The China bandwagon for US chip manufacturers has started to roll once again, with NVIDIA and AMD introducing aggressive solutions in the market, but this time, the dynamics are different.
By now, the reports of NVIDIA offloading the Hopper H200 AI chips to China have become mainstream, but there's also a breakthrough with AMD, which we'll discuss ahead as well. During the initial quarters of 2025, American chip manufacturers faced significant challenges selling their products to China, as geopolitical tensions led to policy changes that heavily impacted firms like NVIDIA and AMD. The situation reached a point where NVIDIA's CEO, Jensen Huang, declared that their market share in China was at 'zero percent', indicating that the regional business had been completely disrupted.
NVIDIA/AMD China Breakthrough: A Mix & Match of Lobbying & Architectural Adjustments
Politics has played a significant role in shaping the Chinese business landscape for AI chip manufacturers, and as a result, both NVIDIA and AMD have had to consistently work with the current US administration to ensure that their chip offerings comply with the relevant architectural regulations. We saw a breakthrough in policy restrictions back in August, when President Trump declared that the US government would be taking a 15% cut of the total shipments being sent to China by AMD and NVIDIA, which allowed the tech giants to sell chips like the Hopper H20 and Instinct MI308 AI chips.
However, in a twist of events, Beijing actually showed resistance towards the H20 AI accelerators, even launching a regulatory investigation to determine whether the chips contain a security backdoor. China enforced a 'national policy' that forced domestic AI giants to focus on adopting chips from companies like Huawei, Cambricon, and BirenTech, which eventually meant that the prospect of NVIDIA and AMD re-entering China had significantly diminished. Here's what Jensen had to say around this timeline.
At the moment, we are 100% out of China, and so China is 0%. We went from a 95% market share to 0%, and so I can't imagine any policymaker thinking this is a good idea. In all our forecasts, we assume zero for China. If anything happens in China, it will be a bonus.
Meanwhile, China continued to make rapid advancements in the AI segment, with firms like Huawei introducing impressive chip roadmaps which focused on shifting towards self-built elements, and this prompted a massive wave of skepticsm within the AI industry, since Beijing had been rapidly gaining ground against the US, to the point where NVIDIA's CEO Jensen Huang revealed that the China is now just "nanoseconds" away from America. Such statements eventually paved the way for another regulatory breakthrough for NVIDIA/AMD, which we saw a few weeks ago.
President Trump announced in the second week of December that NVIDIA would be allowed to sell its Hopper H200 AI chips to China, subject to a 25% tax, a higher percentage compared to what we saw with the H20. The development was seen as both optimistic and a concern for NVIDIA, considering that it meant that Team Green would be coming back to China with a more powerful option, but at the same time, the approval has brought in supply chain complexities and added costs, which we'll summarize in this image:
We'll dive into how the H200 and Instinct MI308 AI chips would turn out for China moving ahead, but the important point out of this section alone is the fact that NVIDIA and AMD are back in the region, and while AMD is yet to still with the MI308 accelerators, Team Green is now coming with a much more aggressive option, which is why the interest around the chip is massive, according to earlier reports.
NVIDIA's H200 AI Shipments Now On-Track For Mid-February; AMD Sees Interest In Instinct MI308 From Alibaba As Well
According to a report by Reuters, NVIDIA's H200 AI shipments are expected to begin rolling out by mid-February, following an "extensive review" of license applications for H200 chip sales to China, which will provide insight into the end customers of these AI chips. It is estimated that the first batch could offload up to 40,000 to 80,000 H200 AI chips, and NVIDIA plans to add more capacity to meet the demand from Chinese customers.
We do know that, despite H200 being several years old, it's still heavily sought after in the region, mainly because China has faced challenges in training frontier models on domestic AI chips, and NVIDIA's tech stack is the undisputed choice for training workloads. Similarly, according to a report by MLex, AMD plans to provide the Chinese tech giant Alibaba with up to 50,000 Instinct MI308 AI chips, marking one of the company's largest orders for the region, even during the pre-export control era. This indicates that China is seeking all the computing power it can get.
It would be interesting to see how the Chinese AI market turns out for American chip manufacturers, since the 'floodgates' have now been opened, after, of course, the regulatory process works out, and for NVIDIA, which has ruled out "billions" in chip demand from China, the H200 approval and demand would be a sigh of relief.
About the author: Muhammad Zuhair is a hardware and technology reporter for Wccftech, specializing in the semiconductor industry and the complex interplay between technology, manufacturing, and geopolitics. His coverage focuses on the corporate strategies and technological roadmaps of industry giants like TSMC, NVIDIA, Samsung, and Intel. Zuhair's expertise lies in deconstructing complex topics such as fabrication nodes (e.g., 2nm process), the economic impact of policies like the CHIPS Act, and the strategic development of AI infrastructure from NVIDIA, AMD and Intel.
Follow Wccftech on Google to get more of our news coverage in your feeds.
