US, China tussle over advanced chips for key technologiesーNHK WORLD-JAPAN NEWS

Key Concepts

• Semiconductors/Chips: Integrated circuits crucial for AI, military systems, and digital devices. Specifically, the Nvidia H200 and AMD MI325X are highlighted as advanced AI processors.
• Export Controls: US regulations governing the sale of advanced chips to China, balancing national security with economic interests.
• Generative AI: A type of artificial intelligence focused on creating new content, heavily reliant on powerful GPUs like the H200.
• Decoupling: China’s strategy to reduce reliance on US technology and build independent supply chains.
• GPU (Graphics Processing Unit): A specialized electronic circuit designed to rapidly manipulate and display computer graphics; essential for AI processing.

US-China Semiconductor Competition: Export Controls and Strategic Considerations

The United States and China are engaged in a significant competition for dominance in semiconductor technology. The current focal point is the US government’s regulation of advanced chip exports to China, specifically concerning Nvidia’s H200 AI processor and AMD’s MI325X. These chips are vital components for artificial intelligence (AI), military applications, and a broad range of digital technologies.

New US Export Rules & Financial Implications

Last week, the US Commerce Department announced new rules requiring applications for exporting the H200 and MI325X to China to meet specific security requirements. Notably, former President Trump previously imposed a 25% tariff on these chips exported to China, stating, “We’re allowing them to do it, but the United States is getting 25% of the chips in terms of the dollar value and I think it's a very good deal.” This policy aims to generate revenue while permitting sales. These chips are manufactured by US companies, but physically produced in Taiwan.

The H200’s Technological Advantage

Experts emphasize the significant technological gap between the H200 and currently available Chinese chips. The H200 is specifically designed by Nvidia for generative AI and possesses “much greater functionality” compared to GPUs previously exported to China or those produced domestically. It is estimated that the H200’s price-performance ratio is three times that of domestically produced Chinese chips. As one expert stated, “If Chinese companies are able to use the H200, I believe China's AI industry will make considerable advances in AI technology.” This highlights the potential for rapid advancement in China’s AI capabilities if access is granted.

US Strategic Calculations: Balancing Competition and Profit

The US decision to allow some chip exports isn’t solely driven by easing restrictions. It reflects a strategic calculation to maintain US technological competitiveness and secure financial benefits. Concerns within the US government suggest that overly restrictive measures could inadvertently accelerate China’s indigenous chip development. Furthermore, the US chip industry relies on China for approximately 30% of its profits, creating significant economic pressure against stringent export controls. This internal conflict between security concerns and economic interests shaped the policy.

China’s Response: Blocking H200 Imports & Decoupling Efforts

Despite US approval for potential sales, Reuters reports that China has decided not to allow H200 chips into the country. This decision is attributed to security concerns and China’s broader strategy of “decoupling” its supply chains from the US. As one analyst explained, “AI is a core part of China's technology development. It would be a big risk for China in terms of security to depend on Nvidia's GPU as Nvidia is a US company. China has been wary of just such a situation.” This demonstrates China’s prioritization of technological independence and data security.

Long-Term Outlook: China’s Catch-Up Potential

While China’s technology currently lags behind the US in semiconductor development, analysts believe it will eventually catch up. The timeframe for this convergence remains uncertain. The analyst concluded, “China’s own technology is not likely to catch up to the US anytime soon, but over time it almost certainly will. The question is how long it will take.” This suggests a long-term strategic rivalry in the semiconductor industry.

Conclusion

The US-China semiconductor competition is a complex interplay of technological advancement, national security, and economic interests. The US is attempting to balance maintaining its technological edge with the economic realities of its chip industry’s reliance on the Chinese market. China, meanwhile, is pursuing a strategy of technological self-reliance, even if it means foregoing access to currently superior US technology. The situation remains dynamic, with the ultimate outcome dependent on the pace of innovation in both countries and the evolving geopolitical landscape.