算力产业链内部明显分化，基金密集调研这些方向

Core Viewpoint - The computing power industry chain is experiencing a resurgence, driven by the demand for AI and related technologies, with significant stock price movements in related sectors [1][3]. Group 1: Market Performance - On February 26, stocks related to PCB, CPO, liquid-cooled servers, and computing power chips surged, with companies like Shenzhen South Circuit and Dazhu Laser hitting the daily limit [1][3]. - The CPO concept stock Tianfu Communication saw a remarkable increase of 78.31%, while PCB stocks like Mingyang Circuit and Dazhu Laser rose by 91.29% and 69.26% respectively [1][10]. - The overall performance of the Sci-Tech 50 Index fluctuated, initially dropping over 1% before closing up by 0.85% [1]. Group 2: Financial Results and Projections - NVIDIA reported a Q4 revenue of $68.13 billion and earnings per share of $1.62, exceeding Wall Street expectations, and provided a revenue forecast of $78 billion for Q1 of FY2027 [3][4]. - The strong financial results from major chip companies validate the expanding demand for computing power, with expectations of continued high demand in the AI sector [4]. Group 3: Investment Trends - There is a noticeable increase in fund interest in computing power stocks, with multiple funds conducting research on companies involved in data centers, PCBs, and optical switches [5][6]. - Funds like E Fund and Southern Fund have shown significant interest in Dazhu Laser, which is experiencing strong demand for AI computing data center servers and related products [6][10]. Group 4: Industry Insights - The semiconductor industry is witnessing a price increase in passive components, driven by AI and automotive demand, with price hikes ranging from 5% to 30% expected by early 2026 [4]. - The AI computing power sector is forming a positive feedback loop, with increased capital expenditure expected to sustain high demand for related hardware, including optical modules and PCBs [4][11]. - The upcoming GTC 2026 conference is anticipated to focus on NVIDIA's next-generation Feynman architecture, which could significantly enhance the market value of PCBs used in AI applications [11].