Nvidia Ends Chip Sales to China Following CEO Transition and New Export Controls

Image: TechCrunch · Source
Nvidia will cease selling H200 AI chips in China amid leadership changes and tightening US export rules, potentially impacting up to 20% of its revenue linked to the Chinese market.
Nvidia has announced it will stop selling its H200 AI accelerator chips in China, marking a significant shift in its business strategy following the recent departure of CEO and co-founder Jensen Huang. The interim CEO, Colette Kress, revealed that the company is closing a controversial loophole that had allowed advanced chip sales to China under a special arrangement brokered by Huang and former President Donald Trump. This deal routed shipments through the US, enabling tariff collection but also drawing criticism from US policymakers eager to restrict China’s access to cutting-edge AI technology.
Huang had long advocated for maintaining Nvidia’s presence in China, viewing conceding the market as detrimental to the US technology sector. However, his ouster—linked in part to a personal scandal involving ties to a company under investigation for illicit chip exports—and the imposition of stringent new export controls have altered Nvidia’s stance. New regulations now prohibit advanced US chips from being used in foreign data centers serving Chinese customers remotely, further constraining Nvidia’s ability to supply Chinese AI labs.
Analysts at investment bank Jefferies estimate that these export controls could reduce Nvidia’s data center revenue by $44 to $62 billion next year. Yet, Kress suggested that immediate revenue impacts might be muted because of existing demand backlogs for chips like Vera Rubin. Still, the long-term implications are considerable: Nvidia may lose customers not only in China but also in Southeast Asian markets such as Singapore, Malaysia, Thailand, and Indonesia, where demand is tied to Chinese compute needs.
China itself has shown reluctance to accept the H200 shipments, aiming instead to cultivate its domestic chip industry despite preference among AI developers for superior Nvidia hardware. Moreover, Chinese AI models have reportedly been trained using Nvidia’s Grace Blackwell chips located outside China, an approach now curtailed by new export restrictions requiring data centers to certify no service to Chinese clients. Without such certification, data centers may shift to alternative chip suppliers, further eroding Nvidia’s market share.
Nvidia’s new board-level committee on national security and export compliance underscores the company’s focus on aligning with global trade policies amid escalating US-China tech tensions. This strategic retreat from China marks a pivotal moment for Nvidia and the broader AI hardware ecosystem, reflecting the intersection of geopolitics and technology innovation in shaping the industry’s future.
Sources and original reporting
Read the original source ↗

Comments (0)
No comments yet. Start the discussion.
Write a comment
Comments are published after moderation. Your name and comment will be visible publicly. Account