你以为“美国国王”是特朗普，其实是黄仁勋？

Core Viewpoint - The article expresses concern over the United States' heavy reliance on AI models and computing power, suggesting that this focus may lead to an economic bubble rather than sustainable high-quality growth [1][8]. Group 1: Market Concerns - There is significant volatility in the U.S. stock market, primarily due to differing opinions on whether the AI bubble will burst [2]. - Major investors, including SoftBank and Michael Burry, have taken actions such as selling Nvidia stocks and shorting AI companies, indicating a growing concern about the sustainability of AI valuations [4]. - Wall Street perceives current market conditions as reminiscent of the 2000 internet bubble, with companies' valuations diverging significantly from their fundamentals [6]. Group 2: AI Valuation and Energy Concerns - Nvidia's price-to-earnings ratio stands at 63, suggesting that investors would need 63 years to recoup their investment, which is seen as unrealistic for a hardware manufacturer [7]. - OpenAI is projected to incur losses exceeding $5 billion in 2024, yet its valuation is estimated at $300 billion, raising questions about the sustainability of such high valuations [7]. - The energy consumption of AI models is a critical issue, with OpenAI's GPT-3 requiring 1,300 MWh of electricity, and the newer GPT-5 consuming 9 to 20 times more energy per query [12][15]. Group 3: Economic Growth and Investment Dynamics - A study by Harvard economist Jason Furman indicates that nearly all U.S. GDP growth in the first half of 2025 will stem from data centers and information processing technologies, with other sectors showing a mere 0.1% growth rate [10]. - The current economic growth is heavily driven by capital investments in AI models and data centers, which are also leading to increased electricity demands that the existing grid cannot support [12][15]. - The need for substantial investments in energy infrastructure to support AI growth is highlighted, with projections suggesting that the U.S. will need to double its current electrical grid capacity to meet future demands [15]. Group 4: Competitive Landscape and Future Outlook - The article discusses the competitive landscape between the U.S. and China in AI, noting that while the U.S. has advanced technology, China possesses significant advantages in energy production [31][34]. - China's electricity generation capacity is projected to reach 10 trillion kWh in 2024, with a substantial portion being renewable energy, positioning it as a potential leader in AI development due to lower energy costs [31][34]. - The ongoing competition in AI technology is ultimately tied to energy resources, with the article suggesting that the U.S. may have technological prowess but lacks the energy infrastructure that China possesses [34].