Hedge funds have been reducing their exposure to AI stocks, led by semiconductors, memory, and hardware, according to UBS Prime. Despite the shift from overwhelmingly long selling to a more balanced mix of long and short positions, hedge funds have continued to add to their AI hedging shorts. This trend has been observed since early October and has resulted in the sixth-largest re-risking day for US equities year-to-date.

Hedge funds have been focusing on higher-quality names with available compute and stronger build profiles, with hyperscalers being a small net buy. This has resulted in the addition of +1.7 standard deviations of long exposure and +2 standard deviations of short exposure, indicating a largely market-neutral or relative-value risk add rather than a directional long or short bet.

Zain Nizami, Head of Cash Equity Trading at UBS, remains cautious as correlations rise, which could lead to crowded sector and relative-value trades, particularly semis versus software and semis versus hyperscalers, spilling over into broader portfolio de-risking. Nizami recommends buying semis while selling the recent “hedging beneficiaries,” which looks attractive given net exposure in semis is near July and Liberation Day lows, while hyperscaler and software exposures remain high.

Overall, hedge funds have been diversifying their risk through AI hedging shorts, while maintaining a cautious outlook due to rising correlations. This strategy has resulted in a more balanced exposure to the AI value chain and could lead to further additions in the coming days.