Michael O'Rourke witnessed the internet boom of the late 1990s and its subsequent bust firsthand as an institutional trader for Spear, Leeds & Kellogg on the floor of the New York Stock Exchange.

O'Rourke, now chief market strategist at JonesTrading, an institutional brokerage and investment banking firm, sees eerie parallels between the dot-com era and today's artificial-intelligence, or AI boom, as he has repeatedly warned in The Closing Print, his daily newsletter. Simply put, he believes too many investors are chasing the same tech stocks and paying too much for them, and that the AI "bubble" will burst, much as the internet frenzy did.

"I was trained to have independent views and like to think they are honest, concise, and upfront," says O'Rourke, who joined Jones in 2013 after stints at BTIG and the hedge fund Marshall Wace. "I'm going to share my thoughts, whether they're popular or not."

O'Rourke shared his thoughts about the stock market and the economy with Barron's in mid-August, and he has a point: They may not be popular with stock speculators, and even some investors bidding shares of companies with uncertain prospects to new highs. But judge for yourself: An edited version of the conversation follows.

Barron's: You have warned that the artificial-intelligence trade is a bubble resembling the late-1990s dot-com era, which ended with a crash. Why do you draw that comparison?

Mike O'Rourke: I feel like I'm reliving history. I believe in AI. It is incredible and life changing. Yet, I frequently reference a Warren Buffett piece in Fortune from 1999. Buffett talked about how technologies can change society and make people's lives better, but that doesn't mean they will work out well for investors.

Any asset bubble is founded in something we all believe in. But when everyone is in agreement about the outlook, investors tend to be willing to overpay for assets, as they are doing with many AI-related stocks.

Premiums are already deflating, and we should expect to see continued contraction in price/earnings multiples as the cycle matures. Investors will see share prices decline as valuations "appear" more attractive, but that reflects the market expressing concerns that the growth cycle has peaked.

When people overpay for assets in a bubble-type investment environment, there is usually an overshoot to the downside, as well. Take Amazon.com, one of the greatest American success stories ever. Yet, its early success didn't stop its shares from falling 85% to 90% from 2000 to 2002, during the dot-com bust. Microsoft and Cisco Systems shares fell dramatically, too, after the dot-com market crashed. It can take years to undo the damage caused by overpaying for stocks.

Hyperscalers and chip companies are investing directly in the AI labs OpenAI and Anthropic. How concerned are you about vendor financing propping up the AI economy?

Nvidia, the leading AI chip maker, is another incredible success story. But the urgency to use vendor financing to grow your system instead of allowing it to grow organically can be a problem.

Lucent was the late-1990s equivalent of today's Magnificent Seven stocks [technology leaders, including Nvidia]. It used borrowed money to fund customer purchases and that was its undoing after the tech bubble burst. When you artificially try to extend the growth and spending cycle, you are taking on new risks.

Nvidia just announced a deal with OpenAI to provide financing for a massive data center in Ohio. Nvidia is backstopping at least $105 billion of the deal. In taking on financial obligations of OpenAI, Nvidia has tied itself to the risks and fate of that cash-burning company.

Pulling forward demand that isn't organically there yet is a risk. The demand is real, but it is subsidized. That is why the hyperscalers deserve lower price/earnings multiples than they have now. These companies grew to be behemoths primarily as asset-light, free-cash-generating, share-buyback machines with quasi-monopoly status in their primary business lines. Investors are concerned that these companies are destroying some of the most successful, scalable, investor-friendly business models the equity market has ever seen.

They are transforming into asset-heavy enterprises. Capital-intensive businesses generally trade at high-single-digit or low-double-digit P/E ratios. [Google parent Alphabet is trading for 20 times forward-12-month earnings, below its peak valuation of 30 times over the past five years.]

Cisco Systems was the network-equipment leader in 2000. Then its growth started to decelerate. Once that happened, investors started selling. Decelerating growth is a danger in this market, too.

Are the big tech stocks no longer "safe" for investors?

