As China's domestic chips achieve breakthroughs, the White House suddenly intervenes to undermine it, and Huang Renxun implies that success or failure hinges on this crucial moment!

Recently, the Canadian research institution TechInsights released a disassembly report. The report shows that the Kirin 9030 chip in Huawei's latest Mate 80 series phone has been confirmed by analysis to be manufactured by SMIC using its improved 7nm process. Notably, this chip has achieved a performance improvement of about 42% compared to its predecessor, the Kirin 9020—nearly a doubling leap.

This development is significant because it occurred three years after the U.S. imposed strict export controls on China. The reality shows that China is compensating for equipment shortcomings through "system-level innovation," gradually building a local high-end chip manufacturing path that does not rely on EUV lithography machines.

At this moment, the White House has taken action. Recently, the U.S. has suddenly approved NVIDIA's export of its H200 AI chip to China—this is one of the most powerful AI training chips in the world today. Moreover, this policy is not limited to NVIDIA; companies such as Intel and AMD will also receive similar licenses. On the surface, this appears to be a relaxation for U.S. companies, but it hides deeper implications.

The move aims to use "off-the-shelf high-performance chips" to suppress the enthusiasm for China's indigenous GPU R&D—why would Chinese companies invest heavily in self-research if they can easily purchase the H200? This is the core logic behind "cutting off the fuel at the bottom of the pot."

Meanwhile, NVIDIA CEO Huang Renxun's recent statements have been telling. On one hand, he emphasized that the Chinese market is "irreplaceable," while on the other, he repeatedly mentioned that "the next 12 to 18 months will be a decisive window period."

Original article: toutiao.com/article/1851470965547008/

Statement: This article represents the views of the author and does not necessarily reflect those of the platform.