“看，皇帝没穿衣服”！对冲基金经理：万亿美元的AI投入，赚得回来吗？

Core Viewpoint - The investment return gap in AI data center construction is significant, requiring trillions of dollars in investment over the next 3-5 years, with a comparison to the U.S. Department of Defense's annual budget of $1 trillion [1][2][4] Investment Requirements - To achieve a 10% capital return, AI data centers need $1-2 trillion in revenue, while good returns may require $3-4 trillion [2][4][9] - The current AI industry revenue is estimated at $15-20 billion, indicating a need for 30-fold growth to break even [10][11] AI Business Model Flaws - AI models like ChatGPT and Claude are highly substitutable, leading to low user loyalty and intense price competition, potentially reducing profits to just above energy costs [2][4][17] - The rapid advancement of large language models (LLMs) means free versions will remain sufficient for users, limiting the willingness to pay for premium services [14][15] Circular Investment Concerns - The concept of "circular investment" is likened to the telecom bubble of 2000, where companies like Lucent and Nortel created false revenues through financing and leasing [2][4][23] - Major companies like Meta and Microsoft are accused of using equity and leasing data centers to create "safe" assets, misleading investors about the true nature of their revenues [2][4][19] Infrastructure and Market Dynamics - The construction of AI data centers is compared to building railroads, with investors facing repeated failures throughout capital cycles [18][19] - The current trend of purchasing land for data centers mirrors the housing market speculation of 2006-2007, indicating a potential bubble [6][40] Future Outlook - The expectation of massive investments in AI infrastructure raises questions about the source of funding and the sustainability of such growth [10][14] - The potential for significant losses in the AI sector is highlighted, with predictions that the financial fallout could reach trillions of dollars [23][24]