Artificial intelligence-linked trades are confronting their most formidable structural test since the onset of this bull market. Goldman Sachs is sounding the alarm over an unprecedented schism within the momentum factor, with the performance dispersion between short-term and long-term momentum widening to a five-year extreme. Meanwhile, the AI sector's deepest drawdown since the launch of ChatGPT is fundamentally reshaping the logic underpinning momentum strategies.

In a client note dated September 15, Goldman strategist Guillaume Soria highlighted that the three-month momentum index climbed 5% on that day, while the twelve-month momentum index tumbled 6.7%, marking the largest single-day performance gap in half a decade. Concurrently, the bank's AI-tracking index has plunged approximately 45% from its peak, representing its steepest decline since ChatGPT was introduced.

Goldman explicitly advises that portfolios holding AI exposure should consider hedging by purchasing put options on the mid-term winners basket or the AI beneficiaries basket. The elevated overlap between AI trades and momentum strategies has emerged as a key focus for hedging, given the associated risks.

The Internal Fracturing of Momentum: Short-Term vs. Long-Term Divergence Hits Five-Year High

The internal architecture of momentum trading is undergoing rapid transformation. Goldman's data reveals that the single-day performance gap between three-month and twelve-month momentum has reached its widest point in five years, signalling a clear exodus from the previous leaders that dominated the momentum landscape.

From a positioning perspective, overall momentum factor exposure has declined. On a one-year horizon, current positioning sits at the 41st percentile, a notable retreat from prior levels. However, extending the view to five years, positioning remains elevated at the 88th percentile, suggesting that systematic deleveraging has yet to fully run its course.

Additionally, volatility within the momentum factor is markedly higher than the broader market. While the S&P 500 fell by less than 30 basis points on the day, momentum experienced dramatic internal divergence. This environment of low correlation and heightened volatility renders short-term hedging of the momentum factor increasingly cost-effective.

Capital Shift From Semiconductors to Software: The Year's Most Avoided Sector Stages a Comeback

The sector composition of momentum trades is also witnessing noticeable rotation. Goldman's report indicates that short-term momentum capital is migrating from semiconductors to software, which was among the least favoured sectors at the start of the year.

This shift is already evident in relative performance: Goldman's software versus semiconductors relative performance index recorded its second-largest single-day gain in history. Should this trend persist, Goldman believes capital flows could gradually influence six-month and even twelve-month momentum baskets, driving future rebalancing that entails buying software, through both long additions and short covering, while simultaneously selling semiconductors.

Notably, software has long held the largest short position weight within the twelve-month momentum index. Its rapid rebound not only signals a sector style rotation but also serves to amplify the divergence between short-term and long-term momentum.

The High Correlation Between AI and Momentum Faces the Risk of Decoupling

Although the AI sector and the momentum factor still exhibit an exceptionally high correlation, with coefficients ranging between 90% and 96% across one-month to one-year horizons, Goldman cautions that this long-standing tight bond is showing signs of strain.

Goldman contends that sustained weakness in the AI sector will have profound implications for the composition of the momentum factor. As software gradually supplants semiconductors as a key driver of short-term momentum, the previously aligned trajectories of momentum and AI may progressively diverge.

Currently, the themes with the highest momentum exposure in the U.S. market remain concentrated in AI-related domains. However, if the rotation from software to semiconductors persists, the sector makeup of momentum baskets will undergo further adjustment. For investors, this implies that strategies which previously relied on AI as a proxy for momentum exposure require reassessment, as momentum trading transitions into a new phase of risk and reward dynamics.