AI trading stalls! Goldman Sachs warns of the biggest momentum divergence in 5 years, as funds shift from chips to software

AI-related trading is facing the most severe structural test since the beginning of this bull market. Goldman Sachs has warned that an unprecedented split is emerging within the momentum factor—the divergence between short-term and long-term momentum performance has widened to its largest in five years, while the AI sector is undergoing its largest drawdown since the launch of ChatGPT, reshaping the fundamental logic of momentum trading.

In a latest client report dated September 15, Goldman Sachs strategist Guillaume Soria pointed out that the three-month momentum index (GSPRHMO3) surged 5% that day, while the twelve-month momentum index (GSPRHIMO) dropped 6.7%, the largest single-day performance gap in five years. Meanwhile, the Goldman Sachs index tracking AI-related stocks (GSPUARTI) has fallen nearly 45% from its peak, marking the largest decline since ChatGPT debuted.

Goldman Sachs explicitly recommends that portfolios with AI exposure can hedge by buying put options on the medium-term winners basket (GSXUHMOM) or the AI beneficiaries basket (GSTMTAIP). The highly overlapping risk between AI trades and momentum strategies is becoming a major focus for targeted hedging.

Internal Fragmentation of Momentum Factor: Short-Term vs Long-Term Performance Hits Five-Year Extreme

The internal structure of momentum trading is changing rapidly. Data from Goldman Sachs shows that the single-day performance gap between three-month and twelve-month momentum has expanded to the largest in five years, indicating that capital is clearly shifting away from the previously dominant "old momentum leaders."

In terms of positioning, overall exposure to the momentum factor has also declined. On a one-year horizon, current positioning sits at the 41st percentile, a clear retreat from previous highs; however, zooming out to a five-year perspective, exposure is still at the high 88th percentile, suggesting that systemic deleveraging has yet to fully play out.

Meanwhile, the volatility of the momentum factor is significantly higher than the broader market. For instance, the S&P 500 dropped less than 30 basis points that day, yet the momentum factor itself showed sharp internal divergence. This low-correlation, high-volatility state also makes the cost-effectiveness of short-term momentum hedges more attractive.

Capital Rotates from Semiconductors to Software: The Year's "Least Favored" Sector Makes a Strong Comeback

The sector composition of momentum trading is also undergoing significant rotation. According to Goldman Sachs, short-term momentum capital is shifting from semiconductors to software, with software being one of the least favored sectors at the beginning of the year.

This shift is already reflected in relative performance: The Goldman Sachs software versus semiconductor relative performance index (GSPUSOSE) logged its second-largest single-day gain in history. Goldman Sachs argues that if this trend persists, capital flows may gradually be transmitted to six- and even twelve-month momentum baskets, driving future momentum basket rebalancing—manifesting as buying software (including increasing longs and covering shorts) and selling semiconductors.

It is noteworthy that software previously held the largest weight on the short side of the twelve-month momentum index for an extended period. The rapid rebound in software not only indicates a shift in industry style but will further exacerbate the divergence between short- and long-term momentum.

High Correlation between AI and Momentum Faces Breakdown Risk

Although the AI sector and the momentum factor currently remain highly correlated—with one-month to one-year correlation coefficients holding between 90% and 96%—Goldman Sachs warns that this long-standing, tight relationship is starting to loosen.

Goldman Sachs believes that continued weakness in the AI sector will have a profound impact on the composition of the momentum factor. As software gradually replaces semiconductors as the main driver of short-term momentum, the previously tight alignment between momentum and AI may begin to decouple.

Currently, the highest theme exposure for momentum in the US market remains concentrated in AI-related fields. However, if the rotation from semiconductors to software persists, the sector composition of the momentum basket will be further adjusted. For investors, this means that strategies using AI as a proxy for momentum exposure require reassessment, since momentum trading is entering a new phase of risk and return.