Analysis
Wall Street forecasters are publicly split on whether AI-driven equity gains constitute a "late-stage bubble" nearing its end, Fortune reported, with the disagreement sharpening just as the Federal Reserve weighs a rate decision that historically has been the trigger for prior boom cycles to unwind.
Capital Economics, led by senior markets economist James Reilly, has spent months building the most detailed bearish case in the market: a forecast for the S&P 500 to close 2026 at 8,250, up roughly 7.7% from current levels, before falling 21% to 6,500 by the end of 2027. That's a genuinely unusual call -- predicting further near-term gains followed by a sharp reversal -- and it stands in direct opposition to Morgan Stanley, whose wealth-management CIO Lisa Shalett has the index at 8,000 by year-end and 8,300 by mid-2027, implying continued, if slower, upside.
The valuation data sits in between those two stories rather than clearly supporting either. The S&P 500's forward price-to-earnings ratio has actually compressed this year, from 22.5x in January to roughly 19.5x now, even as the index has climbed -- a sign that earnings growth, running above 30% year-over-year, has partly kept pace with price gains rather than valuations expanding on hope alone. Core CPI inflation hit a new post-pandemic low of 2.4% in August, the kind of data point that normally supports further Fed easing rather than tightening.
“Core CPI inflation hit a new post-pandemic low of 2.4% in August, the kind of data point that normally supports further Fed easing rather than tightening.”
Yet UBS expects the Fed's September vote to still land as a quarter-point hike to a 3.75%-4% range, with the increase splitting the Fed governors roughly 10-2 -- Christopher Waller and Michelle Bowman dissenting -- a genuinely contested internal decision rather than a consensus move. Behavioral economist Owen Lamont of Acadian Asset Management has pointed to single-day market swings this year as evidence sentiment, not fundamentals, is driving a meaningful share of AI-linked stock moves: on July 30-31 alone, Microsoft gained $450 billion in market value while Apple lost $360 billion, and Amazon added $388 billion while Meta shed $102 billion -- swings of that magnitude in single trading sessions are not something stable, fundamentals-driven valuation typically produces.
The historical parallel getting cited most is 1999-2000, when the Fed raised rates from 4.75% to 6.5% between June 1999 and May 2000, a tightening cycle that preceded an S&P 500 decline of nearly 50% as the dot-com bubble unwound. Reilly's argument leans on that precedent: rate increases into an already-stretched, narrow-leadership market have historically been the mechanism that ends the cycle, not a slow fade.
What the debate misses is that Capital Economics and Morgan Stanley aren't actually disagreeing about next year -- both expect the S&P to rise through the rest of 2026. The disagreement is entirely about 2027, a full year out, which means neither firm's public forecast is actionable evidence of anything happening in markets today. A bearish 15-month-out target from one shop and a bullish one from another is normal forecasting variance dressed up as a binary bubble call.
The earnings growth and CPI data are the more useful real-time signals than either firm's multi-quarter target: AI-linked megacap earnings are still growing fast enough to be compressing forward multiples even as prices rise, which is a materially different setup than 1999's pure multiple expansion. Whether that holds through another two or three quarters of AI capex spending -- without a slowdown in the underlying revenue growth funding it -- is the actual variable to track, not which economist's price target sounds more dramatic.
The next concrete test is the Fed's September decision itself: a 10-2 split vote with two sitting governors dissenting publicly is a signal the committee itself doesn't have conviction on the rate path, and how markets react to that ambiguity over the following weeks will say more about bubble risk than any single forecaster's 2027 target.
