NVIDIA Raises GPU Prices as U.S. TSMC Production Costs Surge

NVIDIA has reportedly increased the official pricing of nearly its entire product lineup in response to market conditions and profitability goals. This move has also led industry partners to adjust prices accordingly.

One notable example is the ASUS RTX 5090 high-end graphics card, which initially launched earlier this year at NT$90,000. Recently, its market price has surged to NT$100,000, reflecting a price increase of over 10%. Similar price adjustments have been observed in H200 and B200 GPU modules, as well as in AI server systems.

In parallel, NVIDIA is accelerating its U.S.-based manufacturing strategy in response to policy changes. Since April, following the U.S. administration's introduction of reciprocal tariffs, the company announced plans to invest $500 billion over the next four years to build AI server manufacturing facilities in partnership with major players like TSMC and Foxconn. These new facilities, located in Arizona, have already begun producing Blackwell chips using TSMC's advanced 4nm process technology.

NVIDIA also emphasized that U.S. policies should focus on enhancing domestic competitiveness. The company cautioned that restricting chip exports to countries such as China could undermine America's leadership in R&D and innovation.

Meanwhile, industry sources confirm that NVIDIA is implementing further price increases for the H200 and B200 chips and modules to offset rising costs associated with transitioning between chip platforms. As a result, AI server prices are being revised upward by approximately 10–15% across the board.

Analysts within the supply chain note that the shift of Blackwell chip production to TSMC's Arizona facility has led to significantly higher costs in manufacturing, materials, and logistics. Like TSMC, NVIDIA has the ability to pass on these costs, making these price hikes largely anticipated.