Huachuang Securities: Scarcity of computing power highlights the potential for structural expansion in the AIDC industry.

Huachuang Securities released a research report stating that recently, global AIDC demand has continued to grow rapidly, with domestic and foreign cloud vendors simultaneously releasing positive signals in terms of capital expenditure, business expansion, and pricing strategies. The price increase of cloud vendors and the expansion of AI spending jointly reshape the logic of the AIDC industry. The price hike enhances the expected return on computing power assets, while the surge in demand raises the industry's growth ceiling, driving AIDC from a heavy asset industry to a core infrastructure race with high barriers to entry and high certainty. Leading vendors with technological iteration capabilities and efficient resource integration are expected to continue to benefit from structural dividends. Huachuang Securities' main points are as follows: Overseas cloud demand expansion combined with rising hardware costs confirm the scarcity of computing resources. According to MarketsandMarkets data, the global cloud computing market is rapidly expanding, expected to grow from approximately $1.29 trillion in 2025 to around $2.28 trillion in 2030, with a CAGR of about 12%. Meanwhile, the supply side faces structural constraints: rising prices of storage chips and CPUs increase cloud service costs. On January 25th, according to the Korean Electronic Times, Samsung Electronics raised the supply price of NAND flash memory by more than 100% in Q1 2026. TrendForce previously predicted a price increase of 33% to 38% for NAND in Q4 2025, and it is expected to maintain a similar increase in Q1 this year. KeyBanc data shows that Intel Corporation and AMD plan to raise server CPU prices by 10-15%. At the same time, energy costs are rising, with American Electric Power Company, Inc. reporting a 6.7% year-on-year increase in costs in December last year. Since 2020, the cumulative increase is about 38%, while overall consumer prices have only risen by 2.7%. The expansion of AIDC demand and cost pressures are driving cloud service providers to start a price hike cycle, breaking the industry's long-term "price reduction trend" and confirming the scarcity of computing resources. Domestic AI model competitions drive accelerated computing infrastructure deployment, intensifying demand on the application side. Companies like DeepSeek, Kimi, Alibaba Qianwen, have launched model upgrades, highlighting the huge demand for computing power for training and inference due to rapid model performance iteration. The industry's competitive focus has shifted from single-model performance to overall competition in computing power supply and system engineering efficiency. Alibaba Group's CEO Wu Yongming's "cloud + AI + chip" strategic support is driving leading vendors to build competitive barriers through coordinated software and hardware integration. Competition at the application layer is also heating up. Tencent's Elephant Valor quickly gained users by integrating the DeepSeekR1 model with large-scale marketing resources. Its promotional strategy relies heavily on WeChat's ecosystem traffic and continuously iterates on features to optimize user experience and retention. The explosive growth of such C-end applications may directly high-frequency calls for back-end intelligent computing centers, high-speed networks, and other basic infrastructure, further accelerating the scale deployment and upgrade of related technologies. The AIDC industry may enter a period of structural expansion with high barriers to entry. In response to the technical challenge of single-rack power consumption leaping from the traditional 5-15kW to 200kW, innovative solutions such as liquid cooling and high-density power supply are rapidly penetrating (e.g., Huawei's iCooling solution reduces PUE by 8%-15%). With the maturity of the domestic chip ecosystem, Alibaba's "Zhenwu 810E" performance is benchmarked against mainstream international products, helping the domestic computing power system achieve software-hardware synergistic optimization. In this context, the AIDC market is expected to expand at a compound growth rate of 31.5%, with the industry's core barriers shifting from capital investment to technology integration and operational efficiency. Huachuang Securities believes that the price increase of cloud vendors and the expansion of AI spending jointly reshape the logic of the AIDC industry. The price hike enhances the expected return on computing power assets, while the surge in demand raises the industry's growth ceiling, driving AIDC from a heavy asset industry to a core infrastructure race with high barriers to entry and high certainty. Leading vendors with technological iteration capabilities and efficient resource integration are expected to continue to benefit from structural dividends. Recommendations for investment targets: 1) Cloud computing: BABA-W (09988), Cloudflare (NET.US), Sangfor Technologies Inc. (300454.SZ), KINGSOFT CLOUD (03896), Sunevision (01688), Ucloud Technology Co., Ltd (688158.SH). 2) AIDC: Range Intelligent Computing Technology Group (300442.SZ), Shanghai Baosight Software (600845.SH), Shanghai AtHub (603881.SH), Beijing Sinnet Technology (300383.SZ), Guangdong Aofei Data Technology (300738.SZ), INESA Intelligent Tech Inc. (600602.SH). 3) Computing power services: Sharetronic Data Technology (300857.SZ), Glory View Technology (301396.SZ), Dawei Technology (Guangdong) Group (600589.SH), Shenzhen Neoway Technology (688159.SH), Jiangsu Lettall Electronic (603629.SH), Jwipc Technology (001339.SZ), Coreweave (CRWV.US). 4) CDN: Wangsu Science & Technology (300017.SZ). 5) Chips: Hygon Information Technology (688041.SH), Cambricon (688256.SH), MetaX Integrated Circuits (688802.SH), ILUVATAR COREX (09903), Moore Threads Technology (688795.SH), Loongson Technology Corporation (688047.SH). 6) Large models: Minimax-WP (00100), KNOWLEDGE ATLAS (02513), Iflytek Co., Ltd. (002230.SZ), etc. Risk warning: Unexpected technological progress, intensified competition, project delivery delays, and collection risks.