Capital Economics predicts AI spending will fall 10% in 2028, weighing on stocks

AI investment will keep growing in 2027 before falling about 10% in 2028, Capital Economics Chief Economic Adviser Vicky Redwood predicts. Her forecast, reported by Fortune on November 8, puts stocks at risk before spending actually retreats—but stops short of predicting an abrupt investment collapse or a major U.S. economic downturn.

In a note issued Friday, Redwood questioned whether revenue can grow fast enough to sustain the projected spending. She expects the S&P 500 to finish 2027 roughly 12% below its level at the time of the forecast, with larger declines in technology-heavy indexes.

The revenue hurdle behind the spending forecast

Fortune puts expected spending by a handful of hyperscalers—the largest cloud operators—at $800 billion in 2026, twice the previous year's amount. The projected bill reaches at least $1 trillion in 2027. Many forecasts assume annual capital expenditure, or spending on long-term assets, will stay above that threshold for several years.

Redwood calculates that annual AI-related revenue from external customers would need to grow roughly tenfold by 2030 to justify the projected investment. That is a revenue requirement in her analysis, not a forecast that customers will actually spend that much.

She considers that growth possible, but says it would require faster and more severe economic disruption than is currently visible. Existing businesses and workers would face substantial displacement. Her spending forecast therefore assumes a slower expansion in 2027, followed by an outright decline, rather than several more years at the projected pace.

Stocks would react before investment falls

The market forecast brings the consequences forward. Redwood expects investors to price in the coming spending decline before it happens. Her end-2027 S&P 500 target would leave the index 17% below its early-October peak, not just below its level when she made the forecast.

Falling shares could then weaken consumer spending through the wealth effect: households spend more when their investments make them feel wealthier. Redwood estimates that the market rally over the past year lifted U.S. gross domestic product by around 0.4%. She expects a selloff to create an economic drag of similar size.

Weaker investment would have a separate effect. Redwood estimates that higher AI spending added 0.3 percentage point to U.S. GDP growth in 2025. A slowdown could remove a similar contribution from the recent growth rate, while a 10% spending decline could reduce GDP by about 0.1%.

Her regional outlook is harsher for the more AI-dependent Asian economies. She expects the slowdown to leave Taiwan's real GDP growth about 3 percentage points lower in 2027 and South Korea's about 1 point lower, with a further drag in 2028.

The bottom line for the U.S. economy is that we expect slower GDP growth in 2028, but not a major downturn.

Vicky Redwood, Capital Economics chief economic adviser, quoted by Fortune

A smaller buildout can still support wider use

Redwood's forecast separates the spending cycle from the technology's usefulness. She expects AI adoption to keep expanding, supported by infrastructure already built and additional investment still to come. She estimates that AI will raise U.S. productivity by about 1% a year for several years.

Even after a decline, capital spending could remain relatively high. Cloud operators would still need to replace depreciating equipment, capture the cost advantages of operating at scale and stay competitive in an AI market Redwood describes as winner-takes-most. Companies could also spread investment reductions over several years rather than cut spending all at once.

Redwood also notes that falling capital spending alone has generally not been enough to cause a recession, although downturns have followed some earlier investment booms and busts. Her conclusion, quoted by Fortune, is that a more sustainable investment pace can still add enough capacity to support further growth in AI usage.