Goldman Sachs believes that the current risk lies in the fact that when the market's most favored single trade—artificial intelligence—is also the primary driver of rising overall capital costs, a private company's revenue figures can trigger a synchronized move in both stocks and bonds. OpenAI already provided a preview last Thursday, and this Tuesday's major banks' earnings reports along with the CPI data will determine whether this signal is merely noise or a genuine warning.
As the S&P continues to hit new highs, diametrically opposed dynamics are unfolding within the market. Hedge funds are simultaneously dumping tech stocks while pushing their "Mag 7" positions to record levels; confidence indicators have plummeted to extreme lows, yet positioning data across various asset classes keep shattering records. Goldman Sachs's latest weekly report reveals that beneath the surface calm, structural tensions in the market have never been so pronounced.
Last week$S&P 500 Index (.SPX.US)$It rose 1.2% to 7,811 points, hitting another all-time high. However, according to Goldman Sachs, hedge funds' net leverage stands at just 46.9%, placing it in the 2nd percentile over the past year; the long-short ratio has further declined to 1.565, now at the 1st percentile over the past five years. Meanwhile, technology stocks have faced consecutive net selling, with information technology—previously the most favored sector for four straight weeks—leading the decline.
Market structural divergence is also reflected in derivatives pricing. Goldman Sachs' derivatives team points out that,$CBOE Volatility S&P 500 Index (.VIX.US)$Upside convexity is in high demand, while downside protection for the S&P 500 remains "extremely cheap"—in the words of one colleague, "the probability of the VIX rising above 25 is nearly greater than the probability of the S&P 500 falling by 2%." This unusual signal suggests that investors' pricing of tail risk has shifted from concern about a market downturn to guarding against a spike in volatility itself.
Rotation direction: From chips to aircraft and missiles
Fund flows are the clearest narrative thread this week. According to Goldman Sachs' prime book data, after five consecutive weeks of net selling in industrial stocks, hedge funds this week have been net buying the industrial sector at the fastest pace since the start of the year (+1.8 standard deviations), driven almost entirely by long positions. The leading subsectors include aerospace and defense, construction and engineering, airlines, and industrial conglomerates.
There was also a clear divergence within the technology sector. Semiconductors and tech hardware were the subsectors most heavily reduced, while software was the only area to see net buying. Goldman Sachs TMT expert Pete Callahan noted that over the past week, software outperformed semiconductors by roughly 800 basis points, with Goldman Sachs' software–semiconductor hedge portfolio posting a weekly gain of 11.8%.
The trigger for this rotation is clear. Last Thursday, the Financial Times reported that OpenAI's revenue figures were disappointing, sending the Nasdaq sharply lower that day. Goldman Sachs data show that Thursday was the trading day with the most concentrated net selling of tech stocks, as both longs reduced their positions and shorts increased, "pointing to a simultaneous rise in profit-taking and hedging demand."
Mag 7's holdings hit an all-time high, yet confidence has plummeted to a record low.
The paradox is this: hedge funds are slashing their holdings of tech stocks, yet their exposure to the largest tech stocks has never been higher. According to Goldman Sachs' equities division, the Mag 7 position as a share of its total U.S. equity exposure has surged from a year-to-date low of 14% in June to 22%, marking the highest level on record in the prime book.
Morgan Stanley's prime brokerage data corroborates this assessment: information technology currently accounts for 32% of hedge funds' net exposure in North America, placing it at the 98th percentile since 2010. In other words, sector‑level selling does not equate to underweighting—no one has actually reduced their exposure to this sector.
Meanwhile, the performance of fundamental long-short funds has remained under pressure. According to Goldman Sachs' Prime Services weekly report, from October 2 to 8, average fundamental long-short funds fell by 0.70%, while the MSCI Global Index rose by 0.78%; the losses "almost entirely stemmed from alpha‑related losses on both the long and short sides." Year-to-date, the return of fundamental long-short strategies stands at 11.82%, lagging behind the MSCI's 12.42% and the S&P 500's 14.45%. During the same period, systematic strategies gained 21.40%, with quantitative approaches continuing to outperform human judgment.
