Goldman Sachs warns top clients: AI momentum trading shows unprecedented cracks, recommends hedging
Goldman Sachs has issued a rare warning to top clients: deep structural cracks are emerging in AI momentum. The AI-themed basket has dropped nearly 45% from its peak, with the one-day performance gap between short-term and long-term momentum reaching a five-year high. Capital is accelerating its shift from semiconductors to software. Goldman Sachs explicitly recommends that investors with AI exposure start hedging.
Goldman Sachs is issuing a warning to its top-tier sales and trading clients: The AI momentum structure that has supported this decade’s strongest trading theme is showing deep cracks, and investors with related exposures are advised to start hedging.
According to the latest report sent by Goldman Sachs strategist Guillaume Soria to top clients, on September 15, the 3-month momentum factor (GSPRHMO3) surged 5% in a single day, while the 12-month momentum factor (GSPRHIMO) fell 6.7% on the same day—a performance gap that marks the largest in five years. Meanwhile, the Goldman Sachs AI theme basket (GSPUARTI) has dropped nearly 45% from its peak, making it the most severe drawdown since the launch of ChatGPT. Goldman Sachs believes that if this trend continues, momentum factors will begin to decouple from the AI theme.
On the market impact front, Goldman Sachs bluntly states that in the short term, it remains “reasonable” to hedge momentum factors—especially as momentum factor volatility is significantly higher than that of the broad-market index. On the day, the S&P 500 Index fell less than 30 basis points, yet momentum factors experienced intense internal swings. Goldman Sachs advises investors with AI exposure to hedge via buying put options on the Medium-Term Winner Basket (GSXUHMOM) or the AI Beneficiary Stocks Basket (GSTMTAIP). The indicative cost for a one-month put option with a strike price at 95% of current price on GSTMTAIP is about 2.02%, with a 27-day term.
Historic Split Within Momentum Factors
The core warning in the Goldman Sachs report is that the performance divergence between short-term and long-term momentum has reached extremely abnormal levels. The single-day difference between the 12-month and 3-month momentum factors’ performance is the largest in five years, indicating that the market is rapidly abandoning former long-term winners in favor of new short-term leaders.
From a positioning perspective, the concentration of positions in the momentum factor has already decreased—currently at the 41st percentile over a one-year lookback, down notably from previous levels. But over a five-year lookback, position concentration remains at a high 88th percentile, meaning there is still systemic unwinding pressure that has not been fully released.
Goldman Sachs points out that the correlation between the momentum factor and the AI theme remains extremely high, with coefficients maintained between 90% and 96% across periods from one month to one year. However, this high level of binding makes the continued decline in AI especially impactful on the structural integrity of momentum factors.
Capital Rotation: From Semiconductors to Software, Landscape Redefined
Another important change within the momentum factor is a reversal in sector rotation. Short-term momentum is currently exiting the semiconductor sector in favor of the software sector—which was among the least favored at the start of the year. Goldman Sachs data shows that the software-versus-semiconductor performance metric (GSPUSOSE) recorded its second-best single-day performance in history.
Goldman Sachs believes this rotation has spread from the 3-month to the 6-month momentum, and if selling pressure persists, it is expected to further transmit to the 12-month momentum. Looking at the likely direction of upcoming rebalancing for momentum baskets, Goldman Sachs forecasts increased long exposure and short covering in software, as well as reductions in semiconductor holdings, structurally reshaping the composition of the momentum factor.
It is worth noting that previously, the software sector was the largest weight in the short leg of the 12-month momentum factor. Its rebound is also exacerbating the pressure on long-term momentum factors.
Goldman Sachs Advises Top Clients to Initiate Hedges
The target audience for this warning itself sends a message. It is reported that this Goldman Sachs report was not distributed to the regular sell-side lists, but rather was selectively sent to the bank’s highest-priority sales and trading clients, carrying much higher priority than usual research channels.
In terms of specific trading recommendations, Goldman Sachs suggests that clients with AI exposure in their portfolios consider hedging using put options on GSXUHMOM or GSTMTAIP. For example, the indicative cost for a one-month put option on GSTMTAIP with a 95% strike price is about 2.02%, for a 27-day term—a manageable hedging cost.
Goldman Sachs also emphasizes that the current volatility in the momentum factor is already significantly higher than the broad-market index volatility, a divergence which itself constitutes a reasonable basis for hedging. With the S&P 500 Index falling less than 30 basis points that day, the dramatic internal volatility of the momentum factor underscores its current vulnerability.
Disclaimer: The content of this article solely reflects the author's opinion and does not represent the platform in any capacity. This article is not intended to serve as a reference for making investment decisions.
You may also like
Tonight, a "dovish rate hike"?
The Federal Reserve is almost certain to raise interest rates tonight, but the key issue is "what will be said after the hike." Citi characterizes this move as a "fine-tuning" adjustment, suggesting there is no inevitable future rate hike, yet warns that if Chairman Powell does not provide clear forward guidance, it will trigger significant market volatility. Goldman Sachs bluntly stated that there is insufficient economic foundation for this rate hike, with inflation being merely a one-off factor, and expects this to be a "signal-less rate hike."
JPMorgan: "Open source disruption" and "AI safety" are not issues, there is still room for capital expenditure in the next two years, semiconductor equipment will become the "new bottleneck"
JPMorgan believes that open-source models are not a threat, regulatory disruptions are only short-term, and cloud vendors’ leverage remains low—the fundamentals of computing power investment have not changed. It forecasts that the capital expenditure of the seven major tech giants will soar from $443 billion in 2025 to $1.577 trillion in 2027, with semiconductor equipment becoming the core bottleneck of the supply chain and a new round of price increases expected in wafer foundry and advanced packaging.
Micron executive: Storage determines AI limits, substantial new capacity will come after 2028
Micron executive Sumit Sadana stated that memory bandwidth and capacity have become the core factors determining the performance ceiling of AI systems. In the face of structural imbalances between supply and demand, Micron expects to increase capital expenditure to over $45 billion in fiscal year 2027, but due to process complexity, substantive new capacity will not be released until 2028. Additionally, long-term supply agreements are reshaping the industry's business model, and humanoid robots will trigger the next wave of massive demand.
Crypto prices
More
