US Approved Nvidia H200 Chip Sales to China, But Beijing’s Hesitation Stalls Deliveries

In a move that appeared to signal a thaw in the ongoing tech cold war, the U.S. Commerce Department granted approval for the sale of Nvidia H200 AI chips to approximately 10 major Chinese firms earlier this year. These aren't just any chips—H200s are the workhorses behind next-gen AI, from realistic NPCs to cloud gaming infrastructure. The list of approved buyers reads like a who's who of China's tech elite: Alibaba, Tencent, ByteDance, JD.com, Lenovo, and Foxconn, among others. Each approved customer was authorized to purchase up to 75,000 H200 chips under strict licensing terms. Yet, months later, not a single chip has crossed the Pacific. While Washington has relaxed export controls for select buyers, Beijing is holding back, wary of undermining its domestic semiconductor ambitions. This standoff reveals a high-stakes geopolitical tug-of-war over AI dominance, with Nvidia caught in the middle and its China market share in freefall.

The U.S. Green Light—What Was Approved and Why

The approvals came with stringent conditions designed to balance economic interests with national security concerns. U.S. rules require Chinese buyers to demonstrate "sufficient security procedures" and guarantee non-military use of the chips. Nvidia, in turn, must certify that it maintains sufficient inventory in the United States to meet domestic demand before any exports can proceed. For context, unlike consumer GPUs like the RTX 4090, the H200 is a data-center chip designed for massive AI training workloads—the kind that powers everything from game development tools to cloud gaming servers.

Perhaps the most eyebrow-raising condition is the 25% revenue share from chip sales negotiated by the Trump administration. This arrangement requires all chips to physically pass through U.S. territory due to legal constraints, ensuring American authorities maintain oversight throughout the transaction. The approval reflects a nuanced U.S. strategy: allowing limited sales to preserve economic ties with China's massive AI market while retaining control over advanced technology flows.

U.S. Commerce Secretary Howard Lutnick framed the approvals as a matter of business pragmatism. "We're not going to let our companies go out of business because they can't sell to China," he stated, acknowledging the delicate balancing act Washington must perform.

Beijing's Cold Shoulder—Why China Is Holding Back

Despite the American green light, deliveries remain frozen. The reason, according to Lutnick, is simple: China's central government has not allowed purchases to proceed. "They want to keep investment focused on their domestic industry," he explained.

Beijing's hesitation is driven by strategic concerns that imports could weaken China's push to develop homegrown AI chips. This is not mere rhetoric. Firms like DeepSeek are already increasingly using domestic chips from Huawei, signaling a tangible shift in China's tech ecosystem. The Chinese government has backed this transition with concrete policy measures: China's State Council recently issued two supply chain security regulations, prompting a government-wide effort to identify and eliminate foreign dependencies in critical technology sectors.

For Beijing, allowing Nvidia's H200 chips into the country now would risk undermining years of investment and political capital poured into building a self-sufficient semiconductor industry. It's a calculated gamble—sacrificing short-term access to world-class hardware for long-term technological independence.

Nvidia's Dilemma—From Market Dominance to Near Zero

The numbers tell a stark story, though the timeline requires careful unpacking. Before U.S. export curbs took effect (pre-2023), Nvidia commanded approximately 95% of China's advanced chip market, with China alone accounting for roughly 13% of the company's global revenue. CEO Jensen Huang once estimated China's AI market alone would be worth $50 billion in 2025—a tantalizing prize for any chipmaker.

But the reality has been brutal. By April 2026, Huang acknowledged that Nvidia's China market share had effectively fallen to zero. The company's current market share in China now sits at under 60%—a catastrophic collapse by any measure.

The irony is that demand for H200 chips remains feverishly high in China. Some firms are reportedly considering black market purchases to circumvent official channels, a testament to the chip's indispensable role in AI training and inference workloads. Huang has publicly disagreed with a total ban on AI chip exports to China, arguing that such restrictions could inadvertently allow Chinese AI chips to dominate the global market by shielding them from competition.

The Geopolitical Chessboard—Washington's Hardliners vs. Nvidia's Lobbying

The delay has been welcomed by China hardliners in Washington. Chris McGuire of the Council on Foreign Relations argued that even limited sales would reduce U.S. firms' access to Nvidia chips and diminish America's AI lead over China. "Every chip that goes to China is a chip that doesn't go to an American company," he noted.

Nvidia has not taken the situation lying down. CEO Jensen Huang joined a White House delegation to Beijing after an invitation from President Donald Trump, aiming to unlock the stalled H200 chip sales. The delegation's mission underscores a fundamental tension: U.S. national security interests versus the economic interests of American tech giants. For Nvidia, losing the Chinese market is not just about lost revenue—it's about ceding technological ground to competitors that will only grow stronger in isolation.

The standoff highlights how semiconductor policy has become a central battleground in U.S.-China relations. The outcome will shape the future of global AI supply chains and the balance of technological power between the two superpowers.

What's Next—Will the Chips Ever Flow?

The immediate future depends on whether Beijing relents or doubles down on domestic chip development. If China continues to block imports, Nvidia's market share could fall further, accelerating the rise of Chinese AI chip makers like Huawei. If sales proceed, it could temporarily boost Nvidia's revenue but risk undermining U.S. strategic objectives by providing China with advanced hardware.

The black market demand for H200 chips suggests that even if official channels remain blocked, unofficial channels may emerge—a development that would create its own set of regulatory headaches for both governments. The situation remains fluid, with both sides using the chip sale as leverage in broader trade and technology negotiations.

The approval of Nvidia H200 chip sales to China represents a rare moment of U.S. export relaxation, but Beijing's refusal to allow deliveries underscores a deeper strategic shift. China is prioritizing domestic AI chip development over short-term access to advanced foreign hardware, even as Nvidia's market share plummets and demand remains high. The outcome of this standoff will not only determine Nvidia's future in China but also set the stage for the next phase of the global AI arms race. For now, the chips remain in limbo—a symbol of the delicate balance between economic interdependence and technological sovereignty. The question is: who will blink first?

What This Means for Gamers: If China successfully develops its own AI chips, it could accelerate competition in the GPU market, potentially driving down prices and increasing availability for consumer graphics cards. On the flip side, continued restrictions could slow AI advancements that benefit gaming, such as smarter NPCs, more realistic physics, and improved cloud gaming infrastructure.