It’s been (another) good year for the mega-cap market leadership.

There are exceptions. Namely Apple and Tesla, where a lack of product innovation and — ahem — product sales, respectively, have weighed on stock performance.

Everything else to do with high-stakes US tech’s performing quite well. You can tell because everyone’s wailing about market breadth and extreme concentration again. Just like they did last year and the year before. As a reminder you don’t need: Market breadth, like valuations, isn’t a reliable tool if you’re silly enough to think you can call tops and trade them.

The table below shows a Goldman basket of AI-exposed stocks outperforming the median S&P 500 constituent by 23ppt YTD, and the cap-weighted index (which is a third big-tech) by 18ppt.

That all-too-familiar divergence continues to manifest as elevated dispersion. Measured using three-month rolling stock-level returns, dispersion’s 80%ile or so looking back three decades, Goldman’s David Kostin remarked.

Dispersion rises when either single-stock vol is high or correlations are low. Vol’s been “relatively contained” this year, as Kostin put it (“relatively” to account for the tariff fireworks in April), but “the varying impacts of key macro themes, including tariffs, Fed policy, fiscal policy and AI, have pushed correlation among stocks back down near record lows.” (Remember: Short correlation is just long dispersion, and vice versa.)

That’s all a bit esoteric. The point (my point) in mentioning it is to set up the figure below, which shows valuation dispersion.

As the chart text notes, that’s the P/E premium for the most expensive stocks, sector-neutral, versus the least expensive stocks, expressed as a percentage.

What does that tell us? Why am I (hopefully not) wasting your time with this? Well, it gives you a more nuanced sense of the “rich”/”extreme” valuations criticism of the rally.

On that more subtle metric, the discrepancy between the richest stocks and the cheapest is both extreme (in the context of the historical median) and not (in the context of the two “froth” episodes we typically cite when making comparisons (the dot-com boom and the 2021 go-go “stimmy” bonanza).

As Kostin put it, valuation dispersion’s “very elevated” at nearly 200%, which ranks 91%ile since 1980, but it’s “well below the peaks in 2000 (328%) and 2021 (287%).”