Hong Kong Stock Concept Tracking | Google's Computing Power Targets Revealed! The domestic AI computing power industry chain is expected to flourish by increasing a thousand times in the next 4-5 years (with concept stocks)

The Zhitong Finance App learned that after the release of the world-stunning next-generation Gemini model last week, the question of how Google can meet the surge in computing power demand has also become the latest focus. According to several people familiar with the matter, Amin Vahdat, the head of Google Cloud's AI infrastructure, stated bluntly at the general meeting that Google must “double its computing power capacity every 6 months,” and that the overall goal for the next 4 to 5 years is to achieve a “1,000 times increase in capacity.”

“Competition for AI infrastructure is the most critical and costly part of the AI competition,” Vahdat said. Google needs 'at least a 1000x increase in cost, energy consumption, and computing power, storage, and internet connectivity while maintaining the same power and energy levels'. It won't be easy, but through collaboration and co-design, we will reach our goal.”

CEO Sundar Pichai further warned at the conference that 2026 will be an “extremely intense year,” and that AI competition, cloud demand, and computing power pressure will simultaneously drive up the company's investment pace. Google's parent company Alphabet has raised its capital expenditure forecast twice this year to $91 billion to $93 billion, and predicts a “significant increase” in investment in 2026. The four major cloud service providers, including Microsoft, Amazon, and Meta, are expected to spend more than $380 billion together this year.

According to reports, under the two-wheel drive of the global wave of artificial intelligence and the accelerated deployment of domestic computing power infrastructure, listed companies in the domestic computing power industry chain will usher in a bright moment in 2025. Looking at the segmented industry chain, AI computing power infrastructure companies such as servers, PCBs, and optical modules have performed well. IFF achieved net profit to mother of 22.487 billion yuan in the first three quarters, an increase of 48.52% over the previous year; Zhongji Xuchuang's profit level ranked second, but the performance growth rate was even more impressive. In the first three quarters, it achieved net profit of 7.132 billion yuan, an increase of 90.05% over the previous year. Companies such as Cambrian, Shengyi Electronics, and Shenghong Technology all achieved a year-on-year increase in net profit to mother. Among them, revenue for the first three quarters of the Cambrian period reached 4.607 billion yuan, a year-on-year increase of 2386.38%; net profit to mother changed from a loss of 724 million yuan in the same period last year to a profit of 1,605 billion yuan.

Demand for AI computing power and hardware infrastructure continues to be strong. The capital expenditure of Google, Microsoft, Meta, and Amazon, the four largest North American cloud vendors in 25Q3 was 96.4 billion US dollars, up 67% year on year, and 9% month on month; the capital expenditure of domestic Internet vendors showed a divergent trend: Ali plans to invest more than 380 billion yuan in AI and cloud infrastructure within three years, and its capital expenditure for fiscal year 2025 has reached 86 billion yuan. Tencent's capital expenditure fell 24% year on year in the third quarter of this year.

Open Source Securities believes that in 2026, with the gradual shipment of Nvidia GB300NVL72, Rubin and RubinCPX series cabinets, the increase in shipments of the latest computing power chips from ASIC giants such as Google and Meta, as well as large-scale shipments of domestic computing power chips and the gradual launch of domestic supernodes, the liquid cooling penetration rate may increase dramatically; under the joint impetus of national policies and domestic giants, the domestic AI computing power industry chain is expected to flourish in 2026. It is recommended to focus on the five major segments of domestic computing power chips, computing equipment, network equipment, AIDC, and cloud platforms.

Related concept stocks:

Alibaba-SW (09988): The Panjiu 128 supernode AI server released by Alibaba Cloud at the 2025 Yunqi Conference supports 128 AI computing chips in a single cabinet, reaching the highest density in the industry. Through a self-developed CIPU 2.0 chip and a high-performance network card, the server increases inference performance by 50% under the same computing power, and supports 100,000 card cluster interconnection.

Baidu Group-SW (09888): Baidu released two new AI chips, Kunlun Core M100 and Kunlun Core M300, as well as two new supernodes, Baidu Tianchi 256 supernode and Baidu Tianchi 512 supernode, and announced Kunlun's roadmap for the next five years.

Shangtang-W (00020): At the end of October, Lyon released a research report saying that Shangtang has established a strategic partnership with Cambrian, a move that will help it expand its artificial intelligence computing capabilities through domestic chips.

ZTE (00763): The company focuses on layout and investment in the field of artificial intelligence. With leading software and hardware digital intelligence base capabilities, the company provides end-to-end intelligent computing solutions, including intelligent computing infrastructure, AI software platforms, large models and applications, to meet the needs of different scenarios and help enterprises transform digitally.