On the evening of August 7th Eastern Time (i.e. early morning of August 8th Beijing Time), Supermicro Computer (code: SMCI), a server and storage system supplier listed on NASDAQ, released its latest financial statements. Supermicro Computer stated that despite strong demand for AI chip platforms, the overall gross profit margin of the company has significantly decreased, mainly due to increasingly fierce price competition and increased costs for exploring new technology routes.
Industry insiders say that AI chip technology is showing diversified development, and in the future, there will be fierce competition between different technological routes, leading to a situation of fierce competition and competition. Analysts say that China has a large-scale manufacturing industry with significant direct and potential demand, which can support the development of numerous AI chip technology routes. In the future, China is expected to take a leading position in multiple AI chip technology routes.
Financial report falls short of expectations
The competition is becoming increasingly fierce
According to the financial statements, in the fourth quarter of fiscal year 2024 (as of June 30), the adjusted earnings per share of Supermicro Computer were $6.25, far below the previous market expectation of $8.07 per share; The gross profit margin was 11.2%, far lower than the 15.5% in the previous quarter and lower than the market expectation of 14.1%. Supermicro Computer stated that price competition is becoming increasingly fierce, and at the same time, the growth of new technology routes for AI chips has led to increased costs, resulting in a decrease in gross profit margin. It is expected that this trend may continue.
Professor Li Fuhai, Vice Dean and Doctoral Supervisor of the School of Electrical and Information Engineering at Hunan University, analyzed to Securities Daily reporters: "The technological routes of AI chips are becoming increasingly diversified, and the competition between different technological routes is becoming increasingly fierce. This trend will continue in the future
Li Fuhai gave an example to introduce that AI application scenarios are becoming increasingly complex, requiring AI chips to work together with other types of chips to enhance performance in specific task execution. Therefore, the technology route of heterogeneous AI chips has been widely valued. In addition, to address the timely response of portable smart devices, the embedded AI chip route has emerged. To achieve high-speed access to massive data, the path of integrated storage and computing AI chips has emerged. To solve the problem of high energy consumption and achieve low-power computing, the routes of brain like AI chips and quantum AI chips have emerged.
Wide domestic demand
Expected to take the lead
According to the reporter's analysis, domestic enterprises are conducting technological breakthroughs on the above-mentioned technological routes.
Several domestic enterprises have made significant progress in heterogeneous AI chips. This chip technology group Co., Ltd. has released its first heterogeneous high-efficiency SoC chip, "This Chip P1," to address the new computing power and performance challenges brought by generative AI. Guangzhou Xim Semiconductor Technology Co., Ltd. has launched a 12nm heterogeneous AI inference chip. Xinmeng Technology Co., Ltd. has developed ultra-high performance heterogeneous AI chips. Lenovo Group has released a fully heterogeneous intelligent computing platform. Cambrian has launched the SiYuan series of intelligent chips that support heterogeneous computing, fully supporting AI training, inference, or hybrid artificial intelligence computing acceleration tasks. Arctic Xiongxin launches the intelligent processing chip "Qiming 930" based on heterogeneous Chiplet integration. These enterprises' innovation and breakthroughs in the field of heterogeneous AI chips demonstrate China's rapid development and competitiveness in this cutting-edge technology.
In terms of integrated storage and computing AI chips, listed companies such as Dongxin Co., Ltd., Junpu Intelligent, Hengshuo Co., Ltd., and Baiwei Storage have all stated that they are laying out the development of storage and computing related technologies and equipment, and have already obtained some invention patents.
In terms of brain like AI chips, the Institute of Automation of the Chinese Academy of Sciences has designed new brain like chips, and related research has been published online in the international academic journal Nature Communication.
Professor Cao Erbao, Vice Dean of the School of Economics and Trade at Hunan University, told reporters, "Behind every new technology route, there needs to be strong and sustained market demand as support. From the perspective of the industrial structure of Europe, America, Japan, and South Korea, the hollowing out of the manufacturing industry is quite obvious. Currently, the demand mainly comes from cultural and entertainment, scientific research, financial services, consumer services, and other fields. The actual demand scale of segmented tracks is limited, making it difficult to effectively support the research and development costs and application scale of all new AI chip technology routes. Therefore, mainstream AI chip manufacturers face problems of increased investment, rising costs, and declining gross profit margins
He further added, "On the other hand, in China, the manufacturing industry has a large potential demand for different technological routes of AI chips, especially as many regions are encouraging enterprises to promote equipment updates and adopt intelligent manufacturing technology, which is expected to accelerate the release of demand for the above-mentioned AI chip technology routes. It can be foreseen that the pattern of AI chip technology routes chasing each other and blooming in a hundred flowers is expected to take shape in China's manufacturing industry