After $20 Billion Withdraws from TMT, Goldman Sachs Trading Desk Recommends Selectively Buying the Dip in AI: Focus on Four Major Directions

Goldman Sachs FICC and Equities Trading Desk’s latest assessment believes that the sharp correction in tech stocks over the past two weeks is essentially a hedge fund-led deleveraging event, rather than a sign of weakening AI demand—this judgment directly determines whether it is worth entering the market now and how to do so.

According to information from the "Chasing Wind Trading Desk", Goldman Sachs FICC and Equities Trading Desk released a report on October 26, stating that the net sell scale of the entire TMT market in the past two weeks has accumulated to about $20 billion, approaching the historical peak of $23 billion in two weeks during this year’s July AI sell-off. Meanwhile, Goldman’s risk appetite indicator posted one of the largest single-day declines in history, dropping to its lowest level since April last year; the implied volatility spread between the Nasdaq 100 and the S&P 500 widened to a five-year high at the 98th percentile. The Goldman Sachs trading desk concludes that selling pressure is concentrated at the positioning level rather than the fundamentals, crowded positions have largely cleared out, and now is the time to selectively re-enter AI positions.

Goldman Sachs recommends reopening AI exposure in the following order: first Power, then Hyperscalers, followed by Inference and Cybersecurity. Additionally, the report explicitly lists verification criteria for the judgment of “deleveraging rather than demand weakening,” with a verification window from October 28, 2026, to November 6, covering the FOMC rate meeting, Microsoft, Google, Amazon, Intel earnings week, and Oracle Analyst Day.

A Positioning Event, Not a Demand Event

Goldman Sachs’ prime brokerage data shows that US equities have suffered net selling for a third consecutive week (with four out of the past five weeks in net selling), and the single-ticket deleveraging scale is the largest since the end of July. Industrial stocks have been sold by hedge funds at the fastest pace in over six months, while Consumer Staples and Real Estate became the most net bought sectors last week, continuing the defensive positioning trend.

From a structural perspective, during the previous adjustment period, the market saw a distinct rotation of funds from AI to defensive sectors; however, last week, although Consumer, Utilities, and some Financial stocks rose, this rotation nearly disappeared. Goldman’s trading desk views this structural difference as a key signal—capital is not fleeing the stock market but is focusing on reducing positions in crowded AI trades.

Performance of two AI baskets validates this judgment: the GS TMT Broad AI Basket fell by 4.1% last week, the TMT Mo’ Basket dropped by 7.8%, while the Nasdaq 100 overall declined by only 80 basis points for the week. The selling pressure is highly concentrated on crowded AI positions rather than spreading systemically across the market.

Crowdedness Has Been Released, Limited Drawdown at the Index Level

From a quantitative metrics perspective, Goldman Sachs’ AI vs AI-risk-pair (GSPUARTI) is down 13% year-to-date, while the S&P excluding AI components (SPXXAI) has already outperformed the S&P 500 year-to-date—this means that the damage from this round of correction has been almost entirely concentrated in crowded AI trades, with much less index-level damage than in the AI baskets.

Volatility structure also sends a signal of crowdedness unwinding. The implied volatility spread between the Nasdaq 100 and the S&P 500 has widened to a five-year high at the 98th percentile, while the volatility of volatility (vol of vol) is at the 83rd percentile over three years. The Goldman trading desk notes that last Thursday saw a genuine panic-driven deleveraging event, with the risk appetite indicator’s single-day decline among the largest in history; subsequently, volatility quickly fell before VIX expiration but immediately rebounded, showing a more steady upward trend.

Goldman believes that quantitative research on momentum factors and AI valuations both point to now being a time to buy on dips, although there is a prevailing sense of “dip-buying fatigue” in the market.

Why the Order: Risk Stratification on the Financing Side

Goldman Sachs recommends reopening AI exposure in the order of “Power → Hyperscalers → Inference & Cybersecurity,” rooted in the risk stratification of the financing side.

Last weekend, Musk posted on social platform X, stating that the bond market has effectively closed financing channels for all companies that do not have revenue independent of AI data centers. SpaceX, though successful in completing $40 billion GPU debt financing, enjoys investment-grade ratings and independent income streams from both Starlink and its launch business. Musk bluntly stated that over 200 neocloud companies will face mass extinction and consolidation, with only a handful surviving.

In contrast, SK Hynix has outperformed against the trend thanks to real cash returns. According to Bloomberg, SK Hynix plans to issue a special dividend of about 16 trillion KRW (approximately $12 billion), distributing its roughly 20 trillion KRW gains from holding Kioxia; the company is also considering a shareholder return plan totaling 50 to 60 trillion KRW, including potential buybacks. SK Hynix’s stock price thus rose against the trend, forming a sharp contrast with the general decline in AI-related stocks.

These two sets of facts together outline the logic of risk hierarchy underlying the order: targets with real cash flow and independent revenue support (Power, Hyperscalers, storage leaders) should be prioritized over emerging cloud companies reliant on a single AI data center revenue stream.

Four Verification Criteria, Window Until November 6

Goldman’s trading desk explicitly lists the verification criteria for the judgment of “deleveraging rather than demand weakening,” with a window from October 28, 2026, to November 6:

(1) Positioning Side: Whether net TMT selling in the next two weeks converges to less than half of July's round, and whether the risk appetite indicator stabilizes and rebounds;

(2) Volatility Side: Whether the implied volatility spread between the Nasdaq 100 and the S&P 500 drops from the 98th percentile—if it continues to widen and hits new highs, it is more likely a fundamental re-pricing rather than a positioning flush;

(3) Results Side: Whether hyperscaler capex guidance remains unchanged—if guidance is lowered and neocloud financing remains shut, the premise of “deleveraging rather than demand weakening” will be overturned;

(4) Verification Window: This week, 44% of the S&P 500’s market cap will report earnings, led by Microsoft, Google, Amazon, and Intel, with 10-year US Treasury yields flat at 4.99% and crude oil down 2%. This week’s hyperscaler and storage semiconductor earnings (including SK Hynix, SNDK, STX) and Oracle Analyst Day will provide further verification.

At the same time, Goldman Sachs Global Investment Research points out that current global growth risks are skewed to the downside, Middle East tensions are driving energy prices back to March–April highs, putting negative pressure on consumers and energy-importing economies. This macro backdrop constitutes an external risk factor for the above AI trade call.