., Nvidia aims to mitigate risks tied to Taiwan and other key manufacturing regions in Asia. Collaborations with TSMC, Foxconn, and other American-based suppliers are central to this strategy.

Nvidia's product portfolio extends beyond GPUs for PCs and data centers. The company also designs CPUs, DPUs, networking chips, and system-on-chips (SoCs) for vehicles and embedded applications — many of which could be produced in upcoming U.S. semiconductor fabs.

Furthermore, Nvidia relies on a wide range of components from other manufacturers, including AMD, Intel, Micron, Samsung, and SK hynix. With companies like Micron, SK hynix, Analog Devices, GlobalFoundries, and Texas Instruments expanding their U.S. production capacities in the coming years, Nvidia is positioned to source more components locally. Micron's new fab is expected to come online in 2027, with SK hynix following in 2028 and Texas Instruments launching its SM1 fab in 2025.

Nvidia's latest hardware advancements also fuel this shift. The upcoming Blackwell Ultra NV72 system, set for release in late 2025, is designed to deliver unprecedented performance. Each node integrates two upgraded Blackwell GPUs, one Grace CPU, and up to 288GB of HBM3e memory, achieving Dense FP4 performance of 15 PFlops per node. A complete system boasts 72 nodes, packing 144 GPUs, 72 CPUs, 20TB of HBM3e memory, and 40TB of DDR5 RAM — a 50% performance leap over its predecessor.

As Nvidia continues to expand production in the U.S., the company's monumental investment signals a shift toward greater self-sufficiency and technological leadership in the semiconductor industry. While the exact breakdown of the $500 billion procurement remains unclear, it marks a pivotal moment for American chip manufacturing.