China banned Nvidia 5090d v2 while CEO Jensen Huang was in town, report claims — move comes as Beijing pushes its AI tech companies to use homegrown chips

At a glance:

• Nvidia RTX 5090D V2 added to Chinese customs blacklist during Jensen Huang's visit
• The de‑fanged GPU has reduced VRAM and bandwidth compared with the vanilla RTX 5090
• Beijing continues to favour domestic chips such as Huawei’s, while keeping the U.S. H200 AI processor off‑limits

What happened

China’s customs authority reportedly placed Nvidia’s RTX 5090D V2 on a list of prohibited imports while CEO Jensen Huang was in the country as part of President Donald Trump’s state visit. The Financial Times broke the story, noting that the chip was flagged as a banned good even though it is marketed as an export‑friendly version of the flagship RTX 5090 GPU. Huang was a late addition to Trump’s entourage, boarding Air Force One in Alaska after initially being omitted from the guest list.

The RTX 5090D V2 is a slightly trimmed‑down variant of Nvidia’s top‑of‑the‑line gaming card. It was engineered to meet U.S. export‑control requirements while still offering Chinese gamers and 3D artists a high‑performance solution. Key differences from the standard RTX 5090 include:

• Lower video memory (VRAM) capacity
• Reduced memory bandwidth
• Slightly lower clock speeds These compromises make the card easier to ship to China, but they also leave it attractive to AI developers who have been cut off from Nvidia’s newer Blackwell‑based AI GPUs.

Why it matters for AI developers

Even though the RTX 5090D V2 is primarily a gaming GPU, Chinese AI firms have begun repurposing it for machine‑learning workloads. The chip sits below Nvidia’s most powerful AI processors, the H200 series, which were unexpectedly approved for export to China in late 2025 during a surprise move by the Trump administration. Despite that approval, Beijing has not granted Chinese AI companies permission to purchase the H200s, preferring instead to push domestically produced alternatives.

The H200 chips represent the current pinnacle of Nvidia’s AI hardware, featuring dedicated Tensor cores and architecture optimised for large‑scale model training. By denying access to these chips, the Chinese government is effectively forcing AI startups and research labs to either rely on the less capable RTX 5090D V2 or turn to home‑grown solutions such as Huawei’s Ascend series. This policy could reshape the competitive landscape of AI hardware in the world’s second‑largest market.

Beijing’s broader chip strategy

China’s alleged ban on the RTX 5090D V2 coincides with a longer‑term push to replace foreign semiconductor technology with domestic designs. The government’s stance is two‑fold: limit the flow of advanced U.S. AI chips that could be used for military applications, and accelerate the adoption of Chinese‑made processors to close the technology gap.

Huawei, in particular, stands to benefit from this policy shift. With the U.S. chip ban tightening, Huawei’s AI accelerators could capture market share that Nvidia once dominated. Analysts see this as a strategic move to allow Huawei to “leapfrog” Nvidia’s position in China, especially as the country’s AI firms look for alternatives that comply with local regulations.

Jensen Huang’s response and outlook

During a Bloomberg TV interview, Huang expressed optimism that market forces would eventually override political restrictions. He said, “My sense is that over time, the market will open.” The comment reflects Nvidia’s broader belief that demand for high‑performance AI hardware will drive a convergence of supply chains, even if short‑term barriers exist.

Nvidia has not commented publicly on the specific customs ban, but Tom’s Hardware has reached out for comment. The company’s leadership continues to monitor the situation closely, aware that a prolonged ban could erode the United States’ hardware advantage in the global AI race.

What to watch next

The next few months will be crucial for several reasons. First, Chinese AI firms may accelerate development of proprietary chips, potentially narrowing the performance gap with Nvidia’s H200 series. Second, the United States could respond with additional export‑control measures or diplomatic pressure, especially if Chinese AI capabilities begin to rival those of the West.

Finally, the broader tech community will be watching how other countries react. Europe and Japan have their own export‑control regimes, and any coordinated effort could further constrain the flow of cutting‑edge AI hardware. For now, the RTX 5090D V2 ban serves as a reminder that geopolitics remains a decisive factor in the semiconductor supply chain.