What the price already assumes

Every price is a forecast. When a stock trades at a high multiple of its earnings, that multiple is not a mood — it is a specific, decodable claim about how much bigger those earnings will be, and how reliably, and for how long. Which means you don't need to hold an opinion about AI to be exposed to one. If you own a broad US index fund, a historically large share of your money now sits in a handful of companies whose prices encode a strong view about how the AI buildout resolves. The opinion is already in your portfolio. The only question is whether you know what it says.

The arithmetic, with made-up round numbers

Here is the mechanism with deliberately hypothetical figures. Suppose a company earns $10 a share and trades at $400 — a 40x multiple — while the market's long-run average multiple sits somewhere in the mid-to-high teens. Now suppose that over the next decade the company performs well: earnings triple, to $30. If, over that same decade, its multiple drifts down toward 20x as growth matures — still a premium rating — the stock lands at $600. That's about 4% a year. Tripling earnings bought a savings-account-plus return, because the starting price had already spent most of the growth in advance.

Run it the other way and you get the bull case honestly stated: for that stock to deliver strong returns, earnings must grow fast enough to outrun the multiple compression that historically accompanies maturity — or the multiple must never compress. Both are possible. Neither is a neutral assumption, and the second one, dressed up, is what "this time is different" always cashes out to. The point of the exercise isn't that high multiples are wrong. It's that a price is a claim, the claim can be extracted with division, and once extracted it can be argued with — which is more than can be said for vibes.

What the base rates say about the claim

The load-bearing assumption is durable high growth, so it's worth knowing what the record shows: academic work on growth persistence has found that very few companies sustain well-above-average earnings growth for as long as a decade, and that identifying the ones that will, in advance, is close to a coin flip. That research predates this era and deserves a fresh look against it — today's giants are more entrenched than the average firm in those samples, and that's a genuine counterargument. But the burden of proof sits with the price that assumes persistence, not with the base rate that doubts it.

The part that concerns an index investor

Concentration is the channel through which all of this reaches a Boglehead. When the largest names in a cap-weighted index carry a near-record share of it, "just buy the index" and "make no bet on AI" quietly stop being the same sentence. You still own everything — but you own the encoded optimism of a few mega-caps in size, and your next decade of index returns depends meaningfully on whether their prices' claims come true.

What follows from that is not a trade. Selling the winners because they've won is an active bet; so is doubling them because the story is exciting. The index position is still defensible for the same dull reasons it always was: you'll misidentify the winners if you try, concentration has been high before and resolved in both directions, and the alternative to holding the market's opinion is holding your own, which has a worse track record. What changes is your expectations and your posture. Expected returns are set by starting prices, and starting prices are demanding — so plan around modest numbers and treat anything better as weather. Rebalance on schedule, which trims the encoded optimism mechanically without requiring you to be right about it. And when the gap between the adoption curve and the return curve resolves — in whichever direction — the entire job is to be someone who doesn't flinch. That part was never arithmetic.