Rob Arnott: On AI, 'Never Short-Sell a Bubble'

What You Need to Know

Artificial intelligence is an example of companies opening a new industry deemed likely to be pathbreaking.
The emerging technology draws parallels to electric vehicles and PalmPilots.
With sticky inflation and higher interest rates, inflation is unlikely to be sorted out in the next year or two.

Investors shouldn’t bet against the artificial intelligence bubble that’s driving the U.S. stock market but don’t need to participate in it either, financial analyst Rob Arnott suggests.

“One of the points that I like to make with regard to bubbles is never short-sell a bubble. It can go further than you can possibly imagine,” Arnott, Research Affiliates founder and chairman, told ThinkAdvisor in a phone interview this week. “Be very, very careful about the notion of shorting bubbles but you don’t have to own them.”

Nor should advisors and clients assume that an S&P 500 index fund would leave them diversified enough to avoid damage from a bursting AI bubble, he said.

“The dot-com bubble was special and rare but shockingly similar to today,” he said, adding that investors who were broadly diversified across the S&P saw a roughly 45% loss by the time the market reached its lows.

From March 2000 to March 2002, the bear market’s first two years, the S&P 500 was down about 20% while the median Russell 3000 stock was up 20%, meaning that for most companies, the bull market that ended in 2000 didn’t actually end until 2002, Arnott explained.

“Then there was a short, sharp bear market that took everything down for the second and third quarter of 2002, and then you were back off to the races,” he said.

While an S&P investor was down 45% at the market lows, someone who was broadly diversified equal weighting the Russell 3000 was, net net, down 15% or 20%, he said.

Today’s market has the same kind of stretched multiples, Arnott added.

Understanding ‘Big Market Delusion’

The AI bubble is an example of what researchers have dubbed “the big market delusion,” in which a roster of companies opens a whole new industry that is deemed likely to be pathbreaking, Arnott said.

Stock prices are based on the best plausible scenarios but fail to take into account “the fact that the companies compete against one another so they can’t all win,” he said. ”And the result is a collection of companies whose aggregate market capitalization can’t be justified by plausible outcomes.”

The delusion also fails to consider that groundbreaking changes will likely take many years to unfold and that today’s dominant players may disappear in a few years, Arnott said.

Arnott and two others wrote about this in the EV context in early 2021, when there were nine companies that produced only electric vehicles.

“The point of that paper was not to say these companies won’t succeed,” he said. “It wasn’t to say this isn’t an important market, it’s going to go away. Quite to the contrary, market prices are set based on narratives, and narratives have the advantage of being largely true and the huge disadvantage of being entirely reflected in the current share prices. So if you bet on a narrative, you’re betting on nothing because it’s already reflected in the share price.”

Big market bubbles and delusions also go wrong in expecting things to change very quickly, “and they expect the current winners to be the future winners,” he said, citing a dot-com bubble example inPalm, maker of the PalmPilot.

“Everybody had a PalmPilot,” but in a few short years, “BlackBerry blew them out of the water,” and a few years after that, “iPhone blew both of them out of the water,” Arnott added. “So disruptors get disrupted.”

The changes also happened slowly, he noted.

“Handheld devices are central to almost everything we do today. That wasn’t true 10 years ago. It certainly wasn’t true 20 years ago,” Arnott said. “And yet, 25 years ago, as the dot-com bubble was taking shape, the presumption was everything was going to change in the next five years.”

None of the 10 largest market-cap tech sector stocks in the S&P 500 in 2000 were ahead of the index over the next 15 years, Arnott noted. Only one, Microsoft, pulled ahead by 2018. “How many today? Two,” Microsoft and Oracle, “and you had to wait 24 years.”

Today, the narrative is that AI will change everything, Arnott noted.

“It’ll change how we transact, how we, it’ll change the nature of search engines. It’ll change the ways we interact socially. It’ll leverage our time and efforts in communicating with clients, with friends. It’ll leverage the way we do research. It will accelerate business decisions. It’ll replace millions of white-collar workers, but it’ll also create millions of new jobs,” he said.

“OK, that’s actually probably all true. So that’s where narratives are seductive because they’re largely true,” Arnott said, adding that it’s unlikely that Nvidia’s competitors will let the leading AI chip maker keep its roughly 100% market share in those ultra-fast chips.