China Bans Nvidia Gaming Chips During Huang’s Beijing Visit, Expanding Semiconductor Decoupling Beyond AI
Beijing blocks RTX 5090 D v2 imports as Nvidia CEO joins Trump delegation, signaling shift from targeted AI restrictions to comprehensive technology sovereignty.
China blocked Nvidia’s GeForce RTX 5090 D v2 gaming GPU from entering the country on 19 May, refusing to issue import permits for a chip designed exclusively for the Chinese market—a ban that coincided with CEO Jensen Huang’s presence in Beijing as part of President Trump’s diplomatic delegation.
The move marks a strategic escalation beyond AI-specific hardware. Chinese customs authorities notified logistics companies that the RTX 5090 D v2 would not receive import clearance, according to WCCFtech, effectively extending Beijing’s semiconductor protectionism from data center AI accelerators to consumer electronics. The timing—during Huang’s first visit to China alongside a sitting US president—suggests deliberate messaging about technological sovereignty rather than coincidence.
Huang joined Trump’s Beijing trip at the last minute after the president personally called him during a refueling stop in Alaska on 13 May, CNBC reported. The Nvidia CEO’s presence was intended to facilitate semiconductor trade discussions, but Beijing’s gaming chip ban delivered an unambiguous counter-signal.
From Surgical Restriction to Blanket Exclusion
China’s approach to US chip imports has evolved from precision targeting to categorical rejection. While previous restrictions focused on high-performance AI accelerators like the H100 and H200, the gaming GPU ban reveals Beijing’s willingness to forgo consumer-grade hardware entirely in pursuit of parallel domestic supply chains. The RTX 5090 D v2 represented Nvidia’s attempt to comply with Chinese performance restrictions—a specially designed variant that met Beijing’s specifications yet still triggered an import ban.
The shift reflects coordinated state policy rather than technical concerns. In December 2025, China added Huawei’s Ascend processors and Cambricon’s AI chips to the official government procurement list, creating billions in guaranteed revenue for domestic chipmakers while systematically excluding Nvidia, TrendForce reported. Cambricon now targets 500,000 AI accelerators in 2026—including 300,000 units of Siyuan 590/690 processors—more than tripling 2025 output, per Tom’s Hardware.
The H200 That Never Shipped
Beijing’s gaming chip ban arrived as the contradiction in US-China semiconductor policy became undeniable. In January 2026, the US Department of Commerce approved approximately ten Chinese firms—including Alibaba, Tencent, ByteDance, and JD.com—to purchase Nvidia H200 chips under a 25% revenue-sharing arrangement, with a limit of 75,000 units per customer, according to Built In. Yet as of mid-May, not a single H200 has reached Chinese customers.
“The Chinese central government has not let them, as of yet, buy the chips, because they’re trying to keep their investment focused on their own domestic industry.”
— Howard Lutnick, US Commerce Secretary
US Commerce Secretary Howard Lutnick confirmed the impasse in May 2026 statements to Tom’s Hardware, acknowledging that Beijing is actively blocking imports despite Washington’s export approvals. The dynamic reveals asymmetric leverage: the US can grant licenses, but China controls the final gate. Beijing’s strategy prioritizes domestic chipmaker development over immediate performance gains from imported silicon.
Semiconductor Leverage Extends to Taiwan
The gaming GPU ban occurred within days of Trump’s decision to withhold commitment on the latest round of Taiwan arms sales following direct warnings from Chinese President Xi Jinping. Trump indicated on 15 May that he would not approve the sales after Xi cautioned that mishandling the Taiwan issue could lead to conflict, ABC News reported.
The convergence suggests Semiconductors function as leverage in multiple directions simultaneously. Washington uses chip export licenses to extract concessions from Beijing, while China weaponizes import controls to both accelerate domestic substitution and pressure US chipmakers. Taiwan—the world’s dominant advanced semiconductor manufacturer—becomes both a technical dependency and a geopolitical bargaining chip, with arms sales serving as a visible proxy for US commitment.
Beyond AI: Comprehensive Decoupling
The expansion from AI chips to gaming GPUs signals that semiconductor decoupling has entered a new phase. Analysis from Startup Fortune notes that if Beijing is prepared to block gaming GPUs, then almost any advanced semiconductor product can become a target. The RTX 5090 D v2 ban eliminates the premise that consumer electronics would remain exempt from geopolitical friction.
China’s domestic chip ecosystem now spans AI accelerators, gaming GPUs, and general-purpose processors, with state procurement lists ensuring demand regardless of technical parity with US products. The Council on Foreign Relations has documented the performance gap between Huawei’s Ascend series and Nvidia’s H100/H200, yet Beijing’s strategy accepts near-term capability deficits in exchange for long-term autonomy.
- Gaming GPU ban extends decoupling beyond data center AI to consumer electronics, eliminating safe categories
- Zero H200 deliveries despite US export approvals reveal China controls final import gate regardless of Washington’s licensing decisions
- Domestic chipmaker procurement mandates create guaranteed revenue base independent of technical performance parity
- Taiwan arms sales delay suggests semiconductors function as leverage in both trade negotiations and security commitments
What to Watch
Monitor whether China extends import restrictions to other Nvidia product lines, including professional visualization GPUs and automotive chips. Track Cambricon and Huawei production ramp timelines—500,000 units represents significant volume but remains fractional compared to Nvidia’s global output. Watch for US responses to the H200 import blockade, particularly whether Commerce revokes licenses that cannot be fulfilled. The next round of Trump-Xi talks will reveal whether semiconductor trade restrictions ease or whether both sides commit to parallel technology stacks with minimal interoperability. Huang’s presence in Beijing during the gaming GPU ban suggests the era of US chipmakers serving both markets simultaneously is ending—the question is whether Washington and Beijing negotiate a managed separation or allow categorical decoupling to accelerate.