Hedge funds have been trimming positions in U.S. semiconductor companies following the sector’s outsized gains, choosing to secure profits while still maintaining strong exposure to artificial intelligence investments, according to Goldman Sachs data referenced by Bloomberg on Thursday.
Information from Goldman’s prime brokerage unit reportedly indicated that semiconductor and semiconductor equipment stocks were the most heavily net-sold U.S. subsector over the past month. The activity was driven primarily by investors reducing long exposure rather than building significant short positions against chipmakers.
That shift has pushed the sector into a net-selling position for the year to date.
The profit-taking follows a steep rally across AI-linked chip stocks. Goldman Sachs’ AI semiconductor basket has outperformed the S&P 500 by more than 50% in 2026, while the broader index itself gained over 18% between late March and a recent three-day pullback.
South Korea’s Kospi index — often used as a gauge of worldwide demand for AI infrastructure — briefly rose above 8,000 points for the first time in mid-May after climbing more than 80% year to date before later retreating.
Goldman’s prime brokerage team reportedly said the recent moves reflected portfolio adjustments rather than declining confidence in the AI trade. Exposure to U.S. artificial intelligence shares within the bank’s technology, media and telecommunications basket remains near all-time highs.
Meanwhile, hedge funds have expanded short positions in broad equity indices and ETFs as a hedge against broader market risks, with those bearish positions now sitting at their highest levels in about ten years.
Goldman analysts added that gross leverage across hedge fund portfolios reached a new five-year high this month, while net leverage stayed relatively unchanged — a setup the bank said does not resemble the kind of speculative frenzy currently evident among retail traders.