He cited the Fed no longer cutting, weaker commodities and wider credit spreads, and said that in equities he now looks for "AI use cases as opposed to infrastructure".
He warned that AI infrastructure suppliers whose earnings doubled or tripled cannot keep growing at today's rates, so a low forward P/E may not mean cheap.
Einhorn turned around the question clients ask him, noting that AI providers are investing trillions chasing uncertain returns while being rewarded with “generous (or extremely generous) valuations”.
Griffin said AI is “almost certainly” in a bubble, but that the companies on the other side of such bubbles often become the market leaders for decades.
Grantham's January 2026 essay argues that every two-sigma bubble in a large developed equity market has eventually returned to trend, and that AI is following the classic path of great innovations that lead to great bubbles.