算力“硬通货”当道，英伟达产业链狂奔

Core Viewpoint - The AI computing power competition is reshaping the global technology industry, with AI computing power becoming a "hard currency" in infrastructure as demand for large model inference and training surges [1] Group 1: AI Server Market - AI servers are the core revenue segment, with Industrial Fulian achieving record revenue and profit in the first half of the year, with Q2 revenue surpassing 200 billion yuan and a year-on-year growth of over 60% in AI server revenue [2] - Inspur Information reported a revenue of 801.92 billion yuan in the first half of 2025, a year-on-year increase of 90.05%, with a net profit of 7.99 billion yuan, up 34.87% [2] Group 2: Optical Module Market - Optical modules are crucial in AI investments, with leading companies like Zhongji Xuchuang showing strong growth, reporting a revenue of 147.89 billion yuan in the first half of 2025, a year-on-year increase of 36.95%, and a net profit of 39.95 billion yuan, up 69.40% [2] - Tianfu Communication also demonstrated significant growth, with a revenue of 24.56 billion yuan, a year-on-year increase of 57.84%, and a net profit of 8.99 billion yuan, up 37.46% [2] Group 3: Liquid Cooling and PCB Market - The demand for liquid cooling is expected to rise significantly, with a projected penetration rate of 33% in AI data centers by 2025, and companies like Yingwei Ke achieving a revenue of 25.73 billion yuan, up 50.25%, and a net profit of 2.16 billion yuan, up 17.54% [4] - PCB manufacturers are also experiencing growth, with Shenghong Technology reporting a revenue of 90.31 billion yuan, a year-on-year increase of 86%, and a net profit of 21.43 billion yuan, up 366.89% [4] Group 4: Overall Market Trends - The A-share computing power industry chain is being revalued by the market, with leading companies showing strong performance, indicating a robust demand driven by the expansion of global AI infrastructure [5] - Major cloud service providers like Microsoft, Amazon, Meta, and Google are increasing capital expenditures significantly, with a combined capital expenditure of 87.4 billion USD in Q2 2025, up 69% year-on-year [7] Group 5: Future Projections - NVIDIA expects global AI infrastructure spending to reach 3 to 4 trillion USD by the end of the decade, with China being the second-largest computing market projected at around 50 billion USD, growing at a CAGR of approximately 50% [8]