Shang Hai Zheng Quan Bao·2026-02-12 22:56Core Viewpoint - UBS and Goldman Sachs have significantly reduced their holdings in major US tech stocks like Nvidia and Microsoft in Q4 2025, raising market concerns about the tech sector's future performance [1][3]. Group 1: Holdings Reduction - UBS reduced its holdings in Nvidia by 10,042,089 shares, a decrease of 11.47%, and also cut its positions in Microsoft by 2,320,211 shares (7.64%), Apple by 5,266,649 shares (10.57%), Amazon by 1,658,256 shares (4.57%), and Google by 2,206,303 shares (9.05%) [3][4]. - Goldman Sachs also reported reductions, cutting 319,700 shares of Microsoft (5.86%), 247,000 shares of Tesla (8.27%), 343,300 shares of Broadcom (9.33%), and 241,400 shares of META (13.51%) [5]. Group 2: Capital Expenditure Plans - Major tech companies, excluding Nvidia, have announced plans to significantly increase capital expenditures, with Meta projecting up to $135 billion for 2026 (an increase of 87%), Google planning $185 billion, and Amazon announcing a $200 billion capital expenditure plan for 2026 [7]. - Following these announcements, stock prices for Amazon, Google, META, and Microsoft have seen declines of 14.72%, 8%, 6.67%, and 6%, respectively, since February [7]. Group 3: Market Sentiment and AI Investment - There is growing skepticism in the market regarding the return on investment (ROI) from the substantial capital expenditures in AI, which had previously been tolerated by investors [6][8]. - Despite short-term concerns about ROI, some analysts believe that the current investments represent a strategic positioning for future productivity, affirming the long-term trend of AI development [8][9]. Group 4: AI Development Trends - The AI sector is transitioning from a phase of technological exploration to one of large-scale application and infrastructure development, with a focus on increasing computational power and practical AI applications [9]. - The demand for computational resources in the tech industry continues to exceed market expectations, indicating a rapid expansion phase for AI infrastructure [9][10].