Hedge funds have been heading for the exits on AI stocks. According to Goldman Sachs prime brokerage data, funds sold AI-linked stocks at the second fastest pace in a year, and their net leverage dropped at the same time.
What Goldman is seeing
Goldman’s data from July 2026 showed hedge funds had been net sellers of US tech. Their collective exposure shrank by approximately 10% in market value since early June. Goldman described that as the most substantial retreat it had recorded in over a decade.
Funds were net sellers in six of the eight weeks leading up to mid-July 2026. The pain was spread across most of the tech complex. Selling concentrated in hardware, storage and peripherals, IT services, semiconductors and software.
A rough summer for the smart money
July 2026 marked the worst monthly underperformance for hedge funds relative to the S&P 500 in over 20 years, a slump linked to their AI positions.
Goldman strategists acknowledged significant length reduction within hedge fund portfolios. The bank also flagged early signs of capitulation in some positions. Still, Goldman framed the retreat as a tactical adjustment rather than a wholesale exit from tech. Funds were trimming broad exposure to technology and chips while holding on to select mega-cap names.
Leverage down, conviction intact
Net leverage and AI exposure fell from their Q2 2026 peaks, according to Goldman, but both stayed above historical averages.
By October 2026, Goldman noted that the narrowing in these trades had continued and observed historically low net leverage across various strategies. By October 2026, exposure to the “Magnificent Seven” stocks had rebounded to roughly 22% of hedge funds’ US equity exposure, even as overall tech positioning eased.
Goldman strategists indicated that core conviction in AI positions remains strong for many investors.
What this means for investors
When large holders sell broadly across semiconductors, software and hardware at once, smaller AI names without mega-cap balance sheets can take outsized hits. The latest data suggests funds see more risk in the mid-tier of the AI trade than at the very top.
Funds rebuilding Magnificent Seven exposure to around 22% while trimming everything else points to a flight toward perceived quality. With funds holding less risk, they have more dry powder to buy if AI stocks find their footing.
The signals to watch are whether the selling pace keeps accelerating from the second fastest in a year, whether net leverage keeps sliding toward its historical average or below it, and whether funds eventually start trimming the Magnificent Seven alongside the chipmakers and software firms.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.