Hedge Funds Erase Two Years of AI Stock Buying in Five Days, Goldman Data Shows
Key Takeaways
- •Hedge funds unwound all net AI stock purchases made since the start of 2025 within five trading sessions, from October 8 through October 14, per Goldman Sachs Prime Brokerage data.
- •The five-day net-selling streak was the second largest in Goldman's historical records, behind only the July crash, and funds cut about 7.8% of total equity exposure.
- •Memory chipmakers Samsung, SK Hynix, Micron, and SanDisk lost more than $700 billion in combined value in under two weeks, with chip stocks falling 15.6% in October while the S&P 500 slipped less than 1%.
- •Asian markets were hit hardest, with Korea entering bear-market territory and triggering its circuit breaker, and Taiwan recording its largest single-day foreign outflow on record at about $9.3 billion.
- •Gross leverage in Goldman's prime brokerage book remains near five-year highs at 310.3%, prompting Goldman to warn of further deleveraging ahead, while JPMorgan leans toward viewing the dip as a buying opportunity.

Hedge funds unwound every net purchase of artificial-intelligence stocks made since the start of 2025 in just five trading sessions, according to Goldman Sachs Prime Brokerage data — erasing roughly two years of accumulated buying in the sector within days.
The bank's prime brokerage desk recorded net selling of global AI stocks from October 8 through October 14, with funds offloading about $20 billion of technology shares over a broader two-week stretch. Prime brokerage desks provide financing and trading services to hedge funds, which makes the unit's flow records a direct gauge of institutional positioning. Most of the exit came from long positions being cut rather than fresh bets against the sector: selling ran at roughly 3.5 times the pace of new short positions. Goldman cautioned that the deleveraging may not be finished.
Selling Pace Ranks Second Largest in Goldman's Records
The five-day net-selling streak was the second largest in Goldman's historical data, behind only the July crash. Global equities broadly saw heavy net selling as well, reaching -2.1 standard deviations in a single day. Long positions led the exits, while short flows stayed near a low of about -13%.
Goldman's basket of hedge fund favorites fell 4.2% on October 22, its worst single day since 2011. Across all stocks, funds cut about 7.8% of total exposure — a drop Steno Research said was the biggest in Morgan Stanley data since 2016.
Coin Bureau, a cryptocurrency-focused media account, highlighted the figures on X:
HUGE: Goldman data shows hedge funds wiped out ALL their AI buying since 2025 in just 5 days. The bank's prime brokerage desk saw them net sell global AI stocks for five straight sessions, reversing all their AI buying since the start of 2025. They also sold about $20 BILLION… pic.twitter.com/6KmES4ofZk
— Coin Bureau (@coinbureau), November 1, 2026
Despite the pullback, hedge funds remain heavily positioned over longer horizons. Net allocation to AI stocks declined from 31.7% in early June to 26.1%.
Rachel Hu of Goldman Sachs Asia said trading feedback from Taiwan pointed to a wave that was less of a “capitulation” than July's, and she indicated that more selling pressure may still follow.
Asian Markets and Chipmakers Bear the Brunt
Korea's benchmark Composite Stock Price Index entered bear-market territory, conventionally defined as a decline of 20% or more from recent peaks, and Seoul triggered its market circuit breaker. Taiwan recorded its largest single-day foreign outflow on record, about $9.3 billion, and several Asian central banks intervened in currency markets.
Memory chipmakers Samsung, SK Hynix, Micron, and SanDisk — the companies that supply much of the memory and storage hardware used in AI infrastructure — lost a combined more than $700 billion in under two weeks. Chip stocks fell 15.6% in October while the S&P 500 slipped less than 1%, and the Philadelphia Semiconductor Index posted a seventh straight daily decline.
Eight of 11 U.S. sectors saw net selling, led by information technology and industrials. Hedge funds bought healthcare and consumer staples over the same period — sectors whose revenues are typically less sensitive to economic swings — a move Goldman described as a “marginal shift toward defensive positioning.”
Leverage Levels Keep Deleveraging Risk Alive
Gross leverage in Goldman's prime brokerage book fell only 0.4 percentage points to 310.3%, still near five-year highs, while net leverage dropped 1.5 points to 73.3%. Goldman noted that with gross leverage this elevated, the potential sellable size exceeds what net exposure alone suggests — one reason the bank sees further deleveraging ahead. How quickly that gross figure comes down is one measure of how much selling capacity still sits in the system.
Rich Privorotsky of Goldman's Delta-One business described the dilemma in an overnight report. “In hindsight, the painful trade was going long AI and short bonds,” he wrote, adding that demand for computing power “can grow exponentially, but the capital available to finance it cannot.”
Analysts remain split on the outlook. JPMorgan leans toward viewing the dip as a buying opportunity, while Goldman is more cautious, citing record foreign outflows in Asia and elevated leverage across the system. Positions built in 2023 and 2024, meanwhile, remain intact on hedge fund books.
Originally reported by Blockonomi.