Tech stocks experience their worst week since the pandemic; Morgan Stanley: AI trading hits the "wall" moment in the marathon.
After more than two years of a strong bull run, the market once believed that the AI narrative was already more than halfway completed and victory was within reach. However, last week’s crash suggests that the real test may have only just begun.
U.S. tech stocks experienced the most severe sell-off since the onset of the COVID-19 pandemic this week, as the core logic of AI trading is facing a systemic challenge. Morgan Stanley believes that this market movement is not a short-lived style rotation, but rather a "hitting the wall" moment at the 20-mile mark of a marathon—one of the most painful tests, which often emerges just when the finish line appears within reach.
On Thursday, the Nasdaq 100 Index plunged 6.4% in a single day, marking its worst performance since March 2020; the Philadelphia Semiconductor Index tumbled 12.2%, its second-largest one-day drop in history. Morgan Stanley's U.S. Technology, Media & Telecommunications momentum factor has fallen 70% from its June peak, recording the largest annual drawdown on record. Hedge funds’ net leverage plummeted within a week to 42%, ranking in the 8th percentile since 2010, and at the 0th percentile over the past year.
The midterm election results became the immediate catalyst. The Democratic victories in several state elections sparked fears of tighter regulation on data center construction. Texas Governor-elect Gina Hinojosa publicly stated that data center expansion should be paused, and Georgia, Ohio, Michigan, and other states followed suit, triggering a sharp cooling in expectations for AI infrastructure investment. Meanwhile, anticipated "support" from a Fed rate cut is likely absent this time, as strong macro data actually implies limited room for monetary easing.
What Does "Hitting the Wall in a Marathon" Mean?
Before interpreting this crash, it’s necessary to understand a widely circulated metaphor in English financial circles: "Mile 20".
The full marathon is about 26.2 miles (42.2 kilometers). At mile 20, runners often feel good and maintain a steady pace, but the next 6 miles are the real test—glycogen depletion, sore muscles, mental breakdown, and many runners "hit the wall" here, seeing a sudden drop in speed or even forced withdrawal. This segment, known in running culture as "the devil’s mile 20", symbolizes a predicament of suddenly encountering your limits just as success seems within reach: you think you’ve finished most of the journey, but, in fact, the hardest part is just beginning.
This metaphor fits the current state of AI trading perfectly: After more than two years of a powerful bull run, the market once believed the AI narrative was nearly complete, and victory was in sight. But last week's crash suggests the real test may have only just begun.
Thursday: A Historic One-Day Crash
The core shock last week was concentrated on Thursday. The Nasdaq 100 Index fell 6.4% in a single day, its worst performance since March 2020, and the 31st largest one-day drop in its 40-year history; the Nasdaq Volatility Index simultaneously hit its highest level since the tariff shock.
The Philadelphia Semiconductor Index plunged 12.2% on Thursday, its second-largest one-day drop in history, second only to March 2020. This crash pushed the sector’s two-year forward P/E ratio down to single digits for the first time in five years—the chip stocks once seen as the "picks and shovels" of the AI boom now being valued at "steel stock levels".
According to Bloomberg, Thursday saw the highest ever single-day put option volume on the Cboe Options Exchange; the market move that day statistically constituted a 4.9 standard deviation event, with a theoretical probability of about once every 8,000 years.
For the week, the S&P 500 fell 2.7%, and the Nasdaq 100 lost 4.9%. Morgan Stanley's U.S. TMT Momentum pair trading index is down 70% from its June peak, the largest annual drawdown on record; Morgan Stanley’s Global Risk Demand Index fell to -3.4 standard deviations, its lowest in over a year.
Midterm Elections Trigger a Crisis in the AI Infrastructure Narrative
The immediate trigger for this sell-off was the midterm election results. According to Morgan Stanley, Goldman Sachs research suggests Democratic victories at the state level may lead to moratoriums across various states, freezing about 45% of planned U.S. data center capacity.
Texas Governor-elect Gina Hinojosa said Thursday that all data center-related activities in the state should "cease" until new protective measures are in place. Georgia, Ohio, and Michigan have continued to express intentions to follow suit. Morgan Stanley analyst Stephen Byrd pointed out that ERCOT (the Texas grid operator) currently has about 474 GW in interconnection requests, more than five times the peak power demand in Texas, and warns that the U.S. could see data center capacity shortages before 2028.
Byrd believes that under these circumstances, companies with secured power supply (such as Williams, Equinix, Digital Realty, Bloom Energy) will be relatively advantaged, while companies heavily reliant on expanding in Texas (like Cipher Mining, Galaxy Digital, and Solaris) face greater uncertainty.
