Why Wall Street Analysts Now See a Double Bubble in Stocks

On Wall Street, the bull case for U.S. stocks has rested for months on a single number: the forward price-to-earnings ratio, which has actually fallen while the S&P 500 has climbed. A new wave of analysts argues that ratio is misleading, because the cushion it appears to offer is built on a second bubble hiding inside the earnings themselves, a stock market double bubble now being called out by name. BCA Research says equities could fall 30% to 50% once the second layer bursts.

The warning comes from analysts with three different vantage points: a London-based strategy desk, an independent research house, and a former Wall Street chief strategist. They share a single thesis: even as the S&P 500 keeps printing records, the assumptions behind the index’s earnings have quietly drifted far above their long-term trend, and a string of record earnings beats has stretched those assumptions even further. As Q2 earnings season opens this week, that thesis is about to be tested by the very reports the bull case leans on.

Why Bulls Still Lean on the Forward P/E

The forward price-to-earnings ratio is a favourite of fundamental analysts for a reason: it compares today’s share price with the earnings per share the company is expected to deliver over the next 12 months. As earnings expectations climb, the ratio can fall even as the stock price rises, which is exactly what has happened over the past year. The S&P 500’s forward P/E stood at 22.4 a year ago and slipped to 20.51 at Thursday’s close, even as the index itself rose by 20% over the same period, according to Dow Jones Market Data.

To a bull, that arithmetic is comforting. The index is not stretched on the most-watched valuation metric, and the surge in prices has been earned by a faster surge in profits. The case leans on a second number: S&P 500 companies are expected to deliver double-digit profit growth for the seventh straight quarter when second-quarter results land. Analysts have pencilled in bottom-up earnings growth of more than 23% for the period, according to FactSet analysts.

That alibi is now under attack from a small group of analysts who argue both numbers are themselves inflated. Their case rests on a separate valuation measure, one that adjusts for the pace of recent earnings, and on a pattern they say has played out before. If the pattern they describe holds, the alibi dissolves quickly.

Where the Shiller CAPE Says Valuations Sit

Step back from the 12-month forward view, and a different picture emerges. The Shiller CAPE ratio, which divides the S&P 500’s price by its average inflation-adjusted earnings over the previous 10 years, sits at around 41 today, according to the price and earnings bubble warning from Panmure Liberum strategists Joachim Klement and Francisca Reis shared with MarketWatch. That level is approaching the record-high CAPE reading recorded during the dot-com era a quarter-century ago, when prices ran far ahead of slower-moving earnings. The reading is unusual in another way: the underlying earnings themselves have been rising well above their long-term trend.

Earnings per share for the S&P 500 have grown at a pace that is 1.8 standard deviations above their long-term trend, according to the Panmure team, a divergence that has no analogue in modern post-war data. If one were to deflate those earnings back to a normal rate of growth, the Shiller CAPE ratio would swell to 67.6, a level 4.6 standard deviations above the long-term trend. By the strategists’ measure, that adjusted reading would surpass the peak of every other asset bubble in U.S. history. The dot-com peak sat closer to the upper end of normal variance, propped up by more modest earnings growth rather than by both prices and profits moving together. The result is a configuration the strategists describe as a price bubble stacked on top of an earnings bubble.

A reading of 67.6 is what the index would look like if profit growth were normal, and 41 is what it looks like with the recent growth streak baked in. Both are well above the long-term median, and both lean on a single assumption: that the current run of double-digit earnings growth continues. That assumption is the part three other sets of analysts now argue is fragile.

Panmure Liberum’s CAPE-adjusted view is the cleanest expression of the bubble thesis, but it is not the only one in circulation. Two other strategists are making the same call from different angles, and both point to the same mechanism hiding underneath the index’s earnings. Their shared focus is a small group of technology companies that have come to dominate the S&P 500’s profit growth. The shift inside those names is what comes next.

