Revered investor Warren Buffett admitted at Berkshire Hathaway’s annual shareholder meeting Saturday that “I made the wrong decisions” when he passed up investments in Amazon and Google (now Alphabet). It’s a mistake he made because he stuck to his longtime investing principles—and that might mean it’s time to update those ideas.
“I had [a] very very very high opinion of [Jeff Bezos’s] ability when I first met him, and I underestimated him,” Buffett said in comments reported by CNBC. “I’ve watched Amazon from the start. I think what Jeff Bezos has done is something close to a miracle . . . the problem is when I think something will be a miracle, I tend not to bet on it.”
Buffett’s own limited technical savvy might have contributed to the missed opportunity. Bill Gates, according to Buffett, had to personally tell him to switch from the creaky Altavista search engine to Google—and Buffett also admitted that “stupidity” kept him from investing in Microsoft early on. In that light, turning away from Google and Amazon reflects the investing philosophy that has been core to Berkshire Hathaway’s epochal success. Buffett has long cautioned investors to only bet on businesses they understand, and he admitted that when it came to Google and Amazon, “it would have been far better obviously if I had some insights into certain businesses.”
Get Data Sheet, Fortune’s technology newsletter.
But another aspect of tech investment is more fundamentally at odds with Buffett’s approach. Buffett also said Saturday that “I made the mistake in not being able to come to a conclusion where I really felt that at the present prices that the prospects were far better than the prices indicated.” That’s a summation of Buffett’s emphasis on value investing—buying stocks that underprice the intrinsic value of companies.
But tech companies for at least a decade have been persistently overvalued by the stock market compared to their assets and revenue, on the basis of their potential to totally dominate markets in the future. Buffett has been quite public about the appeal of companies with monopoly-like market positions, but those positions are built differently in the data-driven tech world than in many of the sectors where Buffett has succeeded. Network effects and, more recently, data-backed “feedback effects” act as a new kind of monopolistic moat for tech companies, and building those moats takes huge, long-term spending. Early investors in Amazon had to weather years of deep losses before profits came—but when they arrived, it was in torrents.
Of course, just because pseudo-monopolies are good investments doesn’t mean they’re good for society. Regulators are still catching up with what a monopoly means in the digital age, but a growing number of critics are calling for limits on the size and power of the very giants Buffett wishes he had bet on.