Hong Kong Stock Concept Tracker | AICC2026 Artificial Intelligence Computing Conference Will Be Held; Institutions Suggest Continued Focus on Computing Power Services, Domestic Chips, etc. (with Concept Stocks)

AI and computing power sectors have seen numerous positive developments recently. The 7th Artificial Intelligence Computing Conference (AICC, AI Computing Conference) will be held on September 21, 2026, at the Beijing Centergate Technologies International Innovation Center. Since the beginning of this year, agents have been accelerating toward explosive growth globally and in China, rapidly entering production, scientific research, and social services, and the enormous effect of artificial intelligence transforming into real productivity is emerging. At the same time, the computing power supporting AI operations is rising to become national strategic infrastructure. The conference will focus on the computing industry transformation brought by agents, bringing together different technology routes and industry forces in an open and diverse manner, jointly exploring a development path suitable for China's AI computing industry, and promoting intelligence from a technical capability possessed by a few institutions to a basic capability available and universally accessible to everyone. The 2026 Yunqi Conference will be held in Hangzhou from September 22 to 24. With the theme of "Intelligence for Practical Use" and with Agentic AI at its core, this year's conference will feature three main forums, more than 120 sub-forums, and over 1,000 exhibitors. It will focus on chips, cloud, models, Agents, and other links to present Alibaba Cloud's full-stack AI technology and release Alibaba Cloud's core technologies and annual products. In addition, on September 23, the 5th GMIF Innovation Summit, co-hosted by the Shenzhen Memory Industry Association and the School of Integrated Circuits at Peking University, will officially open at the Renaissance Shenzhen Bay Hotel. With the theme of "AI Storage, Driving the Future," this year's summit is based on the entire storage industry chain and takes the Token economy as a key perspective, gathering leading enterprises, top technical experts, and industry leaders from the upstream and downstream of the global storage and semiconductor industry chain to jointly discuss new trends, new demands, and new cooperation in the storage industry in the AI era. Entering 2026, Agent has become a hot track, with office and lifestyle Agent products at home and abroad springing up like bamboo shoots after rain. Agents are becoming increasingly widespread in daily life. What users see is the convenience of "delegating complex work to an Agent with a single sentence." International Data Corporation IDC recently released the "DAA Research Report" (Daily Active Agents). The report shows that in 2025, the number of active Agents globally was 28.6 million, expected to reach 79.4 million in 2026, and grow to 2.216 billion by 2030. According to a Gartner report in March 2026, the amount of Tokens consumed per task by an Agent workflow is 5 to 30 times that of standard chat Siasun Robot&Automation, because a single user request usually triggers 10 to 20 sequential model calls. Goldman Sachs predicts that by 2030, global Token consumption will increase 24 times compared with 2026, reaching about 120,000 trillion Tokens per month; continuously online agents may consume more than 100,000 Tokens per day. In this round of AI infrastructure investment, the importance of memory chips is rising rapidly. Recently, South Korean brokerage KB Securities analyzed that as AI infrastructure investment expands at an unprecedented pace, the memory chip market is facing a severe supply shortage. The memory chip inventories of Samsung Electronics and SK Hynix are already less than 10 days of supply, and actual sellable supply next year may be clearly insufficient. KB Securities expects that next year AI-related investment by the world's leading cloud service providers will reach $1.3 trillion, a year-on-year increase of 60%. The report also points out that the share of memory chips in AI infrastructure investment is expected to rise from 14% last year to 40% this year, and further to 57% next year. The analysis says that structural changes on the supply side will further amplify the contradiction between supply and demand. Memory chip manufacturers are shifting a large amount of capacity to produce the new-generation high-bandwidth memory chip HBM4, which is specifically used for AI large-model training. Under the same storage capacity, the wafer capacity required to produce HBM4 is about three times that of traditional DRAM. DRAM is mainly used as the running memory for mobile phones and computers and is the core chip of general-purpose consumer memory. Against the background of limited total wafer capacity, the expansion of high-end memory chip HBM4 will squeeze the capacity of ordinary memory chips, which will further increase supply pressure in the subsequent market. China Securities Co.,Ltd. believes that since 2026, the investment logic of the AI industry has gradually shifted from a race in model capabilities and capital expenditure to commercial verification through orders, revenue, and profitability. Overseas Agent products have taken the lead in generating incremental revenue, and cloud providers' high-intensity computing power investment is still supported by cloud revenue and backlog orders; the capability gap of domestic models in Coding and Agent tasks is rapidly narrowing, with inference efficiency, Token call volume, and product revenue rising simultaneously. Looking ahead to the next one to two quarters, domestic model iteration, Agent product upgrades, and repricing of flagship models are expected to advance together. It is recommended to continue focusing on computing power services, domestic chips and super nodes, as well as B-end AI application vendors with scenario, data, and