Five weeks ago, Goldman Sachs derivatives trader Shawn Tuteja strongly recommended buying call options on AI stocks, citing "no one is positioned and options are cheap." That trade subsequently delivered substantial profits.

Now he has personally reversed his own call — the AI sector has transformed from an ignored corner into the most crowded trade in history, and the title of his latest report is telling: "Not So Clear Anymore."

The S&P 500 closed at a record high of 7,818.93 on Tuesday, while the Nasdaq 100 simultaneously hit a record 31,224. Yet Tuteja's latest research shows that Mag 7 net exposure has risen to the highest level in Goldman's database (going back to early 2022), accounting for roughly 22% of total U.S. equity exposure. He warns that the original rationale for buying AI — "low valuation, low positioning" — has essentially disappeared, leaving behind a long bet that requires fundamentals to keep delivering.

At the market level, the implications of this judgment are direct and clear: Goldman is advising clients to shift their chips away from the already-crowded AI leaders toward lagging sectors represented by SPXXAI (the S&P 500 excluding AI stocks), while viewing energy infrastructure-related AI assets as one of the few exceptions that still have the old logic supporting them.

Bond Market Pressure Is Driving Money Into Seven Stocks

Tuteja analyzed the macro backdrop behind this renewed crowding in the AI trade. Since late June, Goldman's financial conditions index has tightened by about 40 basis points, almost entirely from rising long-end rates — with credit spreads, the dollar, and equities contributing very little. Meanwhile, the SOFR December 2026/December 2027 spread has surged nearly 50 basis points since late August, with the market significantly raising the odds of further rate hikes before the end of 2027. The 10-year Treasury yield currently stands at 5.28%, and the 30-year at 5.66%.

Faced with a comprehensive tightening of the rate environment, fast money has made a highly consistent choice: buy a handful of tech stocks that appear immune to high rates, and sell everything tied to rate sensitivity. "Clients did indeed re-buy tech stocks due to this macro uncertainty," Tuteja wrote.

Data confirms this judgment. In September, U.S. tech stocks experienced their largest monthly net buying since February 2025, the only sector in Goldman's prime brokerage book to record net buying. At the same time, retail money has made a strong return — total assets in U.S. long leveraged semiconductor ETFs rebounded from a low of about $72 billion on July 29 to roughly $115 billion, a recovery of about 60% in ten weeks.

Record Concentration Means AI Is No Longer a Low-Valuation Trade

The data Tuteja cites clearly reveals the core of current risk. Mag 7 net exposure accounts for about 22% of total U.S. equity exposure, the highest in Goldman's data history; semiconductor and semiconductor equipment exposure is roughly 12%, double the approximately 6% at the start of the year; Goldman's prime brokerage book net leverage ratio stands at 46.4%, down 8.5 percentage points from the start of the year, at the 0th percentile of the past year's data, with the long-short ratio at the 1st percentile of the past five years.

Market breadth data is equally alarming. In September, the S&P 500 cap-weighted index outperformed the equal-weighted index by about 4.5 percentage points, one of the largest monthly gaps in over 25 years; fewer than 25% of constituents outperformed the index; the median stock is down 17% from its all-time high; and the Nasdaq 100 has beaten the Russell 2000 in 14 of the past 16 trading sessions. Goldman's Tony Pasquariello described large-cap tech as "continuing to play the role of both spear and shield."

BTIG's Jonathan Krinsky noted that the Nasdaq 100 has outperformed the Russell 2000 for 14 consecutive days, "previously seen only in early 2000," and pointed out that the Philadelphia Semiconductor Index's analogy to the 2000 trajectory "continues in astonishing fashion," currently positioned near the top of that rally.

Two Paths, Neither of Which Is "Keep Chasing Nvidia"

Tuteja's judgment on the path forward is the core of this report. He describes two scenarios, and both lead to a highly consistent conclusion: if macro improves, the lagging remaining 490 stocks will outperform the AI leaders; if macro deteriorates, AI stocks will ultimately fall along with the broader market. Under either scenario, chasing AI stocks that are already heavily held is not the right move.

His preferred expression is to go long the lagging sectors with limited risk: buying SPXXAI November-expiry 102% out-of-the-money call options for about 1.25%, "which can serve as a convex right-tail portfolio hedge if the macro environment turns warmer." Notably, from Goldman's perspective, call options on the average U.S. stock are now characterized as "tail hedging tools."

Morgan Stanley's Mike Wilson said over the weekend that current poor market breadth "must be resolved somehow"; Deutsche Bank's Henry Allen put it more bluntly. On the other side, JPMorgan's trading desk last week shifted back to a "tactically bullish" stance, though the foundation for that confidence remains questionable against a backdrop of high rates.

Hedge Funds Have Little Left to Sell

The report is not entirely bearish, because the other side of the positioning picture provides downside support. Although hedge funds are heavily positioned in tech, overall leverage is at historical lows, with the net leverage ratio at the 0th percentile of the past year's data, meaning that even if the market falls, the room for large-scale deleveraging is limited.

Tuteja estimates the S&P 500's implicit "buying zone" at around 7,400, about 5.4% below Tuesday's close. He lists three reasons supporting a "relatively contained" index decline, including fund leverage already being at low levels, seasonal factors, and the reopening of the stock buyback window.

However, critics point out that this floor logic is clearly circular: the so-called "support" depends on those hedge funds that just made record additions to Mag 7 being willing to keep buying more stocks after a 5% decline.

The Only AI Trade Still Logically Supported: Buy Power

Within the AI sector, Tuteja remains bullish on one specific area — AI power infrastructure. The sector's valuation has fallen from about 26 times forward earnings at the start of the year to about 21 times, positioning is relatively light, and there are substantive catalysts. Goldman's AI power basket (GSENEPOW) rose about 5% on the day the report was published, with Constellation Energy (CEG) surging 12% in a single day after Google announced it would connect to its nuclear plant for power supply.

Tuteja's year-end trade idea is: sell GSENEPOW December 31-expiry 85% out-of-the-money put options, using the premium received to fund the purchase of a 110%/135% call spread, with a net cost of about 1.75%.

It should be noted that the November 3 midterm elections are an important variable. Goldman options data shows the market is pricing an additional election-related volatility of 1.69% for the AI power sector, exceeding the S&P 500's own 0.31% and higher than the corresponding figure for any single party's policy basket. As Tuteja's colleague warned on Sunday, the midterm elections are "unlikely to be a positive event for the market."

The Bond Market Is the Real Arbiter

Tuteja believes the core risk in the current situation still lies with CPI data.

An October rate hike was priced at a 64% probability before last week's nonfarm payrolls data, and one "mild" data point has not fundamentally changed the inflation pressure argument, especially with Brent crude near $100 per barrel and the 30-year Treasury yield at 5.66%. The last time the gap between the S&P 500 and the "S&P 500 excluding AI" was this wide was in June, and holders of leveraged semiconductor ETFs still remember what happened in July.