Why Beijing Is Blocking Nvidia’s China Access

Beijing has reportedly expanded its restrictions on Nvidia, effectively halting purchases of the RTX Pro 6000D AI chip, adding to previous reported restrictions concerning Nvidia’s H20 GPU. Industry analysts suggest this move reflects a growing confidence in China’s domestic semiconductor industry and could be a strategic maneuver to gain leverage in ongoing trade negotiations with the U.S.

The reported ban follows earlier news that the H20 GPU, specifically designed to comply with U.S. export restrictions for the Chinese market, was facing regulatory hurdles in China, despite Jensen Huang publicly disclosing that it would resume sales. The H20 was seen as a compromise solution to maintain Nvidia’s presence in the critical Chinese market.

Initially, the H20 faced scrutiny amid concerns, fueled by some U.S. lawmakers, about potential tracking capabilities, leading to a national security review. However, the recent expansion to include the RTX Pro 6000D indicates a potentially broader shift in China’s strategy.

This development surfaces amidst a preliminary investigation by Chinese regulators into Nvidia for potential violations of anti-monopoly laws, further complicating the landscape for the U.S. chip giant in China.

While the motivations behind China’s actions remain multifaceted, industry experts point to several key factors: the accelerating development of domestic Chinese AI chip capabilities, a desire to reduce reliance on foreign technology amidst geopolitical tensions, and the potential use of market access as a bargaining chip with the U.S.

Reports of Chinese regulators’ initial intervention regarding the H20 centered around national security concerns, with some suggesting the chips could be outfitted with tracking systems, as proposed by American lawmakers. This narrative provided a justification for prioritizing domestic alternatives.

The decision to potentially block the RTX Pro 6000D, favored for AI smart factories and logistics applications, has industry experts reassessing China’s overall technological preparedness. Initial reports suggest that some Chinese AI companies had begun testing and verification work with Nvidia’s server suppliers, indicating a real demand that now faces disruption.

Some reports suggest that Chinese regulators concluded that domestic AI chip makers have achieved performance parity with Nvidia’s tailored offerings. But this comes amid analysis that China’s AI chip industry still faces considerable challenges, particularly in achieving the scale and production capacity required to fully meet domestic demand of its current need.

The implications extend beyond performance metrics. The ability to produce AI processors at scale and maintain consistent supply chains are critical for China’s technological ambitions. Developments suggest that local chipmakers are aiming to dramatically increase AI processor output next year in an attempt to further reduce China’s reliance on foreign suppliers like Nvidia.

“These actions underscore a significant shift in China’s confidence in its domestic semiconductor ecosystem,” notes Qingyuan Lin, a senior analyst covering China’s semiconductor industry at Bernstein. “The country is signaling a capacity and willingness to rely on homegrown solutions.”

Signs of progress in China’s domestic AI ecosystem are becoming increasingly apparent, with Huawei leading the charge.

Huawei recently unveiled new AI compute infrastructure powered by its Ascend chips, boldly claiming it as the “world’s most powerful.” While it is hard to verify this claim from an independent source.

Analysis from SemiAnalysis revealed that Huawei’s CloudMatrix system demonstrated competitive performance against Nvidia’s systems on some metrics. Despite limitations in individual chip performance, Huawei leveraged a significantly greater number of interconnected chips to achieve comparable and, in some instances, superior results. This approach highlights the importance of system-level integration and optimization.

Furthermore, Chinese AI startups like DeepSeek are actively working to ensure compatibility with domestically produced AI chips, signaling a broader commitment to supporting the local industry. Tech giants such as Alibaba and Baidu are reportedly integrating internally designed chips into their AI model training processes, reducing their dependence on Nvidia.

Despite these advances, skepticism remains regarding China’s near-term ability to completely wean itself off Nvidia. Ray Wang, research director for semiconductors, supply chain, and emerging technology at Futurum Group, emphasizes that achieving AI dominance requires a comprehensive suite of offerings, not just individual chip performance. He believes it is misleading to suggest that the development of the AI system is only with domestic alternatives and not using any of Nvidia’s systems.

Experts are increasingly framing China’s actions as a strategic attempt to gain leverage in trade negotiations with the U.S. By restricting access to its market, China can potentially influence U.S. policy decisions regarding export controls and technology transfers.

“This is likely a calculated negotiation tactic,” suggests AJ Kourabi, an analyst at SeminAlysis. “China is simultaneously promoting silicon self-sufficiency while also trying to secure access to the best possible chips. This move could be part of broader discussions involving tariffs and other trade-related issues.”

Indeed, the U.S. has signaled a willingness to consider allowing the sale of more advanced Nvidia chips to China, even those exceeding the capabilities of the H20. This willingness reflects the economic importance of the Chinese market to U.S. chipmakers, as well as the need to address concerns about the potential for fueling China’s military modernization.

The Biden administration has implemented increasingly stringent export controls on advanced chips, aimed at limiting China’s access to cutting-edge American technology. However, there are indications that this approach may be undergoing re-evaluation, with some advocating for a more nuanced approach that balances national security concerns with economic interests.

Reva Goujon, director at Rhodium Group, suggests that China’s could be creating an opportunity to negotiate access to more advanced GPUs from US suppliers. She also noted the timing of these announcements alongside other leverages, referencing to an anti-dumping investigation into imports of certain analog chips from the U.S.

“Beijing could be testing the waters, building leverage of its own in response to the evolving trade dynamic,” Goujon concludes.