enterprise delivery capabilities. Dongguan Securities pointed out that from the demand side, AI training, inference, and agent applications continue to scale up, and overseas cloud providers' CAPEX is still being revised upward, driving computing power infrastructure construction to maintain high prosperity and continuously pulling demand for storage, CPU, power semiconductors, semiconductor materials, and other links. Some subcategories have already seen tight supply and demand and rising prices. From the supply side, key links such as semiconductor equipment, high-end GPUs, memory chips, and advanced packaging are still in a stage of accelerating domestic substitution, and the demand for independent and controllable industrial chains continues to strengthen. Related concept stocks: Z.AI (02513): In early September, CMBI issued a research report stating that Z.AI (02513) has a steady annual recurring revenue growth trajectory and strong prospects, and the company's disclosure of new user scale and participation further strengthens the bank's positive view. Driven by stronger annual recurring revenue growth, the bank raised its 2025-2028 revenue CAGR forecast from 199% to 275% and raised its target price from HK$1,503.9 to HK$1,985, maintaining a "Buy" rating. The bank believes that the company has achieved an effective balance among model iteration pace, intelligence level, and cost per task, and is expected to maintain a position on the Pareto frontier of the large language model industry, providing ample room for long-term monetization. The report points out that Z.AI completed 6 GLM iterations in 11 months, with its intelligence index rising from 32 to 60; MaaS platform registered users increased 144% to 7.4 million, paid daily active users increased 603%, Token call volume increased more than 40 times year-to-date, while the average API selling price rose about 101%. As of August 2026, the company's annual recurring revenue from the open platform and API reached $1.6 billion, further rising from $1 billion in early July, with management guiding for $2.4 billion by year-end. The deployment of a domestic 100,000-card chip cluster and the rebuilding of the inference stack reduced cost per Token by about 80% year-to-date, increased the computing power multiplier by 14 times year-on-year, and brought the open platform gross margin to 24.6% in the first half of 2026. CMBI raised its 2025-2028 revenue CAGR forecast from 199% to 275% and expects total revenue in 2028 to reach RMB 38.1 billion. ILUVATAR COREX (09903): ILUVATAR COREX's interim results show that, benefiting from the concentrated explosion in cloud AI inference computing power demand, the Zhikai series achieved breakthrough growth, with revenue of about RMB 654 million in the first half of this year, a year-on-year surge of 651.8%; gross profit of about RMB 233 million, with a gross margin of 35.6%, becoming the core DRIVE of ILUVATAR COREX's revenue growth; the Tiangai series, positioned for high-end training scenarios, still maintained a steady growth rate, with first-half revenue of about RMB 262 million, a year-on-year increase of 38%. KINGSOFT CLOUD (03896): In recent years, the company has continued to strengthen its AI cloud service capabilities and promote the upgrade of its traditional cloud computing business to AI infrastructure services. As one of China's earliest independent cloud service providers, KINGSOFT CLOUD has fully embraced AI in the past two years: relying on the natural scenarios of the Xiaomi and Kingsoft ecosystems, AI-related revenue has continued to scale, and intelligent computing cloud services have become the main growth engine; at the same time, it has continued to increase investment in AI infrastructure and strengthen full-stack capabilities from IaaS to MaaS. As enterprise demand for artificial intelligence applications continues to increase, companies with cloud platform capabilities, customer resources, and ecosystem systems are expected to gain new growth space in the process of AI industrialization. GBA AI COMP (01396): GBA AI COMP (01396) released its interim results for the six months ended June 30, 2026, with revenue growing tenfold: operating revenue of about RMB 2.56 billion, a year-on-year surge of about 1,020.9%, reaching a new revenue level in one step, with surging growth momentum. Profit achieved a leapfrog jump: profit for the period was about RMB 281 million, reaching 5 times the full-year net profit of last year (about RMB 56 million) in one step, with profitability achieving a leapfrog jump and reaching a new historical high. Computing power scale firmly ranks at the top of the industry: FP16 dense computing power scale exceeds 50,000 PFLOPS, firmly ranking at the top of the industry, forging the underlying strength of a 100,000-card-level ultra-large computing power cluster. SENSETIME-W (00020): At the end of August, CMBI issued a research report stating that it maintains a "Buy" rating on SENSETIME-W (00020) and raises the company's total revenue forecast for 2026 to 2028 by up to 7% to reflect the strong performance of the generative AI business. The company's first-half revenue increased 23% year-on-year to RMB 2.91 billion, broadly in line with market expectations. CMBI pointed out that SENSETIME-W disclosed for the first time that first-half recurring revenue (RR) reached RMB 1.14 billion, a year-on-year increase of 124%, accounting for 39% of total revenue, up from 22% in the same period last year, reflecting the successful transformation of its business model from one-time project delivery to a recurring revenue model. During the period, the company's generative AI revenue increased 28% year-on-year to RMB 2.33 billion, accounting for 80% of total revenue, mainly driven by strong demand for Token Factory and agent application services; visual AI revenue increased 14% year-on-year to RMB 497 million, with growth reaccelerating from 3% for the full year last year.