Hong Kong Stock Concept Tracking | Major Release from Shanghai! Concerning AI Computing Power, These Subfields are Under Attention (With Related Stocks)

Good news in the computing power sector. On the afternoon of August 11, the Shanghai Municipal Economic and Information Commission issued the "14th Five-Year Plan for the Development of Software and Information Services Industry in Shanghai" (hereinafter referred to as the "Plan"), which mentions that by 2030, Shanghai aims to build its software and information services industry into an economic growth "Beijing Dynamic Power," a "main battlefield" for AI-enabled applications, and a "bridgehead" for participating in global competition. The main goals are as follows: further expansion of the industry's total volume, aiming for an industrial scale of 4 trillion yuan, and exceeding 1.1 trillion yuan in industrial added value. The "Plan" clearly states the acceleration of the digital transformation of the intelligent computing cloud industry. It aims to enrich and diversify intelligent computing cloud service offerings, guiding quality cloud service providers to create lightweight, high-throughput AI inference platforms with "expert-level intelligence," adaptable to the massive demand for industry-specific intelligent applications. It also promotes the construction of a comprehensive public platform for intelligent computing that integrates computing power scheduling, model incubation, and application adaptation. Additionally, it supports cloud service providers in transitioning to a model of "computing power + development tools + model services" to enhance integrated services and software engineering capabilities in cloud-network security. The ecosystem of the intelligent computing cloud industry will be improved, focusing on key sectors such as finance, manufacturing, education, healthcare, culture and tourism, and urban governance, to create a number of benchmark applications for industry-specific models. The market-driven reform of computing power elements will be innovated, continuously exploring new investment methods such as equity in exchange for computing power pricing and a new investment attraction model of "computing power resources + application scenarios." The "Plan" also outlines the implementation of the "Hundred Thousand" intelligent computing cluster project. This will establish a hierarchical supply system that coordinates "large clusters + small clusters + edge computing power," with the aim of building super-large-scale intelligent computing clusters of 100,000 cards in Songjiang, Lingang, and Qingpu, and constructing thousand-card intelligent computing clusters in Baoshan, Pudong, and Jiading. It encourages traditional data centers and information communication rooms to transition to hundred-card edge intelligent computing centers to meet ultra-low latency computing power demands of businesses and individuals. With the rapid development of large AI models, various intelligent applications are continuously emerging, resulting in a significant increase in societal demand for AI computing power. At the same time, another piece of news has also boosted the global computing power industry chain. NVIDIA announced on Monday afternoon local time that it has established strategic partnerships with six Wall Street institutions: Apollo Global Management, Blackstone Group, BlackRock, Brookfield Asset Management, Goldman Sachs, and KKR, to set up an independent computing power financing platform aimed at mobilizing over $500 billion in third-party capital for AI infrastructure development. There is a continuous stream of positive news regarding the domestic computing power industry. Recently, according to CCTV News, the National Development and Reform Commission has reported that this year, Chinas computing power foundation has been further solidified, with the countrys first domestically produced 100,000-card AI super cluster officially launched. This marks that China's computing power infrastructure construction has entered a new stage of 100,000-card level deployment. Many computing power nodes across the country are experiencing a new round of expansion. At the core node of the National Supercomputing Internet in Zhengzhou, the first domestically produced 100,000-card AI super cluster has been put into operation. Staff stated that the uniqueness here lies in the fusion of scientific computing and intelligent computing power, making advanced preparations for potential new types of computing demands in the future. Moreover, the Shanghai Municipal Bureau of Commerce and six other departments recently issued the "Construction Plan for the National Service Trade Innovation Development Demonstration Zone in Shanghai," proposing to promote the overseas use of computing power and foundational large models, support the export of model products such as intelligent agents and multimodal generation, and encourage enterprises and institutions to establish AI research and development centers, open-source platforms, and cooperative laboratories. Promoting the export of AI products and services will not only help unleash domestic AI advantages but also strengthen Chinas voice in the global AI governance landscape. AI companies with international