This environment is changing quickly and the uncertainty about many businesses is growing. Software stocks on average used to trade for about 25 times earnings. Now there are concerns that AI will commoditize software-as-a-service models. Software has become a 10- to 15-times multiple business because of disruption fears.

Alphabet entered 2026 with a top large language model, but as the year has unfolded, there are concerns it may be falling behind. The company is borrowing money and selling stock to invest in data centers and train its own AI model. Borrowing and selling shares are things that Alphabet hasn't historically done.

What industry sectors do you like?

I have frequently described this as a market I hate at the index level because it is so expensive. [The S&P 500 trades for 20 times forward-12-month estimated earnings.] At the same time, there are many opportunities if you look at stocks from a bottom-up perspective. So much money is chasing the AI trade that valuations are depressed elsewhere in the S&P 500.

There are large-cap pharma companies with single-digit multiples of earnings, and 4% to 6% dividend yields. There are consumer staples stocks trading at 10 to 11 times earnings that pay 5% to 7% dividend yields. These are more defensive, and I admittedly have more of a value bias. But investors haven't crowded into these stocks.

Another set of opportunities lies in stocks considered growth names before the AI era that began in 2023. They are on sale and don't get the same attention that they once did from growth-stock managers and other institutional investors. These aren't recommendations, but Chipotle Mexican Grill, Netflix, Uber Technologies, DoorDash, and Spotify are some of the names I think about. They aren't AI-centric. A growth manager who shares my concerns about AI stock valuations can invest in some traditional growth companies at the most attractive valuations in their publicly traded existence.

AI is one dominant market theme. The Federal Reserve is another this year. You're not in the interest-rate-hike camp, as you recently wrote, meaning you don't think the Fed ought to raise interest rates. What is your view of Kevin Warsh, the Fed's new chairman?

Warsh's nomination potentially could be the most consequential positive decision that President Donald Trump has made during both of his terms in office. Warsh spent the past 15 years working with Stanley Druckenmiller, one of the greatest investors of all time. Due to this experience, Warsh recognizes the importance of allowing markets to function and the importance of price discovery. Those who have spent their careers in markets know that properly functioning markets are an incredible tool for signaling whether the economy or policy is on a healthy course.

Warsh said before his Senate confirmation that he believed an AI-led productivity boom ultimately would allow the Fed to lower interest rates. The key drivers of inflation today are higher energy prices, tariffs, and the AI buildout. Interest-rate hikes don't do anything to reverse those trends.

All rate hikes would do at this point is further squeeze American consumers. Megacap American corporations are winning and Main Street is losing. Warsh and Treasury Secretary Scott Bessent have both talked about it being time for Main Street to prevail.

Does anything other than the AI craze worry you?

There are excesses out there. There is probably more bad credit floating around the system that we aren't aware of because credit is still available. It is just more expensive.

That leads me to all the liquidity in the market. There has been a great pivot in the past 20 years to indexing and passive investments. Given the S&P 500's performance since 2012, investing in an index has been an attractive option at low cost. But index funds are now buying all of these stocks in the S&P 500 and essentially retiring that float.

I would argue that active money managers put time into research and compete to set prices. That gives you good price discovery. But when 30% of the float of a stock is in the hands of a price-insensitive buyer because a committee has designated that company's inclusion in an index, that impairs the price-discovery process.

Some of the biggest market players now are quantitative investors. They are trading signals but they aren't buying businesses because they like the companies' growth prospects. A lot of the daily trading volume comes from people who aren't looking at corporate fundamentals. That includes holders of levered exchange-traded funds and 0DTEs, or zero-day-to-expiration options.

There are also aggressive retail speculators. They have more knowledge and access to information than ever. Some of it is good but a lot is bad. I understand that people want to make their own financial decisions. No one is looking to change the laws and rules around trading because the market is near all-time highs. But when the next downturn comes, and it will at some point, sentiment may shift in favor of regulations to rein in speculation.