Institutions flock to investigate the AI industry chain; upstream companies have abundant orders, and their performance is gradually being realized

Securities Daily reporter Wu Xiaolu

Since this year, the AI (artificial intelligence) sector has remained hot. Major institutions have also been highly focused on companies along the AI industry chain. Wind data shows that, as of July 9, since the second quarter this year, 1,821 A-share listed companies have received institutional research visits. Among them, companies in the AI infrastructure segment have become key targets of institutional attention. Coupled with the outlook for 2026 interim report performance, AI upstream companies have full orders and better earnings fulfillment.

“Currently, the AI industry chain remains in a high-activity cycle, and in the long term it will still be the market’s core investment main line. But due to asset prices and valuation factors, the market has recently shown differentiation,” Li Qiusuo, chief domestic strategy analyst at CICC Research Department, told Securities Daily reporter Wu Xiaolu. After the AI industry chain achieved important breakthroughs in commercialization, the sector’s valuations and growth potential have been revised upward based on fundamental expectations. In the future, investors need to screen using valuations and how tight supply-demand conditions are, among other factors.

Upstream core companies have orders that are well-filled

Currently, China’s AI daily Token (term) usage has already reached several hundred billion, with growth of over a thousand times in two years. Among the 1,821 companies that received institutional research visits in the second quarter, 13 companies received more than 300 institutional visits each. Among them, three companies—Lanzhou Zhenxing Technology Co., Ltd., Huaqin Technology Co., Ltd., and Zhejiang Crystal Optoelectronics Technology Co., Ltd.—each received more than 500 institutional visits. Of these 13 companies, 11 are AI industry-chain enterprises.

Benefiting from the sustained surge in AI compute demand, orders have been sufficient for sub-segments such as silicon wafers, high-speed connectors, PCB (printed circuit boards) equipment, and data center supporting equipment. For example, during an institutional research visit in early July, Hangzhou Li An Microelectronics Co., Ltd. said that the overall shipment scale of its silicon wafer products has remained at a high level and order demand is strong.

Dongguan Dingtong Precision Technology Co., Ltd., a core supplier of high-speed connectors, said in May when it received institutional research visits that, benefited by the爆发 of AI compute demand, the order growth rate for its relevant products has increased significantly.

As core production equipment for AI compute hardware, PCB equipment has seen the industry’s business conditions continue to rise. Kunshan Swanway Technology Co., Ltd. said that in recent years, the proportion of its overseas orders has been gradually increasing. The downstream trend of AI compute demand is clear, and both domestic and overseas orders for high-end equipment have been sufficient, with phased growth receiving good support.

Interim report previews are coming thick with positive surprises

According to Wind data, as of July 9, 171 listed companies have disclosed their 2026 first-half performance previews, with more than 10 AI upstream companies showing positive performance.

For example, Zhuhai Guangke Technology Co., Ltd. expects that the net profit attributable to shareholders in the first half of 2026 will be between 140 million yuan and 150 million yuan, up 170% to 190%. For the reasons behind the change, the company said it is benefiting from the acceleration of global AI compute infrastructure investment and data center construction. Demand for high-speed optical modules and optical components has continued to grow. At the same time, the company has continued to launch new products, actively expanded new customers both domestically and internationally, and operating revenue has achieved steady growth.

In addition, in the first half of this year, the scaling and deployment of edge AI accelerated, pulling a sharp growth in the business performance of edge AI chip manufacturers. For example, Rockchip Electronics Co., Ltd. expects that net profit in the first half will be between 850 million yuan and 910 million yuan, up 60.03% to 71.33%. The company said the main reason is the accelerated scaling and deployment of edge AI (AIoT 2.0). It has leveraged its advantages in the AIoT platform layout and customer ecosystem, and its chip platform has seen many products blossom across multiple product lines.

In early July, Meta planned to rent out surplus AI compute capacity to the outside world, increasing market concerns that the expansion pace of compute capacity by overseas cloud providers may have topped out, and volatility in the AI sector intensified. Recently, updated views released by domestic and overseas institutions believe this is not a signal of industry oversupply, and they recognize the long-term main-line position of AI.

He Li, chief investment adviser at Guotai Junan Securities, said in an interview with Securities Daily reporter Wu Xiaolu that there are differences between international and domestic compute infrastructure. Judging from per-capita compute indicators, China still needs more investment to catch up with international advanced levels. At present, the fundamentals of AI midstream and upstream companies have basically good expectations. But because the current trading concentration of the AI sector is relatively high, the market has been showing fluctuations. Investors need to closely monitor how AI upstream infrastructure-related companies will fulfill their performance expectations in their interim reports and in the future.