Wall Street Sees AI Capex Peak Dragging S&P 500 Earnings Growth From 36% to 11% by 2027

Goldman Sachs, Raymond James, and Ned Davis Research converge on the same signal: the AI capital-expenditure cycle that powered 2026's earnings boom is peaking, and the deceleration will compress valuations into 2027.

The four-year bull market has been an earnings story, and the earnings story has been an AI-capex story. In 2026, the S&P 500 posted a blended Q3 earnings growth rate of 31.4% year-over-year — a figure that makes the 1989–2024 historical range look modest. Amazon, Alphabet, Meta, Microsoft, and SpaceX are on track to spend over $800 billion in combined capital expenditure this year, a record. That torrent of spending has flowed directly into corporate revenue lines across the tech stack and beyond.

Now, three major research desks are saying the same thing: the spending peak is here, and earnings growth is about to decelerate hard. Goldman Sachs projects S&P 500 earnings growth dropping from 36% in 2026 to 11% in 2027. Raymond James sees EPS growth falling from 28% to 23%, with the year-over-year deceleration becoming "more severe" after Q1 as AI-related comparables get tougher. Ned Davis Research calls the 2027 earnings outlook "questionable" and flags a slight decline in the share of companies beating expectations — 84% in Q3 versus 86% last quarter — as a sign of "excessive" optimism. The mechanism is straightforward. Hyperscaler capex drove revenue for chipmakers, cloud providers, data-center builders, and their supply chains. As that spending plateaus or declines, the earnings impulse fades. Meanwhile, the costs associated with the AI boom — interest expenses, infrastructure maintenance, depreciation on massive capital outlays — continue to climb. The margin squeeze is arithmetic, not speculative. Goldman's team, led by Peter Oppenheimer, notes that stock valuations have already begun declining, signaling the market is pricing in slower economic and earnings growth ahead. But valuation compression in a decelerating-earnings environment is a feedback loop: lower growth expectations reduce the multiple investors will pay, which reduces returns, which reduces the wealth effect, which slows the real economy. Ned Davis explicitly warns that the bull market has lasted an "abnormally" long time and that stocks could face a topping risk around mid-2027. Layered on top of the earnings deceleration is a set of acute stressors: the Iran war pushing energy prices higher, inflation re-accelerating (October CPI data is raising concerns), and the sudden departure of Jensen Huang as Nvidia CEO amid a scandal. The Nvidia disruption is particularly relevant because it introduces uncertainty at the exact company that has been the earnings-growth bellwether for the AI trade. Tech stocks have already shown unusual volatility, with investors simultaneously worried about too much and too little data-center spending. The structural question is whether 2026's earnings boom created real, durable productive capacity or simply pulled forward demand. If the $800 billion-plus in hyperscaler capex built infrastructure that generates sustained revenue and efficiency gains, the deceleration is a normal cyclical downshift. If much of it was speculative over-investment in AI capacity that won't be fully utilized, the hangover could be worse than the consensus forecasts suggest — more of a 2028 problem, as Raymond James hints. For now, Wall Street's consensus is not a crash call. Goldman still expects stocks to rise, just with more volatility. But the convergence of three independent research desks on the same deceleration thesis, combined with geopolitical risk, inflation concerns, and leadership chaos at the sector's most important company, makes the setup for 2027 materially different from the ride up.