Hyperscalers’ Move From Asset-Light to Asset-Intensive

Five companies do most of the heavy lifting in the S&P 500’s recent earnings, and they have all announced the same kind of bet. Microsoft, Alphabet, Amazon, Meta Platforms and Oracle, the group of cloud and AI infrastructure operators known as hyperscalers, are pouring cash into new data centers to train and run large AI models. Their capital expenditure has been climbing for two straight years, and the depreciation that comes with the new assets will eventually flow through the income statement. In a Financial Times column on the strategy desk’s report, Klement warned that these so-called supernormal profits probably won’t persist forever, and that at some point, investors will likely need to reckon with financial reality. To be sure, Klement added, earnings could keep galloping higher for a few years, since such runs often continue longer than investors expect.

Supernormal profits probably won’t persist forever, and that at some point, investors will likely need to reckon with financial reality.

The hyperscaler capex cycle is the part of the story that other strategists now lean on hardest. As the data-center bills accumulate, the companies shift from asset-light businesses into asset-intensive ones, the kind of structural change that takes years to wash through reported earnings. The transition does not reverse quickly, and the capex bill is what makes the next leg of profit growth the hardest to find. The earnings hit, when it arrives, will hit a market that has been pricing those profits as permanent.

BCA Research’s 30% to 50% Forecast

Peter Berezin, chief strategist at BCA Research, says the same pattern has played out before, and not in the dot-com era. Earnings bubbles formed in banks and home builders during the run-up to the 2007-’08 global financial crisis, Berezin wrote in a report from late May, with a low price-to-earnings ratio disguising an unsustainable uptick in profits each time. He sees the same combination forming again.

Berezin’s note argues that earnings bubbles tend to form in industries subject to boom-bust cycles. The list includes natural resources, airlines, and shippers, and, in the current cycle, semiconductors, the cohort that has powered the AI trade for the past two years. BCA’s third-quarter outlook, shared with MarketWatch last week, says that once the earnings bubble peaks, equities could fall between 30% to 50%. The strategists wrote that Wall Street is terrible at anticipating when that peak arrives. By Berezin’s reading, the peak only becomes visible after the fact.

Wall Street analysts are terrible at anticipating an earnings bubble’s peak.

Berezin’s framework starts from the same observation as Panmure Liberum’s: that low forward P/Es mask what is happening underneath. The metric only looks cheap because the E in P/E has been inflated by the same forces pushing the P higher, the AI capex cycle, the hyperscaler buildout, the semiconductor supercycle. Once that E stops rising, the cushion disappears. The market then has to reprice both the E and the P at the same time.

Until 2008, banks and home builders looked cheap on a forward-earnings basis, too. The cheapness was real, but the earnings behind it were not, and the lesson is not lost on the strategists watching semiconductors today. Whether or not BCA’s 30% to 50% call proves right, the precedent is now in plain view.

More Analysts See Earnings as the Real Bubble

The Berezin call is not the only one circulating. Andy Costan, chief executive of Damped Spring Advisors, said during a May appearance on the Monetary Matters podcast that the U.S. economy is not growing quickly enough to justify the earnings Wall Street analysts have pencilled in. Wall Street veteran Jim Paulsen said in a recent post that he sees risk in exuberant earnings expectations, the same warning Panmure Liberum and BCA are now publishing. The voices are coming from different desks, but they describe the same setup.

Each analyst lands on the same diagnosis by a different path. Panmure Liberum uses an earnings-adjusted CAPE, Berezin uses the historical pattern of profit-led manias, and Costan uses the gap between real-economy growth and earnings growth. They converge on a single risk: that the next leg down in the equity market, when it comes, will not be a routine multiple contraction.

Until now, every prior earnings bubble in U.S. history has come with a low forward P/E that made the index look cheap by the standard yardstick. Berezin’s note flags the industries that have hosted them before, and the list of prior hosts reads as a primer for the current cycle. Each host has shared three traits: heavy capex, rapid demand growth, and an earnings surge that pulls profitability far above its long-term run rate. Today, the AI trade has concentrated the bubble inside a single sub-industry. The roster of past hosts is short, and it tells investors where to look for the next move.

