Another domestic chip company develops advanced AI chips to accelerate the replacement of NVIDIA
Nvidia has taken advantage of the east wind of AI chips to make great progress, with total assets reaching 17 trillion yuan, but in order to prevent China from developing artificial intelligence, the United States has repeatedly banned Nvidia from producing advanced AI chips for China, but Chinese chip companies are now also struggling with Nvidia's AI chips, and some time ago a Chinese semiconductor company also came out with good news.
Haiguang Semiconductor Co., Ltd., which is Haiguang Semiconductor Co., Ltd., acquired AMD's patents in the early days and developed a high-performance server chip that can be compatible with the current x86 system, occupying most of the domestic server chip market and becoming an alternative solution for domestic manufacturers.
Haiguang Technology also attaches great importance to it, in addition to focusing on server chips, it also focuses on the development of AI chips, and in recent years, the company's research on AI chips has been gradually strengthened, and the research in the field of AI has been comparable to NVIDIA. It is said that the Haiguang No. 1 DCU is only about 50% of the high-end products of the Nvidia A100, and now, the good news from the new Haiguang is that the efficiency of its second-generation DCU reaches 80% of that of NVIDIA. Eighty percent of Nvidia's shares.
Compared with domestic chip manufacturers, Haiguang chips have the advantage of being fully compatible with Intel and AMD server chips, providing a good choice for domestic chip manufacturers, thereby greatly improving the processing ability of domestic chips for AI big data.
This is also what AMD and Intel, the two major server manufacturers, have come up with their own powerful AI chips, wanting to seize a piece of their own territory, and Nvidia has surpassed TSMC and Intel with the rapid expansion of AI chips, sitting in the first position in the world.
However, Chinese semiconductor companies such as Haiguang Semiconductor are facing similar "environmental problems" as Intel and AMD, coupled with the fact that NVIDIA has been committed to CUDA technology in the field of AI chips for more than ten years, forming a relatively complete CUDA ecosystem, so the ecological environment of CUDA is a huge obstacle for other companies.
Because of CUDA, Nvidia launched an A800 processor called H800, although its performance has not been reduced, but Chinese manufacturers also have to buy, Nvidia has reduced the H800 processor by 80%, which makes Chinese companies unbearable, have used domestic artificial intelligence chips, it is said that their artificial intelligence chips, has reached 5000 units. If that's the case, that's more than half a billion profits.
Many Chinese AI chip companies, including Haiguang, have developed AI chips, and their performance is definitely stronger than that of H20 provided by Nvidia in China, and the cost performance between the two has made many local companies explore domestic AI chips, while the AI chips produced by Haiguang can be seamlessly connected with existing data centers, thus having a certain impact on NVIDIA's CUDA ecosystem. Local AI chips will replace Nvidia's products.
It is said that after seeing the rise of domestic AI chips, Nvidia also felt a lot of pressure and no longer insisted on reducing the price when buying H20 in China, which also proves that American semiconductor manufacturers must meet the requirements of Chinese companies through fierce competition, otherwise Nvidia's development speed will be very fast.
The U.S. chip industry has always been arrogant, the United States has cut prices sharply because of its monopoly position in the past two years, but since last year, many local chips have been replaced by local manufacturers, and its price is once again ninety percent, and now, AI will also repeat the rapid development of China's chip industry, but it also tells the U.S. chip industry that it is no longer the only one, and China's chip industry is absolutely competent. The choice of chips is rapidly evolving.