I realize — better than most I’d wager, given how immersed I stay in the market zeitgeist — that ominous narratives centered around the purported pitfalls of extreme market concentration invariably come across as repetitive and anyway fatuous: What’s anyone supposed to do? Not own the leadership? Avoid the AI hyperscalers? Good luck with your old-world widget value stock trading below cash!

Still, anytime a given market metric scores in the 99%ile on a century-long lookback, it’s worth harping on, if not perhaps fretting over.

With five of the “Magnificent 7” set to report in the days ahead, I wanted to draw readers’ attention to the figure below, from Goldman’s David Kostin.

Never so much mind the specifics of Kostin’s “market concentration” calculation. The point is simply that on his metric, the S&P 500 has only been more concentrated one time in over 100 years (see the red “current” annotation).

The scatterplot’s pretty clear: Extreme concentration bodes ill for forward returns outside of recessions. That’s where the proverbial rubber meets the road on the valuation-concentration nexus. As Kostin explained, “the relationship between concentration and valuation” doesn’t show investors always price the risk of higher volatility from extreme concentration, but “returns are ultimately higher-than-average following periods when valuations were low and market concentration was high to compensate for [that] risk.” When the economy’s coming out of recession, low valuations can accompany high concentration.

Goldman’s research team got some pushback last week when Kostin cited extreme concentration in suggesting long-term returns for the S&P 500 could be very subdued looking out over the next decade. A few days later, in a separate note, he emphasized that the analysis “specifically pertains to the S&P 500 capitalization weighted index and does NOT suggest that equities in general will deliver low long-term returns” (all-caps in the original). His point: Growth and innovation aren’t the sole purview of seven (or 10) companies.

He did concede that valuation and concentration might’ve reset “structurally higher” in the AI / Mag7 era, but he seems to doubt it. Consider the figure below.

“Challenging” (Goldman’s word choice in the chart header) is a euphemism. The odds of a given firm sustaining 50% EBIT margins in perpetuity are vanishingly small.

Still, some argue we’re witnessing an epoch in AI on par with the dawn of the personal computer era, or maybe even the taming of fire and the invention of the wheel. Laugh as you will, but contrary to the old adage, sometimes — rarely, but sometimes — “this time” really is “different.”

“[We] could be wrong… and forward returns [for the cap-weighted S&P 500] would likely be stronger than our model projects… if the emergence of AI allowed several of the largest stocks in the market to maintain rapid levels of growth, high levels of profitability and elevated market weights,” Kostin wrote.

So, again: Maybe this time’s different. Maybe generative AI is the wheel. Or fire. Or the PC. But probably not.