Hong Kong Stock Concept Tracking | 2026 Artificial Intelligence Industry Conference to be held to focus on AI application-side and infrastructure-side opportunities (with concept stocks)

The Zhitong Finance App learned that from October 14 to 16, the 2026 Artificial Intelligence Industry Conference will be held in Jinan, Shandong. The conference was co-hosted by the China Academy of Information and Communication Technology, the Shandong Provincial Department of Industry and Information Technology, and the Jinan Municipal People's Government. With the theme of “Accelerating Artificial Intelligence Empowers New Industrialization”, the conference aimed to create a comprehensive industrial platform integrating high-end exchanges, publication of results, and scenario docking in the field of artificial intelligence in China.

The conference set up the opening ceremony and main forum, and arranged more than 20 special forums and industry matchmaking sessions. The event will invite Nobel laureates, academicians from both academies and other top industry experts to take the stage to share, focus on publishing a number of major research results from national ministries, departments, and scientific research institutions, and present the “industry-scenario-subject” trinity development pattern formed by China's artificial intelligence industry in three dimensions.

The Smart Device Innovation Forum, the Embodied Intelligence Industry Forum, the End-side Intelligence Forum, the AI Data Set and Data Elements Forum, and the AI Native Application Forum will be launched one after another. The forum aimed at industry bottlenecks such as iterative optimization of large models, engineering implementation of humanoid robots, lightweight end-side models, and high-quality training data supply. It brought together technical leaders from top domestic research institutes and leading technology companies to discuss basic technology innovation routes in depth, exchange progress in cutting-edge prototype product development, and explore independent innovation paths for underlying artificial intelligence technology.

In addition to the thematic forums, the conference simultaneously hosted a number of high-standard special supporting events. Special events such as the “Artificial Intelligence Standard Tour” of the Ministry of Industry and Information Technology's Artificial Intelligence Standardization Technical Committee and the IEEE Trusted Intelligence and Digital Trust Side Event will be held simultaneously.

According to reports, in the computing power infrastructure section of this conference, Huawei, Inspur, Shuguang, and the three major operators of Mobile, Unicom, and Telecom will appear on the same stage.

Relying on the new Shengteng computing ecosystem, Huawei launched the Shengteng 960 supernode, equipped with the world's first NPO light engine, which can be expanded to 4096 cards. The peak computing power of a single cluster can reach 8EFLOPS FP8. The multi-cluster interconnection can support million-level card scale training, fully adapts to the native training needs of large models with large parameters, and continues to build an open, open, autonomous and controllable national computing power ecosystem.

Wave, on the other hand, is focusing on the decline in computing power in industrial scenarios to create an “AI factory” implementation model. Relying on the national artificial intelligence application pilot base, it has completed more than 300 model training and implemented nearly 1,000 industrial agents, covering 41 major industrial categories, creating 15 large industrial domain models and 43 industry agent matrices. At the same time, through extreme power consumption optimization, edge computing power consumption is reduced to 20 watts, enabling industries such as mechatronic pumps with 45 TOPS high-efficiency computing power, helping enterprises increase production efficiency by an average of 30%, and truly move computing power from computer rooms to production lines.

The three major mobile, Unicom, and telecom operators focus on global computing power network scheduling, opening up channels for computing power interconnection and efficient circulation, and building an inclusive computing power service system covering the entire region.

In addition, companies such as iFLYTEK, Nvidia, Tencent, NetEase, Yushu, Keling, BYD, and Siemens will also be showcased, with a lineup covering all industries, chains, and scenarios. Among them, Nvidia drastically reduced AI training costs with the Vera Rubin next-generation platform. Compared with previous generation products, MoE model training GPU usage was reduced by three-quarters, and the cost of single token inference was reduced to one-tenth of the original, breaking the core economic barriers of AI industry implementation with extreme cost advantages, making large-scale commercialization of industrial AI possible.

Guoyuan Securities said it will continue to pay attention to opportunities on the AI application side and infrastructure side: on the application side, refer to areas where the business model has been initially verified, focus on directions with similar implementation potential and select listed companies that can substantially benefit; on the infrastructure side, track where the gap between supply and demand continues to widen, and lay out listed companies with excellent production capacity climbing and delivery capabilities in related areas.

Related concept stocks:

Inspur Digital Enterprise (00596): Recently, International Data Corporation (IDC) released the “IDC MarketScape: China Enterprise Resource Management Market AI Agent 2026 Vendor Assessment” report. Inspur Digital is among the market leaders. With its leading AI native technology architecture, innovative product solutions and rich implementation practices, it has become the first choice for enterprise customers. Inspur Digital has always adhered to the “1321" strategy with AI native as the main line, and built Inspur Haiyue's AI intelligent native product system, driven by business entities, large models, and agents, covering the four core levels of model and data, digital intelligence platform, intelligent application, and intelligent interaction, to help enterprises leap from “digital intelligent operation” to “autonomous operation”.

China Mobile (00941): Mobile Cloud has completed compatibility with mainstream domestic AI chips such as Shengteng, Pingtou, Kunlun Core, Moore Thread, Suiyuan, Wall, Haiguang, Mu Xi, and Tianshu Smart Chip, and has achieved unified access and management of multiple computing power through a self-developed intelligent computing control platform. The platform can monitor and actively check hundreds of indicators such as computing power, network, and storage in real time to ensure the stable operation of training and promotion tasks.

China Telecom (00728): China Telecom is fully embracing artificial intelligence, thoroughly implementing the “Cloud to Digital to Smart” strategy, making technological innovation the core of enterprise development and accelerating transformation into a technology-based enterprise. In 2025, China Telecom invested 3.8 billion US dollars in R&D, with 58,000 R&D personnel and 17,000 patents; Tianyi Cloud's revenue exceeded 14.5 billion US dollars, becoming the world's largest operator cloud service provider and China's largest hybrid cloud service provider; independently developed the basic Star Model and pioneered open source among global operators; and intelligent computing revenue increased 95% in the first half of 2026.

China Unicom (00762): On March 31, China Unicom officially announced the company's future cloud computing development strategy, fully opening a new era of cloud services. China Unicom simultaneously released new cloud computing products and initiated the establishment of the “China Unicom Wolyun+ Cloud Ecosystem Alliance”. China Unicom is the only self-developed cloud computing service provider with independent intellectual property rights among domestic operators, and is currently implementing an “M+1+N” resource layout covering the whole country. After all of the above data centers are completed, the total number of racks will exceed 320,000, the total bandwidth will exceed 30T, and it has a carrying capacity of 4 million servers.