Valuations Are Nearing Dot-Com Peaks: Is a 2026 Market Crash Imminent?

The S&P 500's most-watched valuation gauges are sitting at levels not seen since 2000, just as a historically rocky midterm-election year gets underway. None of it guarantees a correction — but the signals are stacking up faster than usual.

The Shiller CAPE Ratio Is Closing In On Its Dot-Com High

The Shiller CAPE ratio, which smooths S&P 500 earnings over a decade to filter out short-term noise, stood at 41.76 in July 2026 — 29% above its own long-term average of 32.36, and within range of the roughly 44 level last reached just before the dot-com crash. A separate tracker put the current reading as "second only to the record 44 reached right before the dot-com bubble burst," a framing that understates how rare this territory has historically been.

The metric has real limits, though. CAPE is built to smooth out cycles, which means it also lags — by the time it flags overvaluation, prices may have already run for years. Economist Jeremy Siegel has argued that structural shifts like near-zero-cost indexing justify a higher "normal" P/E than history suggests, while Robert Shiller himself continues to emphasize reversion to the mean. Both views are live among serious analysts; neither has been settled by the data so far.

AI Capital Spending Keeps Revising Itself Upward

Hyperscaler guidance for 2026 AI infrastructure spending has moved three times in a matter of months: an initial estimate near $650 billion, revised up to $725 billion, with Moody's now projecting $785 billion — and $1 trillion pencilled in for 2027. That trajectory is the clearest evidence that the AI buildout isn't slowing, but it's also the detail most likely to feed the chip-stock selloff tied to rate fears: every upward revision raises the bar the eventual returns have to clear.

The genuine structural difference from 2000 is financing. Microsoft, Alphabet, Meta, and Amazon have largely funded their data-center buildout from operating cash flow rather than debt or equity raises, which historically has made bubbles more painful to unwind. That said, the gap is narrowing — reporting this year describes some firms beginning to tap debt markets as buildout costs start to outstrip cash generation, a shift worth watching alongside the AI-subsidy unwind now landing on enterprise buyers.

A Middle East Oil Shock Added a Second, Unrelated Stress Test

Valuation risk isn't the only pressure point. On February 28, 2026, the US launched military action against Iran; within days, Iran closed the Strait of Hormuz, a chokepoint handling roughly 20% of global petroleum exports. The resulting price move was severe: West Texas Intermediate crude jumped from $67.02 a barrel on February 27 to $111.24 by March 8 — a 66% move in under two weeks, described as the fastest oil-price spike in more than 40 years. Gasoline prices rose roughly 50 cents a gallon, and some analysts have floated $150 crude as a further risk if the disruption persists.

The mechanism that matters for equities is inflation. An oil shock of this size raises the odds that the Federal Reserve holds rates higher for longer, which is exactly the combination — elevated valuations plus a less accommodative Fed — that has historically preceded sharper drawdowns rather than orderly ones.

History's Midterm-Year Pattern Cuts Both Ways

Layered on top of both stories is ordinary election-cycle seasonality. Across long-run studies going back to 1932 and 1962, the S&P 500 has averaged roughly 5.8–5.9% during midterm-election years — well below the long-run average near 10% — before rallying an average of 16.3% in the 12 months that follow. Some analyses find that rally runs from the midterm-year low to the following year's high, averaging close to 47% cumulatively.

The nuance worth stating plainly: U.S. Bank's own research team tested whether the "midterms cause weak returns" pattern was statistically meaningful using 31 elections across 125 years, and found the difference was not statistically significant, given the modest sample size and returns ranging anywhere from a 30%-plus loss to a near-50% gain in a single year. A pattern can be real in the averages and still not be a reliable timing tool.

What Separates a Correction From a Repeat of 2000

The bull and bear cases here aren't fringe positions — both come from mainstream institutions reading the same data differently. On the bull side, the S&P 500 currently trades at about 20.5 times forward earnings versus a 19x ten-year average, described by one outlet as "a modest premium," with Wall Street's consensus projecting 24% earnings growth in 2026 and a median 12-month target implying roughly 19% further upside. Goldman Sachs and J.P. Morgan have both argued the AI-driven growth is fundamentally justified rather than speculative.

On the bear side, Capital Economics has forecast that the AI-fueled bubble bursts in 2026 as rates and inflation weigh on valuations, and a Bank of America fund-manager survey found 45% of respondents named "AI bubble" as the market's biggest tail risk in November, up sharply from 11% in September. Investor Michael Burry has taken a public leveraged short position tied to semiconductor stocks, telling subscribers the market has "jumped the shark" — a view echoed, with less colorful language, by Stanley Druckenmiller and David Einhorn.

None of this settles the question, and the historical patterns discussed above are descriptive rather than predictive: they describe what has typically happened, not what must happen this time. Readers tracking the broader cash-burn unwind among "zombie unicorns" will recognize the same tension playing out at the private-market level — real spending, real returns in some pockets, and a valuation debate that won't be resolved by any single data point.