The US government has officially cleared around 10 major Chinese tech firms to purchase Nvidia H200 AI chips, yet not a single delivery has materialized. This unprecedented deadlock has prompted Nvidia CEO Jensen Huang to join President Donald Trump on a high-stakes diplomatic trip to Beijing, aiming to untangle a complex web of export controls and domestic pushback. Before US export curbs tightened, Nvidia commanded about 95% of China's advanced chip market, which historically accounted for 13% of its total revenue. Huang recently estimated that the Chinese AI market alone would be worth $50 billion this year.
The US Commerce Department has authorized tech giants including Alibaba, Tencent, ByteDance, and JD.com to purchase up to 75,000 chips each under strict licensing terms. A select group of distributors, including Lenovo and Foxconn, have also been approved to act as intermediaries. Lenovo confirmed its status in a statement, noting it is authorized to sell the hardware as part of Nvidia's export license. However, despite this regulatory green light from Washington, the deals have completely stalled as Chinese firms pull back following direct guidance from Beijing.
The hesitation in China reflects a strategic calculation to foster reliance on homegrown hardware. Commerce Secretary Howard Lutnick confirmed this dynamic, stating that the Chinese central government has not allowed the purchases because they are trying to keep their investment focused on their own domestic industry. As a result, local AI developers like DeepSeek are increasingly touting their reliance on domestic alternatives, particularly those developed by Huawei, underscoring Nvidia's precarious position in the region.
Any deal that allows Nvidia to sell more chips to China means fewer Nvidia chips for U.S. firms, and a smaller U.S. lead in AI over China.
- Chris McGuire, Council on Foreign Relations
The path to a completed sale is heavily obstructed by thorny conditions on both sides. US rules issued in January require Chinese buyers to demonstrate sufficient security procedures to ensure the hardware is not used for military purposes. Furthermore, a Trump-negotiated arrangement demands that the US receive 25% of the revenue from the chip sales. Because US law does not permit the direct imposition of export fees, this structure requires the chips to physically pass through US territory before being shipped to China.
This specific routing requirement has prompted severe unease in Beijing over potential tampering or hidden vulnerabilities. Scrutiny intensified after the Chinese State Council issued two recent supply chain security regulations, triggering a government-wide effort to eliminate foreign dependencies in critical technology infrastructure. Meanwhile, China hardliners in Washington welcome the delay, arguing that prioritizing Nvidia's financial interests could inadvertently help Chinese rivals close the technological gap.
The Hidden Cost of Hardware Sovereignty
The standoff over the Nvidia H200 AI chips reveals a critical shift in the global tech war: tariffs and export bans are no longer the only weapons; forced domestic adoption is now Beijing's primary shield. By actively blocking its own tech giants from buying approved Nvidia hardware, the Chinese government is essentially subsidizing the Huawei and DeepSeek ecosystem. This is a painful, forced incubation period designed to break the addiction to American silicon.
For Nvidia, the implications are severe. Huang's warning that the company's share of AI accelerators in China has effectively fallen to zero is not just a temporary dip - it is a structural lockout. If Alibaba and Tencent are forced to optimize their massive Large Language Models (LLMs) for Huawei's architecture instead of Nvidia's CUDA platform, reversing that technical migration later will be nearly impossible.
Ironically, the strict US conditions - specifically the 25% revenue cut and the mandatory routing through American territory - have handed Beijing the perfect national security justification to enforce this pivot. By making the purchase process look like a potential supply chain vulnerability, the US may have inadvertently accelerated the exact scenario it sought to prevent: a fully independent, self-sustaining Chinese AI hardware ecosystem.