Earnings Season Opens: High Standards, Low Rewards
Earnings season officially kicks off this week, with the major banks leading the way.
Goldman Sachs' equities division directly highlights the challenge: consensus estimates project S&P 500 earnings per share to grow 27% year over year, the highest since 2021, while the median stock's expected growth rate is only 9% (compared with 14% last quarter). Behind this gap lies the fact that Micron and NVIDIA alone account for roughly one-third of this quarter's earnings growth.
Even more noteworthy is the phenomenon of "no reward for performance exceeding expectations."
Goldman Sachs data show that, in the last quarter, the outperformance of individual stocks whose earnings exceeded expectations relative to the S&P 500 was only 35 basis points, far below the historical average of 95 basis points. Micron posted a 1,003% year-over-year increase in earnings per share, thoroughly beating expectations, yet market reaction was remarkably muted. In light of this, Goldman Sachs asks: Has this "high-standard dilemma" now spread across the entire AI sector?
Bank stocks are facing a particularly unique situation. Goldman Sachs' derivatives team notes that the financial sector has retraced about five times the implied one-day volatility of its options from its peak, yet the market is pricing in an extremely muted reaction on earnings release day. Morgan Stanley has fallen 20%, with option pricing implying a move of only 3.7%; Bank of America has dropped 19%, corresponding to an implied volatility of 3.4%.
Goldman Sachs financial industry expert Christian DeGrasse noted that market sentiment toward banks has "clearly turned cautious," with the key question being whether the slowdown in third-quarter activity is merely a "brief pause" or a "signal of a sustained downturn."
Nick Savone of Morgan Stanley went even further: U.S. bank stocks are now as oversold relative to the S&P 500 as they were in 1993, with a 14-day RSI of around 27, and the BKX forward P/E ratio has fallen below its ten-year median.
The Ripple Effect of the OpenAI Incident: Bond Markets Unexpectedly Benefit
Last Thursday, the market volatility triggered by OpenAI's revenue data not only affected the stock market but also unexpectedly boosted long-term bond prices. Goldman Sachs' futures team noted that long-term bonds "may finally have bottomed out this week," with the rebound sparked by headlines related to OpenAI.
The logic is this: once expectations for a pickup in AI‑related capital spending begin to wane, the demand for longer‑dated U.S. Treasuries—previously squeezed by robust AI‑funding needs—could start to rebound. Earlier, asset managers' short positions in long‑term bonds have surged since mid‑September; if Thursday's price action prompts these shorts to cover, the bond market could gain additional support early next week. However, whether that rally can endure will still hinge on Tuesday's CPI data.
On the gold front, prices rose 1.3% this week, "remaining above $4,100 per ounce." Goldman Sachs' futures team noted that systematic strategy positions have fallen to the 0th percentile over the past year, and CFTC managed funds' net longs have also returned to normal levels, concluding that "the attractiveness of increasing long positions in gold is rising."
Electricity: The "Most Undervalued AI Bottleneck" in Goldman Sachs' Eyes
Goldman Sachs' thematic basket team has dubbed its "Power Up America" basket (ticker GSENEPOW) the "most undervalued AI bottleneck," basing this assessment on three dimensions: positions are at near two-year lows, valuations have normalized, and fundamentals remain solid. This week, the basket gained 4.9%, with the independent power producer sub‑sector surging 12.7%.
The team believes that market concerns about the midterm elections and the politicization of power policy are "fading," and expects that future grid investments and behind-the-meter electricity solutions will boost U.S. power supply. Goldman Sachs forecasts that by 2030, behind-the-meter electricity will meet 25% of data center power demand nationwide. Recent key catalysts include FERC's ruling on the PJM high-load framework on Monday, October 12, as well as the release of PJM-related documents and the results of the ERCOT data center grid-connection audit.