Western Securities: The upward AI boom cycle is optimistic about domestic computing power opportunities

The Zhitong Finance App learned that Western Securities released a research report saying that this round of AI and storage boom cycles resonate, along with the continuous improvement of the technical strength of domestic manufacturers, the process of localizing computing power and memory chips is expected to be further accelerated. The capital expenditure for domestic computing power and storage will be transmitted upward along the industrial chain, which in turn will drive the growth of the foundry and semiconductor equipment industries, and is optimistic about domestic computing power, foundry, and semiconductor equipment investment opportunities.

The main views of Western Securities are as follows:

Demand for AI is booming, and domestic computing power infrastructure continues to expand, opening up a broad market space for domestic chips

The 2026 government work report first proposed “implementing new infrastructure projects such as hyperscale intelligent computing clusters and telecommunication collaboration”. Computing power grids are juxtaposed with power grids and new communication networks, forming the “three trump cards” of new infrastructure. Computing power has become a national strategic resource along with water and electricity. Vice Minister of Industry and Information Technology Zhang Yunming said at the press conference of the State Information Office that by the end of March 2026, the scale of China's intelligent computing power had reached 1,882 EFLOPS (FP16), and the development of the computing power industry showed a good trend. On May 26, 2026, the China Information Security Assessment Center and the National Confidential Technology Evaluation Center incorporated AI chips into the safe and reliable certification system for the first time. Nine domestic artificial intelligence training inference chips from manufacturers such as Huawei HiSilicon, Ali Pingtou, Bizao Technology, Haiguang Information, Tianshu Zhixin, Mu Xi, and Moore Thread were rated at the highest level I, which is expected to further open up market space for domestic chips in key fields of government and key fields.

Changxin Technology is about to go public. Changjiang Storage's listing process is progressing steadily, and the pace of domestic storage expansion is speeding up

Changxin Technology has disclosed the Science and Technology Innovation Board listing prospectus and the “Issuance Arrangements and Preliminary Enquiry Notice”, officially launched the Science and Technology Innovation Board IPO issuance process, and is scheduled to begin the IPO subscription on July 16; Changjiang Storage has disclosed the first IPO Counseling Report, and the listing process is progressing steadily. According to Changxin Technology's prospectus, the company plans to use the raised capital of 29.5 billion yuan to invest in memory wafer manufacturing mass production line technology upgrading projects, DRAM memory technology upgrade projects, and dynamic random access memory forward-looking technology research and development projects respectively to achieve technology upgrades and production capacity expansion. Since the second half of 2025, due to factors such as the increase in DRAM demand brought about by the development of artificial intelligence and production capacity allocation from major global manufacturers, the global supply of DRAM products has been in short supply, driving up DRAM product prices. This is one of the main reasons for Changxin's performance growth since the second half of 2025.

Recommended to follow

1) Domestic computing power: Cambrian (covered), Haiguang Information (covered), Shengke Communications (covered); 2) Foundry: SMIC (covered), Huahong Hongli (covered); 3) Semiconductor equipment: Beifang Huachuang, Zhongwei, Tuojing Technology, Huahai Qingke, Shengmei Shanghai, Xinyuan, Changchuan Technology, Precision Electronics, Zhongke (covered).

Risk warning: AI technology breakthroughs fall short of expected risks, application implementation falls short of expectations, and industry competition exacerbates risks.