AI stocks show bubble-like traits as retail buying surges, says McKinsey’s Tim Koller

McKinsey’s Tim Koller points to ownership patterns in the so-called Magnificent Seven stocks as the clearest warning sign (Pic: McKinsey & Co)

The explosive rally in AI stocks is showing increasingly clear signs of speculative excess, with a sharp rise in retail investor ownership emerging as one of the strongest indicators that the market may be entering bubble territory, according to one of the world’s most experienced voices in corporate valuation.

McKinsey’s Tim Koller — whose foundational valuation framework dates back 35 years — says the fundamentals of valuation have not changed since he first codified them in 1990: long-term value comes from revenue growth and return on capital. “Those two things will drive its cash flows and its profits and then that's what drives the value,” he said. But in the current AI cycle, prices are being driven less by long-term cash-flow expectations and more by investor excitement, particularly from retail buyers.

He points to ownership patterns in the so-called Magnificent Seven stocks as the clearest warning sign. Retail shareholders now hold “double what it is for the typical company,” he said — a dynamic that has historically been associated with overheating rather than fundamentals. “Whenever you see share prices being driven by retail investors… they’re more likely to be deviations from sort of a reasonable value,” he added.

This surge in non-institutional buying mirrors the signature traits of earlier speculative markets. Koller noted that whether it was the dot-com boom or the post-Covid meme-stock spikes, all featured the same amplification: retail enthusiasm pushing valuations far above what any reasonable long-term model could justify. “These things don’t last… they’ll always eventually get back to fundamental,” he said, recalling how several high-growth names that soared in 2019–2022 “were at ridiculously high levels… and then they came back down to earth.”

The concern today, he said, is not just the run-up in valuations but the mechanics of how AI stocks are being priced. Investors appear to be extrapolating transformative potential without anchoring it to eventual revenue or profit models. Yet even for AI, he stresses, the valuation rules have not changed: software companies will still be judged by “the cash flows that they’re going to generate right… five, seven, ten years out.” Chipmaking, data centres and software each require understanding their underlying economics, which he believes are comparable to established sectors. He noted that the main difference in AI-related businesses today is the degree of uncertainty rather than a change in the basic method of valuing them.

Koller said retail investors tend to focus on how exciting or promising a company appears without weighing what price is reasonable to pay for that promise. “Retail investors tend to forget… it’s not just about how great a company it is but am I paying a reasonable price for it,” he warned. This behaviour, combined with a lack of rigorous analysis relative to institutional investors, often leads to momentum-driven overshoots. “You could have a fantastic company that is still overvalued… simply because people are not thinking about the price,” he said. “Some of them seem hard to justify and others the expectations aren’t so unreasonable,” he added, stressing that each company must be assessed individually rather than as a thematic basket.

When asked about interest rates and valuations, he referred to his earlier analysis showing that the drop in rates during the period of near-zero monetary policy did not translate into lower equity return expectations. “The cost of equity didn’t change,” he said. He explained that investors believed rates would eventually rise again, so they maintained the same return requirements. Because of this, the subsequent increase in US interest rates did not have the valuation impact some investors expected.

On overall US market valuations, he said concentrated moves in the largest technology stocks can influence perceptions of market levels. Removing the top seven companies, he said, shows that the rest of the market is only at the high end of normal ranges. While not commenting specifically on Indian market valuations, he said comparing multiples across countries requires accounting for differences in industry mix, growth and returns on capital.