You can’t separate the S&P 500’s YTD gains from the performance of large-cap US tech shares.

And you can’t separate large-cap US tech shares from the A.I. hype.

Put that together and what do you get?

Well, I’ll tell you: You get a pretty strong case for the contention that were it not for ChatGPT, US stock benchmarks would be lower in 2023, not higher.

“Without the A.I.-popular stocks, the S&P would be down 2% this year, not up 8%,” SocGen’s Manish Kabra remarked, in a recent note. He’s hardly the only analyst to suggest as much, although I’m not sure I’ve heard anyone else express it so unequivocally.

Earlier this week, for example, JPMorgan’s Marko Kolanovic wrote that “while there may be substance in the ChatGPT euphoria, the question is whether it is enough to be the basis of a rally for the broader market.” A day later, Nomura’s Charlie McElligott described a market “in thrall to this AI frenzy, which ties into many of the S&P’s largest companies.”

SocGen has an AI newsflow indicator (I’ve mentioned it before), and it’s accelerating on a 90 degree angle — “rising exponentially,” as Kabra put it.

The figure on the right above gives you a sense of just how crucial the A.I. hype is for index returns. This is a “Where would we be without GPT?” market.

Kabra’s view is that equities will likely be rangebound for a while. Thanks in no small part to the contributions of mega-cap tech and other names seen as “AI-popular,” the S&P is trading at the top of that range.

Notwithstanding resilient profits, a sturdy US labor market, a likely resolution to the debt ceiling standoff and better-than-expected quarters from the most important companies in the cap-weighted benchmarks, I should reiterate that at current multiples, stocks are fully priced — and on some pretty rosy earnings assumptions. I know I’ve said that again and again, but it bears repeating because, to roll out another familiar talking point, if we’ve already seen the trough for ex-energy EPS growth, this would count as a very, very shallow earnings recession.

But, as Kabra noted, “it is tough to fight strong hype in the very short-term,” and the hype is strong with A.I. So strong, in fact, that companies in every, single sector talked about A.I. on their analyst calls. 77% of comms services companies mentioned A.I. For tech, the share was 62%. Even 10% of energy companies had something to offer on the subject.

“Our economists have estimated that generative A.I. could potentially lift US productivity growth by roughly 1.5 pp per year over 10 years following widespread adoption [and] based on the historical relationship between productivity growth and corporate profitability, this boost could lift S&P 500 net profit margins by roughly 4 pp over that decade, holding all else equal,” Goldman’s Ben Snider wrote, in a new piece of his own, adding that “the relatively stable share of revenues allocated to SG&A expenses in recent decades, in contrast with large declines in other input costs, underscores the potential gain to profitability from A.I.”