The S&P 500's 7% year-to-date pullback has been largely led by artificial intelligence (AI) stocks suddenly being seen as liabilities rather than assets. Microsoft is 26% below where it ended last year, while Palantir Technologies is off by nearly 20%. IBM is down 20% since the end of 2025, and Oracle shares are 28% lower.
This swell of disinterest is understandable, though. Many of these tickers were already struggling with steep valuations stemming from a little too much hype about AI's potential upside. When Microsoft, Alphabet, Oracle, and Amazon all recently announced they'd be making massive investments in artificial intelligence infrastructure this year, investors were forced to acknowledge that being in this business is neither easy nor cheap. Not all of the names in the AI industry are going to thrive.
Still, this sweeping -- and somewhat indiscriminate -- sell-off is more of an opportunity than an omen. Although the artificial intelligence industry is now suffering a reality check similar to what dot-com names experienced in early 2000 (and like solar stocks did in 2012, and cannabis stocks in 2019, along with meal kit, 3D-printing, and cryptocurrency names each did at some point in their pasts), the world still needs AI solutions, and it needs them right now. This could be a short-lived correction for most of these tickers.
Not exactly an apples-to-apples comparison
Don't misread the message. The artificial intelligence stock sell-off may or may not have run its full course already. Certainly, some of these names have already reached their ultimate bottom, while others still have more downside to dish out.
On balance, though, this industry's stocks as a whole still offer more upside than downside, for a couple of related but overlooked reasons.
One of those reasons is highlighted by Jeremy Siegel, professor emeritus of finance at the University of Pennsylvania's Wharton School of Business. In a regularly published commentary posted late last year -- and in contrast with the conditions seen in the midst of the dot-com craze of the late 1990s -- he explained, "These [AI companies] are real firms with real cash flows, not concept stocks."
DataTrek Research's Jessica Rabe echoed that idea in a research report around the same time, noting, "These companies have much better fundamentals now than in the late 1990s."
And both Siegel and Rabe are right. These stocks may be relatively expensive compared to historical norms. But these companies' artificial intelligence operations aren't bleeding money. This ultimately means they can remain afloat without burning through resources while fine-tuning their approach to the AI business (as the survivors of the 2000s crash did).
Image source: Getty Images.
Investors who believe the steep valuations are enough to steer clear of every name in the artificial intelligence business are also missing another important nuance. As Citi analyst Heath Terry describes, what's happening to many of these right now is more of a price "reset" rather than a full-blown reckoning. AI's profit growth is still plausible this year and next.
He's not alone either. Terry's explanation echoes viewpoints held by Benchmark's Bull Gurley as well as 22V Research's Jordi Visser and with DataTrek's Rabe, just to name a few.
This right-pricing may turn out to be less damaging than anticipated, too, if analysts' near-term earnings projections are any indication. The analyst community is calling for Nvidia (NVDA -0.13%) to report per-share earnings of $8.27 for the current fiscal year ending in January, up 74% from last year's $4.77, en route to a profit of $11.12 per share next year. Microsoft's projected revenue growth of more than 16% this year is expected to drive per-share earnings to 26%.
The point is, although they seem expensive right now, most of these "overvalued" stocks are simply reflecting next year's likely profits. Microsoft's forward-looking price-to-earnings ratio is well under 20, as are Nvidia's and Broadcom's. These aren't unreasonable prices to pay for the growth potential these companies offer.
NASDAQ: NVDA
Key Data Points
Or if nothing else, think about it like this: Any popping of an AI bubble is likely to only lead to a smaller and shorter-lived correction than the dot-com crash of 2000.
Recognize and respond to the reliable pattern
Sure, there may be some lingering pain from individual stocks. That's just the nature of the beast. Every company's situation is a little bit different.
On balance, though, comparing what AI stocks are going through right now to what most technology stocks did during the late 1990s isn't a great apples-to-apples comparison.
While there are certain similarities, the differences are even bigger. And the biggest? In the '90s, investors were unrealistic about what could and would become a viable internet-focused company. Today, the market has a pretty good handle on how there's money to be made in the artificial intelligence business, and who's going to make it. Most people are simply tweaking their expectations here, spurring some (arguably necessary) volatility. That was certainly the case for solar power and the entire internet-tech industries, both of which have more than bounced back from major setbacks simply because there's real demand for both businesses. AI isn't apt to be any less marketable in the long run. The world is simply right-pricing these tickers for what this future is likely to look like.
This spells opportunity for smart investors, of course, who recognize how unbridled hype about a new industry is often followed by strong, fear-driven selling that underestimates that industry's actual long-term marketable value.
