“Yuyuan Tantian”, a Chinese state media, published a commentary on Jensen Huang’s visit to China in the past week via its Weibo account.
The Weibo account of “Yuyuan Tantian,” set up in March 2019, has released many exclusive news items and pictures related to previous US-China closed-door meetings. According to the self-description, “Yuyuan Tantian” is a woman, an “experienced political and economic news reporter” with a PhD in Economics. CCTV has recently admitted that the account belongs to them.
Why did NVIDIA CEO visit China for a week?
Recently, a video has been circulating online. In the video, NVIDIA CEO Jensen Huang, a global chip giant, is seen wearing a Northeast-style vest and waving a large handkerchief in his hand, appearing to be in a good mood. Tan Zhu learned that Jensen Huang attended NVIDIA’s annual meetings in Beijing, Shanghai, and Shenzhen on the 15th, 17th, and 19th, respectively, and the video was shot in Shanghai. On January 20th, Jensen Huang left the mainland and flew to Taipei. Including his itinerary in Taiwan, Jensen Huang’s visit to China this time lasted at least a week. #NVIDIA chips face cold reception in the Chinese market#
Jensen Huang’s last visit to mainland China can be traced back to 2019. As the helm of NVIDIA, what considerations does he have for this visit to China?
Tan Zhu noticed that before Jensen Huang’s visit to China, U.S. media revealed a crucial of information: NVIDIA is facing new challenges in China as Chinese customers do not want its “downgraded” chips.
The U.S. media mentioned that many Chinese cloud computing companies reduced their orders for NVIDIA chips and turned to purchase chips developed by locals such as Huawei.
To understand this change, let’s start with RTX4090 D graphics card that NVIDIA launched for the Chinese market at the end of last December. This “compliant version” of the RTX4090 D graphics card is essentially downgraded version of the RTX 4090 graphics card that has been on the market for over a year.
On October 17th last year, the U.S. Department of Commerce further tightened the rules and two semiconductor control documents, introducing two new indicators, “Total Peak Performance” (TPP) and “Performance Density” (PD), which further restricted the semiconductor exports from U.S. companies to China.
This ban not only precisely targeted NVIDIA’s chip sales to China, but even the RTX4090 graphics card, which is usually used in the gaming industry, unexpectedly fell victim due to exceeding the indicators.
In fact, the U.S. government has been restricting NVIDIA’s exports to China for several years in order to limit China’s development in the field of artificial intelligence.
On October 7th, 2022, the U.S. Department of Commerce suddenly introduced new export control regulations, requiring all “high-performance semiconductors” to be prohibited from being exported to China. As a result, NVIDIA’s flagship chips A100 and H100 were cut off from the Chinese market.
In order to avoid losing the Chinese market, NVIDIA created the A800 and H800, which are lower-performance but compliant alternative models specifically for China.
However, with the introduction of the two aforementioned new documents, even the A800 and H800 became “less high-performance products” that cannot be exported to China. As a result, it was reported that NVIDIA would launch customized versions of “secondary downgraded” chips such as H20, L20, and L2, and export them to the Chinese market.
It can be foreseen that the “cat and mouse game” between NVIDIA and the U.S. government will likely continue in the future.
It should be said that the fundamental purpose of the U.S. government’s repeated tightening of export standards to China is to limit the computing power that Chinese companies can obtain with their advantages in the chip field, thereby reducing the efficiency of Chinese companies in training AI and ultimately establishing technological leadership over China in the field of artificial intelligence.
As the U.S. regulatory agencies emphasized that they will adjust export control policies to China based on the latest trends in the chip industry. Even gaming graphics cards have now been included in the sanctions scope.
Therefore, many Chinese companies believe that their cooperation with NVIDIA has reached a ceiling.
From this perspective, Huang Renxun’s visit to China at this time reflects his concerns about the possibility of losing the Chinese market.
The cold reception of NVIDIA chips in the Chinese market is directly reflected in its revenue. Compared with the 2022 fiscal year, NVIDIA’s revenue in China in the 2023 fiscal year has decreased by 18.4%.
Jensen Huang is well aware of this.
As a professional in the chip and artificial intelligence field, Jensen Huang has recently pointed out multiple times that China cannot be replaced in the global market of the U.S. technology industry, and the various restrictions imposed by the U.S. government will only cause significant damage to U.S. companies and further strengthen China’s determination to achieve independent self-reliance in the semiconductor field. Some analysis institutions have pointed out that if the U.S. continues to implement its current policies rigorously. The market share of some NVIDIA products may drop by as much as 30% in the next five years.
In fact, NVIDIA’s anxiety is not simply about losing the Chinese market itself.
At almost the same time as Jensen Huang’s visit to China, the CEO of OpenAI was in talks with global investors, planning to develop their own chips. Nowadays, the trend of AI companies manufacturing their own chips is becoming a major blow to companies like NVIDIA.
The main problem lies in the fact that various challenges always arise during the development of AI algorithms. When these challenges and data are fed back to NVIDIA, only then can they effectively address them and improve their next-generation products. It is these added values that have formed a mutually beneficial relationship between chip companies and AI companies. Therefore, NVIDIA cannot afford to lose the Chinese market, not only for sales but also for these values that cannot be measured by sales figures.
China’s rapid growth in AI capabilities and its position as a world leader in this field, just like the United States, are well recognized by the world, and NVIDIA is well aware of the stakes. Losing the Chinese market and data would be a double blow to NVIDIA. The U.S. government has always believed that by standing at the top of the chip industry chain, it can easily suppress the development rights of other countries in this field. However, through Jensen Huang’s anxiety, we should update our understanding: having cutting-edge technology does not necessarily mean having the greatest influence in the field.