Key Points
- OpenAI’s recent chip deals could cost the company over $1.3 trillion by 2030, according to Citi.
- The company has promised to deliver 26 gigawatts of new computing power.
- This massive spending is fueling Wall Street’s fears of an AI-driven stock market bubble.
- There’s a huge disconnect between OpenAI’s projected costs and its expected revenue.
OpenAI’s massive plans to build out its artificial intelligence infrastructure could cost the company over $1 trillion in the next five years, according to a new analysis from Citi. The jaw-dropping figure highlights the incredible expense of the global AI arms race and is adding to Wall Street’s growing fears of a stock market bubble.
The warning comes after OpenAI recently announced a series of huge deals with chipmakers Nvidia, Broadcom, and AMD. In these partnerships, OpenAI has promised to deliver a staggering 26 gigawatts of new computing capacity—enough to power the entire state of New York.
Citi analysts did the math and estimated that it costs about $50 billion to bring just one gigawatt of that capacity online. At that rate, OpenAI’s total spending would hit $1.3 trillion by 2030.
And that might just be the start. Internally, CEO Sam Altman has reportedly floated even bigger numbers, suggesting the company could be aiming for 250 gigawatts of computing power, which would cost an astronomical $12.5 trillion.
The problem is, it’s not clear where all that money is going to come from. While OpenAI’s costs are set to soar into the trillions, Citi estimates its revenue will only reach a fraction of that, about $163 billion, by 2030. This huge disconnect has some analysts worried. “[Sam Altman] has the power to crash the global economy for a decade or take us all to the promised land,” wrote one Bernstein analyst, “and right now we don’t know which is in the cards.”
