Acting CEO Colette Kress announced on Tuesday that Nvidia will cease selling its H200 chips in China, marking the end of the company's use of a controversial loophole within U.S. restrictions on advanced chip exports.
Discussions regarding China's access to leading AI accelerators took center stage during Nvidia's first earnings call since CEO Jensen Huang stepped down last week. This announcement occurred alongside Nvidia's declaration to establish a board-level committee focused on national security policies, export control compliance, and global trade risks.
The arrangement to sell H200 chips in China was originally brokered by Huang and former President Donald Trump. This specific deal required chips to be routed through the United States, a process that allowed the government to levy a 25% tariff.
This strategy drew criticism from hardline China hawks who sought to deny Chinese laboratories computing power and maintain the U.S.'s advantage in artificial intelligence. However, Huang had long argued that it was vital for the U.S. to compete in the Chinese market. In April, he stated to podcaster Dwarkesh Patel, "to concede that market for the United States technology industry is a disservice to our country."
Huang's departure, partly attributed to an affair with the CEO of a company under investigation for smuggling chips into China, has coincided with new regulations issued Monday. These rules prohibit advanced U.S. chips from being sold to foreign data centers that provide services to Chinese customers remotely.
Investment bank Jefferies predicts these regulations could reduce Nvidia's data center revenue by $44 billion to $62 billion next year. Responding to the new export controls, Kress noted that the existing backlog for advanced chips like Vera Rubin suggests the rules will not impact the company's top-line revenue.
In the long term, Nvidia risks losing potential customers in Singapore, Malaysia, Thailand, and Indonesia. While critics argue that up to 20% of Nvidia's revenue is linked to China—highlighted by the Singapore subsidiary reaching $104 billion in 2025—the company admits these figures will change.
Even prior to this decision, China largely refused to accept H200 deliveries. Jacob Feldgoise, a researcher at Georgetown's Center for Security and Emerging Technology, noted that the government prefers to generate demand for domestic chips, even if major AI users desire the superior Nvidia equipment.
Chinese frontier labs continue to find ways to access Nvidia chips via alternative locations. Analysis indicates that all of China's leading AI models show evidence of having been trained on prohibited Nvidia Grace Blackwell chips.
While this training may have occurred in data centers outside China, the new rules will bar sales to such facilities. Export control attorney Dan Fisher-Owens explained that if large data centers intend to purchase Nvidia chips, they must certify they do not serve any Chinese clients, or else they will be forced to consider alternative chip options, further eroding Nvidia's market share.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.
