Key Takeaways:
The AI trade that powered two years of equity gains is fracturing, with nearly 6 in 10 S&P 500 technology stocks now in bear market territory.
The S&P 500 technology sector is undergoing its sharpest internal correction since the pandemic, with 59% of its constituents falling more than 20% from 252-day highs, Goldman Sachs data shows.
"The current selloff is a controlled de-risking driven by systematic and factor flows, with core pressure concentrated in crowded AI, semiconductor and memory exposures," UBS's trading desk said in a note.
Goldman's momentum factor pair — which goes long AI winners and shorts perceived AI losers — has dropped 27% from its June 22 peak, with the five-day drawdown the fastest since the pandemic. The Philadelphia Semiconductor Index's constituents broke below their 50-day moving average for the first time since April, a technical signal that has historically preceded further weakness. Momentum factor exposure has retreated from its peak to the 60th percentile of the past year, Goldman data shows.
The divergence matters because AI concentration has masked a broader deterioration in technology breadth. With 59% of S&P 500 tech stocks already in bear territory, the risk is that continued momentum unwinding spills beyond AI names into the broader market. The next test comes as investors reposition ahead of key supply-chain events and reassess which parts of the AI value chain offer the strongest earnings visibility.
AI and Momentum Become Nearly Indistinguishable
The correlation between the momentum factor and AI-related stocks has exceeded 95%, according to Goldman's Marquee platform, meaning the two trades are effectively the same. The sharp drawdown reflects concentrated AI exposure unwinding — a dynamic Morgan Stanley's quantitative and derivatives strategy team described as the fastest 14-day momentum reversal in a decade.
This structural overlap explains why the selloff has been so rapid despite relatively calm aggregate index volatility. The S&P 500's VIX remains below 20, a level that historically signals orderly markets, even as AI-related volatility has surged to its highest since the pandemic. The divergence between headline index calm and factor-level stress is a hallmark of concentrated positioning unwinding rather than broad-based selling.
Rotation From Chips to Cloud Accelerates
Investors are increasingly rotating within AI itself. A strategy of going long hyperscale cloud providers and short semiconductor stocks has emerged as one of the most discussed trades on Wall Street this summer, reflecting a reassessment of where value accrues in the AI supply chain. As questions mount over the marginal returns on compute spending, cloud infrastructure operators with direct exposure to AI application deployment are seen as having more predictable earnings than upstream chip suppliers.
The rotation is visible in sector-level flows. While semiconductor names have borne the brunt of the momentum unwind, cloud and data center stocks have held up relatively better, suggesting capital is being reallocated rather than withdrawn from the AI theme entirely.
One contrarian signal bears watching: technology company insiders have stepped up their buying during the selloff. Historical data shows insider purchases rarely coincide with major market tops, though their timing is not always precise.
This article is for informational purposes only and does not constitute investment advice.
