Markets are starting to scrutinize AI, and that's a good thing, says Griffin

Citadel founder and CEO Ken Griffin says even if AI stocks are in bubble, they might be worth holding onto.

There's a bubble in artificial-intelligence stocks - but it's no reason to sell, one of the leading figures in finance said Tuesday night.

"The big picture is there is almost certainly bubbles within this moment in time. And any radical transformation of our economy is generally intertwined with a bubble," said Ken Griffin at an annual conference run by Citadel Securities, the market maker he mostly owns. Griffin also is the majority owner of Citadel LLC, the hedge-fund giant.

As for where he stands on investing in AI, Griffin drew on his experience in the decades-ago boom and bust where internet companies at times were valued unrealistically.

"I just wish I had kept my Amazon stock from that moment in time. And the internet, which was the great bubble, could it be more of a part of our life today than it is?" he said in an interview that published Wednesday.

"What's left in the wake of these bubbles, if we want to use that word, is often game-changing for the world. And the companies that are on the other side of that bubble are the market leaders for decades to come," he said.

Griffin discussed the recent slump in semiconductor stocks - the Van Eck Semiconductor exchange-traded fund SMH has dropped in 10 out of the past 11 sessions - and uncertainty surrounding some of the biggest public listings to come in the AI space.

The Nasdaq Composite COMP has dropped for seven of the last 10 sessions.

"I think this market is actually experiencing people doing research," he said. That marks a change from the past couple of years when there was a sense of just buying the entire AI supply chain, from hyperscalers and semiconductor makers to high-tech companies in Korea, Taiwan and Japan, he said.

While it's clear that there are "extremely high margins" for makers of memory in the AI supply chain, "we're reaching a point where people are becoming more discerning," he said. "People are now starting to look out two years, three years. How long is this going to last?"

One major issue he sees for the industry is the potential overcapacity of graphics processing units (GPUs), vital in the AI build-out as they help train large language models by making highly technical calculations.

"We saw this actually with the telecom build-out back in 2000," he said. "We built more fiber than we had demand for and the price of telecommunications just went straight down."

The same question now hangs over GPUs. "If we reach a point where the provisioning of GPUs exceeds demand for GPUs, marginal pricing will actually go into free fall. And that really upends all the investment assumptions that people are working with today."

Investors are now stepping back and asking what the "useful life" of a GPU looks like right now. "There are people that are earning extraordinary pricing on GPUs that are five or six years old. That probably doesn't persist for another five or six years," he said.

"So there's a growing set of concerns that 'I can't give the same capitalization rate or the same revenue models to year seven, eight, nine of a chip in the future that maybe I was using 12 months ago.' So I think we're really grappling with just how far will this demand carry us in the marketplace."

Griffin believes U.S. companies also need to improve at figuring out which parts of their business will require "extremely high pricing for the state-of-the-art models from OpenAI and Anthropic," and where can they rely on cheaper open-source models.

"I think all of American corporations is looking at their bills from Anthropic and OpenAI and going, 'OK, we need to think about how to make sure we're getting productivity out of this.' And they will in many areas, but in some areas you've got a lot of waste and people are going to have to pivot around that," he said.

"What we don't know is how as these models become more intelligent, how much are they able to demand high pricing because they help to actually solve the problems that have vexed humanity for centuries," he said.

U.S. stock-market futures (ES00) (NQ00) were higher on Wednesday after upbeat results from SpaceX and Super Micro Computer.

-Barbara Kollmeyer