With the rapid development of artificial intelligence technology, the demand for AI chips is growing exponentially, and China, as the world's second largest economy, has increasingly prominent potential and vitality in its AI market. However, the recent severe challenges faced by domestic tech giants such as Huawei in chip supply have unexpectedly opened the door to new opportunities for international companies such as NVIDIA in the Chinese market. In particular, NVIDIA's "China Special AI Accelerator" is quickly becoming the focus of market attention, and its popularity continues to rise.
Faced with the U.S. government's export restrictions on high-tech products, Nvidia cannot directly sell its top AI chips such as H100 and A100 to Chinese mainland. In order to fill this market gap, Nvidia cleverly launched castrated versions of H800 and A800 chips, and plans to further launch "China-specific AI chips" that are more suitable for China's policy environment, such as H20, L20 and L2 models. Although these special chips have reduced performance compared to the original version, they accurately meet the urgent demand for AI chips in the Chinese market and have become a popular choice in the market. China's largest Internet companies, such as Baidu, Alibaba, and Tencent, have ordered billions of dollars worth of A800 processors and GPUs from Nvidia.
Huawei, as a technology giant in China and even in the world, has its chip business in trouble due to US sanctions. The limited supply of self-developed chips such as Kirin has severely restricted Huawei's development in the AI field. This situation not only puts Huawei itself under tremendous pressure, but also leaves a huge market space for China's AI market. The emergence of NVIDIA's special AI accelerator fills this gap and provides a new choice for enterprises and scientific research institutions that originally relied on Huawei chips.
Figure: Nvidia China's special chip (Source: Wccftech)
The reason why NVIDIA's special AI accelerator has been able to win market recognition in a short period of time is not only due to its accurate market positioning, but also due to its dual advantages in performance and price. While these specialty chips can't match the performance of the original chips, they still meet the needs of most application scenarios at a more affordable price. This cost-effective advantage makes the special chip stand out in the highly competitive market and win the favor of many customers.
Despite the success of Nvidia's AI accelerator in the Chinese market, the Chinese AI industry is also actively promoting the development of domestic AI chips. Domestic companies such as Baidu, Alibaba, Tencent and Huawei have accelerated the pace of developing their own AI chips and are conducting trials on their respective cloud platforms. These companies are starting with the underlying architecture design of chips, trying to seize market opportunities through innovation.
Domestic AI chips perform well in large model inference, but there is still a gap between them and NVIDIA's general-purpose GPUs in terms of large model training. Gao Wen, an academician of the Chinese Academy of Engineering, compared computing power to electricity, emphasizing its importance in the development of the digital economy. Zhao Lidong, founder of Suiyuan Technology, believes that the demand for large model computing power will further promote the research and development of AI chips, and domestic AI chips need to achieve breakthroughs in performance, bandwidth, storage and cluster interconnection.
