AI Momentum, Hedge Funds, and De-Risking: Goldman Sachs Exchanges Recap

On this episode of Goldman Sachs Exchanges: The Markets, host Chris Hussey speaks with Vinny Lin, co-head of Prime Insights and Analytics within Global Banking and Markets at Goldman Sachs, about what hedge funds are actually doing in AI-related stocks as of late July 2026.

Lin explains that his team analyzes trades and positioning that settle on Goldman Sachs’ prime brokerage platform, giving them near real-time, aggregated visibility into hedge fund behavior. Against that backdrop, he argues that the AI trade has just gone through a very sharp momentum unwind, especially in technology names.

According to Lin, tech has been "ground zero" in a violent momentum drawdown that occurred even while the S&P 500 stayed relatively flat. The episode focuses on whether this is a sign of fading conviction in AI or a positioning reset that could set up the next phase of the cycle.

Lin’s main thesis is that artificial intelligence remains, in his words, "among the biggest tech cycles we ever see in our lifetime," but the trade became too crowded and was forced into a de-risking phase.

He highlights that, by early June 2026, Goldman Sachs’ data showed that fundamental equity long/short hedge funds were generating essentially all of their positive alpha from the AI infrastructure trade. That concentration made the trade vulnerable when sentiment turned.

Within that context, the recent episode is framed less as a collapse in AI fundamentals and more as:

• A sharp reduction in exposures after extreme crowding.
• A rotation within tech (from hyperscalers to semiconductors to hardware and infrastructure solutions).
• A move from record or near-record positioning toward more "middle-of-the-pack" levels.

Lin characterizes this as a "healthy reset, not a complete loss in fundamental conviction," assuming conditions described in late July 2026.

Lin cites several specific datapoints from Goldman Sachs’ prime-brokerage analytics to support his view of the AI and momentum unwind:

• The price of their high-beta momentum basket was down 32% from its highs coming into the week of the interview, leaving it only up 16% year-to-date after being up more than 60% in June.
• The TMT (technology, media, telecom) momentum long/short basket was down almost 40% from its highs, which Lin says would be the sharpest drawdown in five years.

On positioning, he notes:

• Global semiconductor net allocation by hedge funds started the year at 10% (meaning $10 in semis for every $100 in global equity net exposure).
• That allocation more than doubled to 20%, peaking as high as 24% in June, which Lin calls the highest level on their record.
• By the time of the interview, that had been cut back to 18% — a clear reduction from the peak, but still well above where the year began.

Lin adds that the realized volatility of the momentum factor over the prior three months had risen to the highest level in the last 45 years, outside recessions, making the de-risking as much about risk control as valuation.

According to Lin, multiple forces contributed to hedge funds stepping back from AI and momentum trades in mid-2026.

First, crowding became extreme. The AI infrastructure complex had grown to dominate hedge fund alpha generation, and global semiconductors reached record allocation levels. Lin argues that this left the trade highly vulnerable to any shift in sentiment or volatility.

Second, leverage and gross exposures were stretched. He notes that, six weeks before the interview (early June), equity fundamental long/short funds’ gross exposure sat at five-year highs, with net exposure at four-year highs. After the unwind, both gross and net exposures had fallen to roughly the 60th–65th percentile relative to the prior three years — "middle of the pack" rather than washed out.

Third, Lin points to broader market mechanics:

• Increased participation from retail investors.
• The proliferation of leveraged ETFs adding more embedded leverage.
• A "heavy dose" of capital markets issuance.
• Technical events such as multiple index rebalancings, pension rebalancings, and monthly and quarterly options expirations.

In his view, these elements amplified the impact of repositioning in AI-related names.

Lin acknowledges several factors that temper the bullish AI narrative and complicate hedge fund positioning.

Market concentration is one. Hussey notes that many investors are reluctant to pay hedge fund fees "to invest in the biggest stocks in the world." Lin responds that, despite elevated concentration, 2026 had still been a strong alpha environment by June, with average hedge fund performance around 9% for the first half, which he says is among the best halves in 20 years.

He attributes this to:

• "Single stock volatility" being "through the roof" even as index volatility stayed contained.
• Multi-year lows in stock correlations.
• Significant sector rotations beneath the index level.

At the same time, the spike in momentum volatility and the simultaneous reduction in both AI longs and macro hedges (what Lin calls textbook "de-grossing") means portfolios could be more exposed if correlations rise again. In his telling, this dynamic makes risk management, not just stock selection, central to how hedge funds navigate AI-related trades.

Lin outlines a few key forward-looking signposts he is watching as of late July 2026.

First, he mentions that, toward the end of the prior week, the momentum factor started to stabilize, and Goldman Sachs’ data showed buying activity resurfacing over the last three trading sessions. This suggests to him a possible early stage of "buying the dip" in certain AI-related names.

He identifies his favored expression of that theme as US AI infrastructure equipment names, arguing valuations had fallen back toward roughly two-year lows. However, he indicates that, given the heightened volatility backdrop, he believes a "limited loss" structure such as call spreads in options may be a better way to express that view than outright stock purchases.

Looking forward, Lin says he is focused on:

• Upcoming mega-cap tech hyperscaler earnings.
• The trajectory of their AI capital expenditure plans for the rest of 2026 and into 2027.
• Evidence that these AI capex investments are translating into incremental revenue and profit growth.
• Whether stock correlations start to rise again after a period of de-grossing and reduced macro hedging.

All of these, in his view, will shape whether the AI trade resumes leadership or faces another round of risk reduction.