兴业证券王涵 | 美国AI神话褪色，市场“三杀”风险上升

Core Viewpoint - The recent adjustments in the US AI sector, coupled with the widening disadvantages in key AI fields, signal potential changes in the US economy and financial markets, challenging the optimistic expectations surrounding "AI hegemony" and hinting at rising systemic risks in the financial market [1][2]. Group 1: Market Adjustments - The US AI sector has cooled down, with the Nasdaq index dropping over 3% from November 3 to 7, and core AI companies like Nvidia seeing a decline of over 7% in market value [5]. - AI concept valuations have become excessively inflated, with the "Big Seven" in the US stock market accounting for approximately 35% of the S&P 500's market value, compared to only 23% during the internet bubble era [5]. - A significant 95% of investments in generative AI have yet to yield actual returns for companies, indicating substantial uncertainty behind the "AI myth" [5]. Group 2: Key Support Areas for AI Development - The four core support areas for AI development are energy, algorithms/talent, data, and computing power, where the US is facing notable disadvantages and challenges [6]. - In the energy sector, the imbalance between electricity supply and demand is a critical issue, with AI data centers driving a surge in electricity demand. The US Energy Information Administration (EIA) projects a 4% increase in electricity consumption by 2024 compared to 2019, with AI-related GPU server demand pushing data center energy consumption to 176 TWh in 2023, expected to rise to 12% of total US electricity consumption by 2028 [7][8]. - On the supply side, the US struggles to enhance its electricity infrastructure, leading to rising electricity prices, with industrial and commercial electricity prices increasing by 20% from 2019 to 2024, and residential prices by over 26% [8]. - In the talent sector, the US produces approximately 645,000 STEM graduates annually, while China exceeds 2.2 million, indicating a significant gap in high-quality talent availability [11]. - The data sector has been adversely affected by the "America First" policy, which has restricted the potential data available for AI training, limiting the US's competitive edge [14]. - In computing power, while the US currently leads with about 74% of global high-end AI computing power, this advantage may diminish due to power shortages and China's rapid advancements in AI chip production and efficiency [15]. Group 3: Market Impact - In the short term, monetary easing may provide some support against downward pressure on US stocks, but the risk of a simultaneous decline in stocks, bonds, and the dollar is rising in the medium term [16]. - The core narrative of the US stock market is heavily reliant on the expectation of "AI hegemony," which may be challenged if doubts about the US's global dominance and the credibility of the dollar arise, potentially leading to a systemic revaluation of asset prices [16].