• Banks and home builders (pre-2008 financial crisis)
• Natural resources
• Airlines
• Shippers
• Semiconductors (today’s environment)

The Indexes Near Records as Q2 Opens

None of this has yet registered in the tape. The three major U.S. indexes sit within striking distance of records, according to FactSet data, with semiconductor names leading Monday’s rebound. The cohort BCA and Berezin have flagged, the same group of AI-linked chip stocks that powered the rally through the spring, was at the front of the move. June had tested the trade, with a powerful momentum run in chip stocks hitting a speed bump in the final weeks of the month.

June’s wobble was the first serious test of the AI trade since the spring, and the early-July rebound suggests the buyers came back. The internal composition of the move has changed, though, with the gains narrowing to a handful of large-cap tech names. When an index’s gains rest on one sector, that sector’s earnings carry the whole forward P/E.

The forward P/E that bulls lean on is built on a narrow base, and the broader market has been quieter. Five companies, the same hyperscalers Klement named, account for a disproportionate share of the S&P 500’s earnings growth over the past year. The other 495 names have lagged, and the gap between the five and the rest is what makes the forward P/E figure look more reassuring than it actually is. Overseas chip markets have flashed similar warnings: the chip selloff that tripped Korea’s KOSPI circuit breaker showed the same fear spreading across the AI trade in early July.

The gap between the index level and the broader market is what the next quarter’s earnings reports will measure. The hyperscalers’ data-center spending, the chip cycle, and the AI capex bill all show up in those reports for the first time on a year-over-year basis. Until the Q2 reports arrive, the forward P/E will continue to look cheaper than it is.

• S&P 500: closed less than 1% below its record high (FactSet)
• Dow Jones Industrial Average: record close above 53,000 (FactSet)
• Nasdaq Composite: gained 1.1% on Monday (FactSet)
• S&P 500 forward P/E: 20.51, down from 22.4 a year ago (Dow Jones Market Data)
• S&P 500 price return: +20% over the same 12-month window

The volatility in chip stocks through June was the first signal that the rally has begun to look like a one-way bet. A meaningful break in either direction will hinge on the next batch of quarterly results, the same numbers that test both the bull and the bubble theses at once. The market’s reaction to those prints will tell traders whether the AI trade has staying power or whether the structural cracks in earnings growth are starting to widen. The split between the leaders and the rest of the index has rarely been this wide. Monday’s rebound left the buyers in control, but the move is concentrated in fewer names than the headline suggests.

The Q2 Earnings Test

As Q2 earnings season opens this week, the second bubble in the strategists’ thesis faces its first major test. S&P 500 companies have been expected to deliver bottom-up earnings growth of more than 23% for the period, according to FactSet, a number that assumes the recent run of double-digit profit growth extends for a seventh straight quarter. A print materially below that estimate is the trigger the bubble thesis has set up, the moment the cushion under the forward P/E starts to unwind. The next six weeks of quarterly reports will reveal which side of that bar the actual earnings land on.

Panmure Liberum’s own note left room for the rally to continue, with Klement conceding that earnings could keep galloping higher for a few years and that such runs often outlast investors’ expectations. BCA Research’s third-quarter outlook frames the same window as the moment a peak becomes visible, with the strategist writing that once the inflection arrives, the air comes out of both bubbles at once. Either way, the next six weeks of earnings calls will set the tone, and a parallel rally in Japan, where SoftBank’s rally that echoes Son’s 2000 peak shows the same AI concentration risk, gives a preview of how the trade unwinds when it does.

Disclaimer: This article is for informational purposes only and does not constitute financial advice. Stock market investments carry risk, including the potential loss of principal. Past performance is not indicative of future results. Figures cited are accurate as of publication date. Consult a qualified financial professional before making investment decisions.