Citadel CEO Ken Griffin says a glut of artificial intelligence (AI) chips could send pricing into freefall, comparing graphics processing units (GPUs) to airline seats that earn money even when sold cheap.
Griffin said bubbles almost certainly exist within AI, though he stopped short of predicting a collapse. Companies that emerge on the other side of bubbles lead markets for decades, he added.
Does Ken Griffin’s Airline Analogy Fit AI Chips?
Griffin spoke with Bloomberg‘s Dani Burger at Citadel’s annual Future of Global Markets Conference in New York.
“GPUs like airlines have a very low marginal cost to run.”
Ken Griffin, speaking to Bloomberg
He argued that an unfilled airline seat sold for as little as $20 still earns money, so fares collapse once capacity exceeds demand.
He cited the telecom build-out around 2000, when companies laid more fiber than demand required and communication prices fell.
If GPU supply outstrips demand, pricing could drop into freefall, he said, upending the investment assumptions behind today’s AI buildout.
Why Is Wall Street Questioning AI Hardware Returns?
Meanwhile, Griffin described the recent AI-linked selloff as investors doing their research, and said hedge fund deleveraging should be the 10th explanation anyone reaches for.
Wall Street firm BTIG now expects a prolonged chip slump that could run into 2027, after the Philadelphia Semiconductor Index (SOX) fell 11.4% in a week, its largest drop since April 2025.
Griffin also questioned the useful life of a GPU. Chips five or six years old still earn rates he called extraordinary, but he doubted that holds for another five or six years.
In addition, corporate buyers are sorting which tasks justify premium OpenAI and Anthropic models and which can run on cheaper open-source ones, Griffin said, agreeing that this strains against the labs’ spending.
Not every strategist accepts the dot-com parallel, however. Yardeni Research President Ed Yardeni forecasts the S&P 500 at 8,400 by mid-2027 and says the 1999 and 2000 bubble comparison no longer fits.
Whether AI revenue catches up with the capacity being built will decide how much of that spending earns a return.
