Is the Nasdaq in a Bubble? CAPE Says Yes, History Says Maybe
The valuation alarm is loud, the comparison set is ugly, and yet the cash flows underneath are nothing like 1999. Here's the honest read.
What the CAPE is actually saying
Let's start with the number everyone keeps pointing at. The Shiller CAPE ratio — price divided by ten years of inflation-adjusted earnings — is sitting in the high 30s to low 40s depending on whose feed you use. Multpl pegs it at 40.87, YCharts has it at 41.02 for June 2026, and GuruFocus shows 40.06 as of June 1, against a record high of 44.2 and a median of 16.06.
Translation: by this one yardstick, U.S. stocks are priced richer than at almost any point in 155 years of data. The only company the current reading keeps is the late 1920s — right before the Great Depression — and the year 2000, when CAPE peaked near 44 just before the dot-com bubble burst.
The Nasdaq's own multiple tells a similar story without the dramatic backstory. The Nasdaq 100 P/E was 35.73 as of June 11, 2026, versus a record high of 38.76 and a median of 24.51. Not a record. But not cheap, either.
Why "history says crash" isn't quite the slam dunk it sounds like
The bear case writes itself. Every time the Shiller P/E has crossed 30 over the last 155 years, a decline of at least 20% has eventually followed for one or more major indexes — and while it isn't a timing tool, the directional track record is clean. Add in the leverage picture: FINRA margin debt has crossed $1.3 trillion, and Bank of America's May 2026 survey shows institutional cash reserves at 3.9% — historically a contrarian sell signal.
So why isn't this an open-and-shut bubble call? Two reasons worth taking seriously.
First, the "E" in CAPE is a ten-year average. That window still includes 2015–2019 margins and the COVID earnings shock. If you believe the structural profitability of the S&P 500 and the Nasdaq has stepped up — software gross margins, hyperscaler operating leverage, AI-driven productivity — then the denominator is artificially low and the ratio is overstating the stretch. You can disagree with that view; you can't dismiss it.
Second, the AI-versus-dot-com analogy is doing a lot of heavy lifting in the bear case, and it's weaker than it looks. Many of the dot-com darlings had no real business model — investors were buying a narrative, not unit economics. The same simply isn't true of AI as a whole. The Mag 7 throw off enormous free cash flow. Cisco in 1999 traded on a story; today's hyperscalers trade on actual capex they can fund out of operating cash. Different animal.
The bond-market tell most people skip
Here's the metric that actually moves the needle for me, and it gets almost no airtime: the Excess CAPE Yield. Shiller built it because raw CAPE ignores interest rates, and at the end of the day stocks compete with bonds. ECY is the inverse of Shiller PE minus the real 10-year Treasury yield — it captures the margin that stocks are paying over bonds.
As of May 2026, the Shiller Excess CAPE Yield was 1.32, against a historical median of 3.31. Still positive — stocks are technically out-yielding bonds on a forward real basis — but the cushion is thin. That's not a bubble verdict. It's a "priced for perfection, with little room for error" verdict. Which is a meaningfully different thing.
What would make the bubble call right — and what would make it wrong
Bubble thesis confirmed if: AI capex stops translating into actual earnings growth at the hyperscalers, real rates push higher, and the ten-year earnings figure in CAPE doesn't catch up fast enough. Then you're paying 1999 multiples without the 2000s productivity payoff, and the math gets ugly.
Bubble thesis broken if: AI-driven earnings come in hot enough over the next few years that the ten-year average in CAPE re-rates upward, compressing the ratio without prices having to fall. The S&P 500 and Nasdaq Composite both notched double-digit returns for the third year running in 2025, with technology, industrials, energy and utilities leading on AI tailwinds — if that earnings growth is durable, the "expensive" label fades on its own.
The bottom line
CAPE at 40 is a real signal, not noise. It's also not a forecast. The honest read: the Nasdaq is priced for AI delivering, the cushion versus bonds is thinner than usual, and historical precedent for this multiple is genuinely bad. None of that tells you what happens in the next twelve months — CAPE has never been a timing tool — and the structural case that this earnings base is different from 1999's is stronger than the doomers admit.
Bubble? By the single most-cited valuation metric, yes. By the underlying cash flows, no. Reality is probably somewhere in between, and "priced for perfection" is the phrase that actually fits the data. That's not advice — it's just what the numbers say when you stop yelling at them.