While Nvidia, the leader in artificial intelligence (AI) chips, is facing a direct hit from semiconductor export restrictions to China, it has significantly increased the prices of graphic processing units (GPUs) by 10 to 25% to secure revenue.
According to Taiwan's DigiTimes on the 12th, Nvidia has recently raised the prices of its GPU products, allowing its distribution partners to increase prices as well. As a result, prices for GPU modules and server products, including AI chips H200 and B200, have risen by 10 to 15%. The price of the RTX5090, evaluated as the highest specification among graphics cards for PCs, is at around 20 million won, reflecting an increase of over 25% compared to early this year.
The comprehensive price adjustment is Nvidia's desperate measure to proactively respond to declining revenue. Taiwan's foundry company TSMC has started producing Nvidia's Blackwell chips using 4-nanometer (with 1 nanometer being one billionth) processes at its Arizona plant starting in the first quarter of this year, during which manufacturing expenses surged. TSMC explains that operational costs, including material and logistics expenses, are about double at the U.S. factory compared to production in Taiwan. The industry believes that chip prices must rise as much as the increased production costs.
In addition to the burden of production expenses in the United States, Nvidia is in a situation where it must reflect a loss of $5.5 billion (approximately 7.7 trillion won) due to semiconductor export restrictions in the first quarter of this year alone. The Trump administration imposed export restrictions on Nvidia's AI chip H20 sold in China last month to hinder China's AI growth. Nvidia stated that not being able to export the H20 chip could result in about 8 trillion won in loss expenses for just the first quarter.
Jensen Huang, CEO of Nvidia, noted that the business has been seriously affected by these regulations. During his visit to Beijing last month, he lamented, "The U.S. government’s tightening of semiconductor export controls has already had a substantial impact on Nvidia’s operations." Nvidia's sales in China amounted to $17.1 billion (approximately 25 trillion won) last year, accounting for 14% of the company's total revenue. Once dominating 90% of the Chinese AI chip market, Nvidia’s market power in China is weakening as U.S. regulations tighten.
Nvidia emphasizes that the Chinese market will become increasingly important. During a recent conference held in the United States, CEO Huang expressed concern, stating, "While exports to China are restricted, the Chinese market will likely grow to about $50 billion (approximately 69 trillion won) in a few years, making it a massive market we are missing out on. If we do not supply products to specific markets and completely withdraw, Huawei will take that place." Industry forecasts suggest that Nvidia will likely release a lower-spec version of the H20 chip, which has limited exports to China, around July.
The market is also remaining vigilant regarding Nvidia's forecasts. Nvidia's revenue guidance for the first quarter of fiscal year 2026 (February to April) is about $430, reflecting a 65% increase compared to the same period last year. Although the growth trend continues, it significantly falls short of last year's growth rate of 262%, raising concerns about performance slowdown. Ahead of the earnings announcement at the end of this month, CEO Huang plans to visit Taiwan from today and meet with major partners to reorganize the supply chain.