capabilities are expected to benefit from favorable policies. Looking ahead, Guosheng indicated that capital expenditure by North American cloud service providers has yet to peak and is expected to continue to rise. As multimodal large models, embodied intelligence, and AI agents penetrate deeply into the industry, demand for AI computing power is spreading from the "training end" to the "mass inference end," with the demand slope increasing rather than declining. Sealand pointed out that overseas CSPs and Neocloud vendors are gradually proving that large-scale capital expenditures in computing power can effectively convert into revenue and profit. Driven by agents, inference demand has long-term sustainable growth potential. Coupled with the significant increase in prices of high-end computing server channels in China, which reflects the current tight supply situation in the industry, it is anticipated that the tension in Chinas AI computing power supply and demand will escalate, and the logic for medium to long-term growth in the computing power rental industry is relatively certain, having entered a period of performance release benefits. In sub-sectors, CICC noted that the expansion of computing power, combined with electrification transformation, is driving a comprehensive upgrade in thermal management technology, with a compound annual growth rate of 59.8% projected for the global heatsink market from 2025 to 2027. Among them, the market space for data center heatsinks is expected to exceed 315.1 billion yuan by 2027. Relevant concept stocks: QINIU (02567): Benefiting from the shift in the AI agent paradigm and the explosion of model demand driven by the rise of one-person companies, QINIU is undergoing a value reassessment. The company expects its AI-related revenue to reach 437 million yuan in 2025, with its contribution to total revenue increasing from 22.2% in the first half of the year to 24.7%; adjusted EBITDA recorded a profit of 900,000 yuan, marking the first positive result since its listing, achieving a key transition from scaling growth to profit realization. Notably, the increase in AI revenue contribution not only confirms the effectiveness of the companys strategic transition to a "MaaS scheduling hub" but also indicates that its core positioning in the AI ecosystem is accelerating in value realization. As of December 31, 2025, the number of registered platform developers has exceeded 1.8 million, with over 160,000 developers activating AI large model services. TIME INTERCON (01729): CMSC released a report stating that TIME INTERCON (01729) is an important component of the Luxshare system, with exceptional positioning in the MPO optical communication and AI server sectors, expected to achieve high-quality growth. Its automotive business is benefiting from the successful acquisition of Leoni Cables, likely to rise rapidly to a top global automotive cable supplier, while the medical device business has long-term potential in frontier sectors, with collaborative growth expected through the Luxshare system's empowerment. Considering the high demand in the data center and server market and Leoni's better-than-expected profitability, the companys total revenue for 2026/27/28 is projected to be 20.7/27.4/34.1 billion HKD, with net profits of 1.35/2.03/2.49 billion HKD, corresponding to P/E ratios of 27.7/18.5/15.0. Maintain a "strong buy" rating. Z.AI (02513): On August 11, Z.AI's programming product ZCode announced a comprehensive upgrade, with Goal, Subagents, Remote Control, and Idle Tasks launched, pushing the Coding Agent from dialog assistance to autonomous delivery of complex tasks. The number of ZCode users has exceeded 1 million, and to celebrate this milestone, the weekly limit for all GLM Coding Plan users was reset on the same day. Z.AI's self-developed Z.ai Code Bench shows that with GLM-5.2 paired with ZCode, the overall task pass rate was 2.39% higher compared to when paired with Claude Code, with ZCode's strengths concentrated in complex tasks that require cross-file and cross-segment collaboration to achieve final acceptance. ZCode has also effectively increased the GLM Coding Plan limits by about 30% through cache hit rate optimization, combined with a 1.5x limit activity before August 31, resulting in nearly double overall usage. This update indicates that Z.AI (02513) is accelerating the creation of an AI programming ecosystem of "foundational models + Coding Harness + subscription services." ILUVATAR COREX (09903): Recently, international investment bank BofA completed its initial coverage of ILUVATAR COREX (09903), with a target price of 1,013 HKD. The report indicates that ILUVATAR COREX is a leading domestic provider of general-purpose computing GPUs (GPGPU), with two major product lines: Tianpai and Zhikai. The institution is optimistic about the companys development based on three core logical points: firstly, amid the trend of domestic replacement, the company fully benefits from the rapid expansion of the domestic AI acceleration chip market; secondly, with new products landing and continuous expansion into large end customers, the companys market share will steadily increase; and thirdly, a diversified supply chain layout can support sustained revenue growth.