US Commerce Secretary Howard Lutnick delivered a surprising assessment of the semiconductor trade war during a Senate Appropriations Committee hearing, claiming that China has not purchased a single Nvidia H200 chip “as of today”. According to the South China Morning Post, Lutnick attributed the lack of sales not to stringent US export controls but rather to Beijing’s own domestic policies aimed at fostering technological self-sufficiency. The testimony highlighted the complex and often contradictory narratives surrounding the flow of advanced AI hardware between the two superpowers.
Lutnick’s assertion that the US is not “selling our best chips to China under any circumstances” was accompanied by a nuanced explanation of the market dynamics. He argued that while Chinese cloud companies are eager to acquire the H200, the central government in Beijing has actively prevented these purchases to ensure that domestic investment remains focused on homegrown alternatives. This perspective shifts the locus of control from Washington’s regulatory apparatus to Beijing’s strategic industrial planning, suggesting that China’s drive for autonomy is as significant a barrier to US tech sales as any export restriction.
(Related: Jensen Huang Called US Chip Export Controls “Extremely Stupid” — Here Is What He Actually Said)
Skepticism from the Senate and the Military
The commerce secretary’s claims were met with immediate skepticism from lawmakers and military officials concerned about the potential dual-use applications of advanced AI hardware. Senator Chris Coons (D-Delaware) directly challenged Lutnick, pointing out that the H200 chips are six times more powerful than the previously approved H20 models. Coons questioned the administration’s confidence in its ability to prevent these highly capable chips from ultimately reaching the People’s Liberation Army (PLA), highlighting the inherent difficulties of tracking the end use of dual-use technologies within China’s opaque ecosystem.
This skepticism was echoed by Navy Admiral Samuel Paparo, commander of the US Indo-Pacific Command, who testified the previous day that providing the PLA with access to more advanced AI chips would tangibly improve their warfighting capabilities. The military’s assessment underscores the high stakes involved in the semiconductor trade, where commercial sales can have direct and profound implications for national security. The debate over the H200 reflects a broader tension within the US government between the desire to maintain technological leadership through commercial dominance and the imperative to restrict adversaries’ access to critical capabilities.
The approval of the H200 for export to China, which occurred following President Trump’s October summit with President Xi Jinping in South Korea, remains a point of contention. While the administration has maintained a strict prohibition on the more advanced Blackwell architecture to preserve a “generational gap”, the H200 represents a significant leap in processing power over the H20. The arrangement, which reportedly includes an “export fee” of 25% of sales payable by Nvidia to the US government, has drawn criticism from those who view it as a dangerous concession.
The Disconnect Between Orders and Shipments
Lutnick’s testimony also highlighted a stark disconnect between the stated demand from Chinese customers and the actual flow of hardware. Nvidia CEO Jensen Huang had previously indicated that Chinese clients had placed orders for the H200, yet Lutnick insisted that zero units had actually shipped. This discrepancy raises questions about the efficacy of the export control regime and the true state of the market. Whether the lack of shipments is due to bureaucratic delays, deliberate slow-walking by the Commerce Department, or genuine intervention by Beijing remains unclear.
(Related: US Lawmakers Target DUV Lithography in New Push to Close China’s Chip Loophole)
The ongoing debate over the H200 and the broader trajectory of US export controls will likely feature prominently in the upcoming summit between President Trump and President Xi, scheduled for May 14-15 in Beijing. While US Trade Representative Jamieson Greer has stated that the summit will focus primarily on “non-sensitive goods”, the underlying tensions over technology transfer and strategic competition are unavoidable. As the two nations navigate this “delicate balance”, the fate of the H200 serves as a microcosm of the broader struggle for dominance in the AI era.
The Commerce Department’s Bureau of Industry and Security (BIS), which oversees export controls, has faced intense scrutiny over its handling of these issues. Representative Gregory Meeks, the top Democrat on the House Foreign Affairs Committee, recently characterized the BIS under the Trump administration as “a picture of dysfunction and delays”. This criticism suggests that the challenges in managing the flow of advanced technology are as much administrative as they are geopolitical, further complicating the US effort to maintain its technological edge while managing its complex relationship with China.