According to a Reuters report dated September 16, Nvidia's RTX 6000D chip, which was tailor-made for the Chinese market, has encountered a significant hurdle. Major tech players in China, such as Alibaba, Tencent, and ByteDance, have reportedly declined to place orders for this product. Designed primarily for AI inference tasks, the RTX 6000D's performance, as revealed by sample tests, falls short when compared to the RTX 5090. Notably, the RTX 5090 is prohibited from being sold in the U.S. by the American government. Furthermore, the RTX 6000D comes with a hefty price tag of up to 50,000 yuan, which is double the cost of the RTX 5090 available on the gray market.

In an effort to adhere to U.S. export restrictions, Nvidia has intentionally reduced the memory bandwidth of the RTX 6000D. This modification has led to a processing speed that is 37% slower than that of the RTX 5090. Additionally, the RTX 6000D is priced 40% higher than international models in the same category. Moreover, Nvidia is mandated to pay the U.S. government a 15% share of its sales revenue generated in China, a move that has drawn criticism from Chinese companies.

One prominent cloud computing giant has already redirected 20% of its AI chip orders to Huawei's Ascend 910C. On September 15, the State Administration for Market Regulation announced the initiation of an antitrust investigation into Nvidia. If violations are found, the company could face a hefty fine of approximately $5 billion.

Huawei's Ascend 910C has demonstrated impressive capabilities, achieving 85% of the computational efficiency of Nvidia's Hopper architecture while consuming 23% less energy. This development poses a direct threat to Nvidia's market standing. Nvidia's share of the AI chip market in China has seen a decline, dropping from 66% in 2024 to 54% in 2025. Meanwhile, the self-sufficiency rate of domestic chips is projected to reach 55% by 2027.