Nvidia Readmitted To China With Conditional License

The US government has granted Nvidia Corp. a license to ship a small number of its less advanced H200 chips to customers in China, marking an important step forward for the leading maker of artificial intelligence accelerators as it seeks to re-enter the world’s largest semiconductor market.

According to Bloomberg Terminal, the US Department of Commerce granted Nvidia a license for H200 shipments, subject to inspection in the US and a 25% duty on the value of the exported chips. This decision follows months of uncertainty and tension between the two countries over China’s efforts to build its own domestic semiconductor industry.

The approval is significant, as it allows Nvidia to resume sales of the H200 chip, which was previously banned from export to China by President Donald Trump. The ban was put in place in April 2025, following concerns that Chinese companies were using the chips for military applications. However, under pressure from Chinese tech giants and other stakeholders, Trump reversed the decision just months later.

Despite this development, it remains unclear whether Beijing will allow even a limited return of H200 shipments to China. The Chinese government has repeatedly criticized US export controls as “malicious” and has accused Washington of using technology and trade to contain and suppress China’s growth.

In reality, China is facing a complex dilemma in balancing its need to support local chipmaking initiatives with its desire to maintain access to advanced technologies like the H200 chip. The country claims that its domestic semiconductor industry can already match the capabilities of Western companies like Nvidia, but it also acknowledges that Chinese firms are struggling to develop high-performance chips on their own.

To address this challenge, Beijing has been urging its tech giants to prioritize local production and reduce reliance on imports. However, many of these companies have expressed frustration with the lack of progress in building up domestic capabilities and have instead turned to imports as a way to keep pace with global rivals.

One solution proposed by Chinese officials is to force companies importing foreign chips to purchase a ratio of locally built processors. This approach has been suggested as a way to encourage domestic production while allowing for the use of more powerful H200 chips for training and other applications.

However, without an official announcement from Beijing, Chinese tech companies have no choice but to wait and see whether they can purchase Nvidia GPUs. The uncertainty is adding pressure on the US government, which must balance its own national security concerns with the need to maintain a positive relationship with China.

The recent approval of Nvidia’s H200 shipment license highlights the complexities of this trade-off. On one hand, the US sees export controls as an essential tool for protecting national security and preventing the proliferation of advanced technologies that could be used for malicious purposes. On the other hand, Washington recognizes that its policies are having unintended consequences, including driving up costs and creating uncertainty in global markets.

As the situation continues to unfold, it is clear that the US-China tech war has entered a new phase, marked by selective access, higher costs, and slower trust. The future of Nvidia’s business in China remains uncertain, but one thing is clear: the outcome will have far-reaching implications for the global semiconductor industry and beyond.

In recent months, there have been numerous reports on how Chinese companies are smuggling NVIDIA GPUs into the country despite US export restrictions. According to a report by The Financial Times (FT), at least $1 billion worth of NVIDIA GPUs were smuggled to China after US President Donald Trump took office and tightened export controls. These chips are highly sought-after by AI companies across various hardware systems, including OpenAI, Google, Meta, Microsoft, and others.

NVIDIA’s B200 is a GPU based on the company’s high-end Blackwell architecture, which is widely used in AI applications. The direct sales of this chip are prohibited in China, as the US government has placed harsh restrictions on companies exporting advanced AI hardware.

The FT report revealed that a Chinese company claimed to have over 100 business partners, including AliCloud, ByteDance’s Huoshan Cloud, and Baidu Cloud. Additionally, it showed that most of the B200 racks sold by this company originated from Supermicro, a US-based company. The packaging and installation pictures and videos seen by the FT show product logos of companies such as Supermicro, Dell, and Asus.

Besides Hopper GPUs, which are a generation of chips preceding the Blackwell architecture, Chinese companies are still lining up to get their hands on these relatively powerful chips that domestic chipmakers struggle to match. The Chinese government claims that its homegrown semiconductors can now match Nvidia’s defanged H20 and RTX Pro 6000D GPUs.

However, these claims remain untested as Beijing has been warning local tech firms to reduce reliance on imports and prioritize domestic production. Companies like Alibaba Group, Tencent, and ByteDance are now buying H200s only for critical projects, while limited supply has pushed prices up sharply on the grey market.

The recent decision by US chipmakers Nvidia and AMD to give 15% of their revenue from advanced chips sold to China in return for export licenses highlights a new phase of the US-China tech war. This arrangement signals an easing in the US-China trade war, but it also raises concerns about the credibility of export controls and the trust of other nations.

In June, US businesses took on about 64% of the costs of tariffs imposed by Trump on various nations’ exports, while consumers bore 22%. However, analysts expect that consumers will soon be bearing the brunt of these costs, with their share expected to increase to 67%.

The recent approval of Nvidia’s H200 shipment license marks an important step forward for the company as it seeks to re-enter the world’s largest semiconductor market. While it remains unclear whether Beijing will allow even a limited return of H200 shipments to China, the development highlights the complexities of this trade-off.

For the US government, allowing Nvidia to export its H200 chips to China is an essential tool for protecting national security and preventing the proliferation of advanced technologies that could be used for malicious purposes. However, Washington recognizes that its policies are having unintended consequences, including driving up costs and creating uncertainty in global markets.

In conclusion, the recent approval of Nvidia’s H200 shipment license highlights the complexities of the US-China tech war. The future of Nvidia’s business in China remains uncertain, but one thing is clear: the outcome will have far-reaching implications for the global semiconductor industry and beyond.

Original Source

• Read the full article here