Unpacking Nvidia’s $5.5 Billion Charge Tied to Export Restrictions on H20 Processors

In a recent and significant development, Nvidia, the renowned American multinational technology company, announced it would record a hefty quarterly charge of approximately $5.5 billion. This substantial charge is directly linked to the exportation of H20 graphics processing units (GPUs) to China and several other destinations.

The announcement led to a nearly 5% slide in Nvidia’s stock in extended trading. This financial jolt is primarily due to the restrictive regulations imposed by the U.S. government on the export of AI chips, which gained traction during the Biden administration. The restrictions were updated the following year to prevent the sale of advanced AI processors.

These changes in trade rules have had a profound impact on Nvidia’s historic growth trajectory. The company’s H20 GPU, an AI chip explicitly designed for China, was structured to comply with the U.S. export restrictions. This particular chip brought in an estimated revenue of $12 billion to $15 billion in 2024 alone.

However, export restrictions have led to a significant drop in revenue from China, according to Nvidia’s CEO, Jensen Huang. During the company’s last quarterly earnings’ call in February, Huang stated that revenue from China had plummeted to half of pre-export control levels. This drop is not only due to trade restrictions but also because of growing competition in China.

In fact, for the second consecutive year, Nvidia listed Huawei as a competitor in its annual filing. China is Nvidia’s fourth-largest sales region after the U.S., Singapore, and Taiwan. Therefore, the export restrictions could seriously harm Nvidia’s bottom line.

The H20 chip’s performance is comparably similar to the H100 and H200 AI chips used in the U.S. and elsewhere. However, it has slower interconnection speeds and bandwidth. This chip is based on a previous generation of AI architecture called Hopper introduced in 2022. Nvidia is currently focusing on marketing its new generation of AI chips, named Blackwell.

Furthermore, DeepSeek, a Chinese firm that recently disrupted markets with its competitive AI model R1, utilized H20 chips in its research. This indicates the prominence and potential of the H20 chip in AI applications, making the export restrictions even more consequential for Nvidia.

On top of existing Chinese export controls, Nvidia is also facing new restrictions on what it can export starting next month. This backdrop resulted from the “AI diffusion rules” first proposed by the Biden administration. However, Nvidia has maintained that additional control on its chips could stifle competition and possibly undermine U.S. competitiveness in technology.

In response to the export controls, Nvidia has relocated some of its operations, including testing and distribution, out of China. The company’s CEO, Huang, emphasized during last month’s company conference that Nvidia strives to comply with the law while recognizing the significant contribution of Chinese AI researchers, many of whom work at U.S.-based AI labs.

Ultimately, the U.S. government declared that the license requirement for H20 chips would be in effect indefinitely. This development, coupled with the anticipated financial impact of the $5.5 billion charge, will undoubtedly have significant implications for Nvidia’s fiscal first-quarter results, which will be reported on May 28.