Hedge Funds Make a Full Retreat, Net Leverage Hits Historic Lows
According to Morgan Stanley prime brokerage data, hedge funds chose to further reduce their exposure during Wednesday’s rebound rather than rebuild positions. Thursday's plunge ranked as the fifth-largest single-day decline in U.S. long-short net leverage since 2010.
For the whole week, hedge funds’ net leverage dropped 8 percentage points to 42%, at the 8th percentile since 2010 and the 0th percentile for the past year; gross leverage fell 4 percentage points to 202%, a 12-month low, with cumulative deleveraging since June equaling about 10.7% of total global assets under management.
AI exposure in U.S. funds' net exposure has dropped from the June peak of 47% to around 32%, and for the first time since 2023, hedge funds have underweighted AI.
This slimming down of positions has somewhat controlled losses: U.S. long-short strategies are down 0.7% month-to-date, better than the S&P 500’s 3.0% drop. But for the approximately 40% of U.S. long-short funds that are flat or in the red for the year, Thursday erased all their 2026 profits, rendering this solace rather hollow.
Strong Earnings Growth, but the Market Refuses to Buy In
From a fundamental perspective, corporate earnings have not collapsed. S&P 500 Q3 earnings growth is tracking at 31.4%, with 84% of companies beating expectations. The issue is that the margin of beats is only 2.7%, below recent averages, and the breadth of upward earnings revisions has fallen from a peak of +26% to +10%.
Specific cases prove out this logic: Palantir achieved its 13th consecutive quarter of accelerating revenue growth, reaching a YoY growth of 96%, but the market reaction was muted; Shopify saw a marked decline even as constant-currency GMV growth hit 30%. Savone explains, "The surprise in free cash flow is now a stronger driver for stock prices than EPS surprises," and market debates have shifted to "AI spending in 2028."
Credit also shows signs of concern. Morgan Stanley analyst Joyce Jiang notes that 16% of software leveraged loans are currently classified as high-risk for AI disruption, with nearly three-quarters rated B- or CCC. Even so, there is a clear divergence between stock and credit markets—the equity market is panicking, while the credit market remains largely unmoved.
Bullish Arguments Remain, but the Rebound May Be Only Short Covering
Morgan Stanley's chief U.S. equity strategist Mike Wilson believes the adjustment that started in June is now in its "final stages", with the equal-weighted S&P 500 outperforming the cap-weighted index by nearly 6 percentage points since September, and sentiment indicators "gradually reaching capitulation levels." Wilson maintains an S&P 500 year-end target of 8,000 points, but isn't rushing to add exposure, opting to wait for further capitulation selling or for the index to rebound to the 7,100–7,300 range before increasing allocation.
Looking at historical data, there is precedent for a bullish case: in the past 15 years, when the Nasdaq 100 experienced a single-day drop over 6% (four occurrences), it posted a gain of more than 25% in the following three months and over 40% in six months.
However, Savone points out the limitations of this analogy: among the four events, three occurred in March 2020—when the Fed cut rates to zero and launched unlimited QE; the other was on April 4, 2025, just five days before Trump announced a pause on tariffs. In every case, a policy "cavalry" was already on the way. At this moment, however, nonfarm payrolls have just massively beaten expectations, Eurozone inflation is still at 4%, and Morgan Stanley's own economists predict a Fed rate hike in December.
"Positioning has been cleaned up, sentiment indicators are signaling fear—these are usually found at market bottoms," Savone writes:
"But as we head into next week’s CPI data—the next key event—this rebound may only be short-covering. In marathon terms, AI trading has just reached mile 20—which is usually where the wall is hit."
Disclaimer: The content of this article solely reflects the author's opinion and does not represent the platform in any capacity. This article is not intended to serve as a reference for making investment decisions.
You may also like
Bitcoin Touches the 200-Day MA — Analyst Calls the 2-Month Correction Over
Assessing if Avalanche’s growing network activity can spur AVAX’s price recovery
Berkshire's stock hits a new high amid tech plunge as return to yen bonds sparks heated "increase positions in Japan" discussion
Berkshire Hathaway has hired banks to prepare for its first yen bond issuance since April, aiming to raise more funds in Japan. According to sources familiar with the matter, Berkshire has mandated Mizuho Securities and Bank of America Securities for a potential benchmark yen bond issuance, which is expected to take place in the near future, though the exact timing will depend on market conditions.
Crypto prices
More
