Do you think the collective market could be pricing in massive expected productivity growth over the next ~10 years driven from AI, which would make current prices relative to pretty much any valuation metric look highly inflated for companies that are best positioned to benefit. Companies which, in this case, may also be the largest ones with the most expertise and capital, like the S&P500?


Sure. But one could abbreviate that to "the market is pricing in [some narrative] and overpaying as usual".

If some novel "displacement" makes a firm more profitable that's good, but the increase in price due to that displacement almost always exceeds the average improvement in business prospects. Solid fast-growing firms are certainly worth more than boring slow-growing firms, but the world seems full of people happy to pay three times as much for a company with twice as good a future. Plus, of course, there is the problem that technology diffuses. If all firms become more productive, no firm becomes more profitable, as they almost all have competitors.

There will be a few giant winners of course, as there are in any five year stretch. The problem is that nobody really knows in advance which ones they will be, so the only thing you can rationally focus on when estimating future returns is how the average will do.

What will LLM models do to the trajectory of (say) future US real GDP growth? Maybe 3% higher a decade from now than it would otherwise have been? 5%? Maybe 8%? What is that per year? Since there is only so much GDP to be split among companies, individuals and taxes, in aggregate the present value of all future corporate profits won't be that much different. Better? Sure, I won't argue. Enough better to deliver good returns starting from these valuation levels? Nope.

The broader problem is that US valuations are pretty stretched even for boring middle-of-the-pack firms. I track as a metric the median P/S among the largest 400 non-financial S&P 500 firms. The tech bubble peaked at 1.639:1. The 2007 credit bubble peaked at 1.844:1. In the whole of US history up till October 2013, this never exceeded 2:1. Around 2019 it passed 3:1 for the first time. The current figure is 3.584:1.
Is each dollar of sales at a boring middle-of-the-pack big firm really worth more than twice its average since 1997?

Jim