The AI‑Fueled Stock Surge Is Showing Its Age – A 21% Crash May Be Coming in 2027

If you’ve been riding the AI‑driven equity rally this year, you might want to savor the last few weeks of 2026. A chorus of analysts is already pointing to the same familiar warning signs that foreshadowed the dot‑com bust, and they’re betting that the market could tumble about 21 % before 2028 rolls around.

James Reilly, senior markets economist at Capital Economics, stuck to his earlier forecast in a note on Thursday: the S&P 500 could close 2026 around 8,250 – roughly a 7.7 % gain from its Friday close – but then slide to about 6,500 by the end of 2027. “The data look consistent with a late‑stage bubble,” he wrote, noting that several key metrics are hovering at levels we’ve seen before a major correction.

What exactly is ringing alarm bells? For starters, valuations. The market’s cyclically adjusted price‑to‑earnings (CAPE) ratio is now perched close to the peak it hit during the dot‑com frenzy. Even the S&P’s price‑to‑Treasury‑bond spread is nudging the same extreme territory. In plain English: stocks are pricey relative to the safe‑haven yields that historically kept excess enthusiasm in check.

Next up, earnings growth. Forward‑looking 12‑month earnings‑per‑share growth for the index mirrors the upside we saw at the height of the early‑2000s tech bubble. That kind of optimism is hard to sustain, especially when you layer on the AI factor.

The AI angle is especially precarious. The biggest hyperscalers – the firms that have poured billions into AI chips, data centers, and talent – are projected to see their combined free cash flow dip into negative territory by 2027. In other words, the cash‑generating engine that usually powers further investment may be sputtering.

Concentration is another red flag. A handful of mega‑caps now dominate the market‑cap weight of the major indexes. History teaches us that when a few stocks carry most of the market’s lift, the rally becomes fragile, prone to sudden reversals.

And there’s the sheer volume of new equity. IPO pipelines are bursting, and follow‑on offerings are lining up like a queue at a buffet. Past bubbles have often shown a flurry of issuances right before the crash – a pattern that many see repeating now.

While Reilly didn’t mention Treasury yields directly, his colleague at Rockefeller International, Ruchir Sharma, has been waving that flag louder than ever. In a recent Financial Times op‑ed, Sharma warned that the AI bubble could pop once the 10‑year Treasury yield “decisively breaches” the 5 % mark – a ceiling that bounded yields during the dot‑com era.

Why does a 5 % yield matter? Higher borrowing costs make it tougher for AI‑heavy companies to raise debt, and they’ll also feel the pinch when issuing new equity, because investors demand a bigger risk premium when safe returns climb. Moreover, yields at that level start to brush up against the nation’s nominal GDP growth, putting additional strain on the already‑inflated public‑debt load, which now exceeds 100 % of GDP.

Even the most bullish market veterans are tweaking their outlooks. Ed Yardeni, a long‑time Wall Street forecaster, trimmed the probability of his “Roaring 2020s” scenario from 80 % to 70 %, and nudged the odds of a bearish outcome up to 30 %. He cited recent turbulence in oil and bond markets as unsettling signs.

All this isn’t a doom‑saying manifesto; it’s a reminder that markets move in cycles. The AI surge has delivered spectacular gains, but the underlying fundamentals – valuations, earnings sustainability, cash flow, concentration, and financing conditions – are starting to look a lot like the ingredients that brewed past crashes. Whether the correction arrives gently or with a bang, investors would do well to keep an eye on those 5 % Treasury yields and the next few months of price action.