Hedge funds offloaded chip stocks for the fourth consecutive week, but continued to invest in AI stocks.

      Hedge funds have reduced their semiconductor stock holdings for the fourth consecutive week, according to data from Goldman Sachs' prime brokerage. This ongoing sell-off has positioned chipmakers and their equipment suppliers as the most heavily net-sold segment in the US market. This trend coincides with some instability in the AI trade and may be viewed by some as the first sign of weakness in a rally that has driven companies like SK Hynix closer to a trillion-dollar valuation. However, the underlying details suggest a more measured approach rather than a market peak.

      The data indicates a retreat rather than a major sell-off. Goldman’s statistics reveal net selling in the sector over eight straight trading days, with semiconductors being the most net-sold US industry subsector tracked by its prime desk during the past four weeks. This reflects a significant and sustained behavioral change rather than a fleeting response.

      Upon closer inspection, this is not an exit strategy. The selling activity mainly involves fund managers reducing existing long positions rather than initiating new bearish stances, which is typically indicative of profit-taking after an extended upward trend. When funds are realizing gains instead of positioning for declines, the emphasis shifts to risk management rather than strong conviction.

      The positioning data further emphasizes this point. Despite four weeks of selling, hedge funds’ net exposure to semiconductors remains at the 98th percentile of the past five years, suggesting it is still near the upper limit of the historical range. It is not feasible to exit a crowded trade within a month, indicating that this trade continues to be quite congested.

      Goldman's assessment indicates that no fundamental shifts have occurred. Their analysts do not perceive the selling as a movement away from the AI theme, with hedge fund positioning in US AI stocks still at historically high levels. While some positions in chips are being reduced, the rationale for holding them remains intact.

      The broader context sheds light on this cautious approach. Rising inflation and increasing bond yields have prompted managers to implement downside protection across their portfolios, and the most apparent area to take some profits, literally, is the sector that has seen the most significant increases. Lowering the most valued positions is a strategy funds employ to hedge against a high-valued market without suggesting it has peaked.

      Recent price movements reflect this sentiment. In the shortened trading week following the holiday, the main index for semiconductor stocks dropped 6.3% in one session and 5.4% in the next, resulting in a pronounced two-day decline. Nevertheless, the overall market still ended higher for the week as capital shifted towards other beneficiaries of AI rather than exiting the theme altogether. This rotation, rather than an exit, has been the focal point.

      This rotation warrants attention in its own right. As established chip stocks have become pricey, some managers are now seeking out the next tier of companies expected to benefit from AI expenditures, including power and cooling suppliers, as well as software firms built on the underlying hardware. The movement of funds away from semiconductors does not equate to leaving the trade entirely; often, it simply indicates a shift to different segments within the industry.

      The concentration of investments presents a persistent concern. When a single theme dominates a portfolio and a few companies drive the index, even orderly profit-taking can become chaotic if numerous funds opt to sell simultaneously. While Goldman's data indicates this isn't currently happening, the same figures that reveal discipline also highlight how much depends on the continuation of this trade.

      Whether this is merely a pause or the beginning of something more significant remains unresolved in the data. The memory and accelerator companies central to the AI expansion, such as Micron and those linked to Nvidia’s memory strategy, carry substantial expectations, and crowded trades can rapidly unwind when circumstances change. For now, funds are opting to trim their winners rather than completely walk away from them.