OpenAI would have to spend over $1 trillion to deliver its promised computing power. It may not have the cash.

Core Insights - OpenAI is projected to spend over $1 trillion in the next five years to achieve its computing power goals through partnerships with Nvidia, Broadcom, and AMD [1][4] - The company aims to deliver 26 gigawatts of computing capacity, which is comparable to the electricity needs of New York during peak summer [2] - Citi estimates that the capital expenditures for OpenAI could reach $1.3 trillion by 2030, with even bolder internal projections suggesting a potential cost of $12.5 trillion for 250 gigawatts by 2033 [3] Financial Projections - OpenAI's revenue is expected to reach $163 billion by 2030, which is significantly lower than its projected expenditures [4] - The company has made substantial commitments, including a $300 billion deal with Oracle for a 10-gigawatt AI infrastructure project and additional projects in the UAE and Norway [5] Industry Concerns - There are concerns regarding the sustainability of AI demand and the potential for a stock market bubble driven by investor optimism in AI [6] - Questions remain about whether the US power infrastructure can scale to meet the energy demands of OpenAI's projects, which could hinder the company's ability to capitalize on its investments [7]