Computing power is national power, related concept stocks collectively surged!

On December 30th, computing power concept stocks collectively surged. As of the time of publication, SINOSTAR CABLE (300933.SZ) rose over 17%, INESA-it (600602.SH) and AtHub (603881.SH) increased by 10%, Aofei Data (300738.SZ) rose over 9%, and ZTE Corporation (000063.SZ), Cambricon (688256.SH), Digital China (000034.SZ), and Topway Information (002261.SZ) all strengthened. Recently, CCTV and the State-owned Assets Supervision and Administration Commission jointly produced a large-scale documentary series "Cornerstone of a Great Nation," with the third episode titled "Computing Power Engine," which emphasizes that computing power is national strength and is a key engine driving social progress. By the end of 2023, China's total computing power ranked second in the world. Computing power is emerging vigorously, opening up unprecedented industries. In the wave of the digital age, computing power is profoundly changing our world. In addition, local governments are continuously increasing their investments. Recently, Shanghai issued the "Implementation Plan for Artificial Intelligence 'Molding Shanghai'," proposing to strive for the city's intelligent computing power scale to exceed 100 EFLOPS by the end of 2025; to build a super-large-scale autonomous intelligent computing cluster, and to accelerate the development of autonomous intelligent computing chips such as general-purpose graphics processors, application-specific integrated circuits, and programmable gate arrays. The continuous policy announcements indicate that the computing power industry will maintain a high prosperity trend, leading to a collective surge in related concept stocks, attracting capital like a magnet. Recently, the industrial aspect of computing power has also seen continuous positive developments, with major tech giants racing to invest in the AI large model field, which will drive a boom in upstream computing infrastructure. Not long ago, Douyin's Doubao family underwent a comprehensive update, with the application scenarios of Doubao's large model continuously expanding. The future of domestic AI looks promising, which has led to renewed interest in AI computing power hardware. According to media reports, ByteDance's capital expenditure budget for 2025 has soared to nearly 160 billion RMB, mainly for AI computing procurement and IDC infrastructure. In addition, Xiaomi is working on building its own GPU cluster, which will heavily invest in AI large models. The Xiaomi large model team already had 6,500 GPU resources at its establishment, signaling an increased emphasis on AI large models. Industry insiders predict that since ByteDance and Xiaomi have clearly adopted an "AI in AI" attitude, other tech giants are likely to follow suit quickly, continuously opening up the imagination space for upstream computing infrastructure It is worth noting that recently, DeepSeek-V3 has emerged, causing a huge stir in the AI industry due to its remarkably low pre-training costs, even sparking discussions among many overseas AI scientists. According to reports, the performance of the DeepSeek-V3 version is quite robust. Preliminary evaluations indicate that its reasoning capabilities are comparable to GPT-4o and Claude-3.5-sonnet, with programming abilities even rivaling the leading 3.5 Sonnet, surpassing GPT-4o. Andrew Carr, a former researcher at OpenAI and now an AI entrepreneur, stated, "China is accelerating its catch-up efforts, attempting to replicate OpenAI's reasoning model. The DeepSeek researchers managed to achieve this in just a few months, and frankly, many of my colleagues are surprised by this." CITIC Securities' research report pointed out that DeepSeek-V3 has significantly improved training efficiency and reasoning speed while ensuring model capabilities. The release of this model signifies that the application of large AI models will gradually become more inclusive, facilitating the widespread implementation of AI applications; at the same time, the substantial increase in training efficiency will also support the high demand for reasoning computing power. With the frequent emergence of favorable policies for domestic AI large models, topics such as computing power and AI+ have garnered significant attention from investment banks. CITIC Construction Investment Strategy stated that it is crucial to focus on the direction of policy efforts and the main line of the AI+ industry. The bank believes that the recent widespread application of the Doubao large model in various scenarios such as information processing, customer service, sales, and AI tools, along with the rapidly increasing call volume, demonstrates a significant enhancement in its commercialization potential. DeepSeek-V3 has achieved excellent performance with relatively low training costs and operational expenses. Favorable policies for the low-altitude economy are frequently emerging, with the National Development and Reform Commission officially establishing a Low-Altitude Economic Development Department, accelerating industry development once again. The rapid emergence of new defense equipment may significantly increase attention on the military industry sector. With multiple thematic catalysts frequently appearing, it is recommended to seize the opportunity for multi-clue rotation in thematic strategies at the beginning of the year