The AI investment narrative is confronting an internal contradiction in its own logic. The market's judgment on the capex path of hyperscale cloud providers has shifted from the earlier "win no matter what on the upside" to the current dilemma of "choosing between two evils" — accelerating would strain free cash flow, while decelerating would cast doubt on the demand outlook, with both paths pointing to a selloff in AI-related assets.
According to a cross-asset strategy report by Nomura's Charlie McElligott dated October 23, the past week saw a succession of negative sentiment catalysts within the AI ecosystem, which, combined with the long positions accumulated during September's "agent hype" rally, triggered a large-scale forced liquidation across the market.
In just the most violent single-day selloff, 47 of the 124 US equity long-short pairs monitored by Nomura recorded single-day moves exceeding 3 standard deviations, with 6 hitting extreme deviations of more than 5 standard deviations.
The turmoil not only hit core AI concept stocks but also spread through cascading liquidations of crowded positions into US Treasuries, foreign exchange, and commodity markets, triggering a cross-asset systemic deleveraging.
Next week brings a dense window of earnings disclosures from hyperscale players, including Google, Amazon, and Microsoft, alongside the Federal Reserve's rate-setting meeting, meaning market direction will face a critical test in the near term.
The "Prisoner's Dilemma" of Capex
The core narrative of the AI trade has abruptly shifted from the earlier one-sided logic of "all positive outcomes are bullish" to a two-way dilemma of "everything looks bearish no matter how you slice it."
McElligott described this predicament in his report as "either choice is poison":
First, if hyperscale capex continues to accelerate. As the trend revealed in Meta's latest earnings shows, the market will focus intensely on the ongoing depletion of free cash flow and further question the overall sustainability of financing.
This path carries nonlinear risk, meaning debt must be issued at larger scale and wider spreads, and the AI trade comes under pressure from the "unsustainable" narrative.
Second, if hyperscale capex sees a substantial deceleration, it would mean the "optimal scenario" already priced in by the market — covering core theses such as AI demand, technology adoption, profitability, and end-user productivity gains — is now breaking down, and the AI trade would face an even more ferocious shock.
The formation of this impasse is closely tied to the rapid rebuilding of long positions during September's "agent concept" rally. At that time, large amounts of capital poured into core AI names, accompanied by sizable speculative call option positioning, laying the groundwork for the subsequent stampede-like exit.
Stop-Loss Triggers Set Off Cross-Market Deleveraging
Crowded positioning is the core reason this round of volatility has been so unusually severe.
According to Nomura trading desk data, in the most violent single-day selloff, a total of 124 US equity long-short pairs triggered stop-losses, of which 47 pairs moved beyond plus or minus 3 standard deviations, 26 pairs exceeded plus or minus 4 standard deviations, and 6 pairs exceeded plus or minus 5 standard deviations.
McElligott believes the root cause of this stop-loss wave lies in the extreme concentration of "high Sharpe ratio trades" throughout 2026, with carry strategies and trend-following strategies resonating over the long term, creating a rare broad-spectrum winner effect.
Yet the current juncture coincides with the end of the fourth quarter, when fund managers face year-end performance pressure. In an already stressed drawdown environment, the drive to actively hunt for a rebound is far weaker than the defensive instinct to "preserve capital and cross the year safely" — a "de-risk, close positions" mode has come to dominate market behavior.
At the same time, the reset of the AI narrative is accelerating position adjustments in systematic trend strategies, particularly concentrated in the previously extremely crowded short positions in US Treasuries and short-term rates.
According to Nomura QIS data, over the past two weeks, CTA trend strategies have closed out roughly $68 billion of Treasury futures shorts and $38 billion of short-term rate futures shorts, but signals remain in a net short state, meaning that if the bond rally continues, the potential scale of forced covering remains considerable.
McElligott believes that with the short-term window of covering largely complete, the squeeze momentum on bond shorts is pausing in phases. Both the 2-week and 1-month cycle models have flipped to "long" signals, but for the 3-month cycle model to flip long from current levels, bond prices would still need to rise substantially further.
Macro Resonance: The Correlation Switch Between Bonds and Equities
This reset of the AI narrative has simultaneously triggered structural changes in correlations within both the bond and equity markets, a signal worth watching at the market-mechanism level.
On the bond side, the protective property of "negative stock-bond correlation" has reappeared. McElligott noted that a large amount of capital completed an asset allocation switch on the previous trading day through selling equity futures and buying Treasury futures, with bonds once again playing the role of a risk hedge.
This feature, which persisted through most of the year before temporarily failing, has now returned as the "growth scare" scenario gains credibility.
The equity market has seen the opposite switch. For most of 2026, the Nasdaq and the semiconductor sector displayed a rare feature of "higher volatility on up days than on down days"; now, this relationship has reversed back to a "negative spot-volatility correlation," meaning volatility is greater when prices fall.
This switch is most pronounced in the Nasdaq and SMH (semiconductor ETF) space, directly reflecting the market's repricing of the sustainability of AI capex.
It is worth noting that VIX volatility at the overall S&P 500 index level has been relatively mild. McElligott attributes this to a "reverse dispersion" effect — tech longs have been badly hit, while rate-sensitive sectors and other short targets have rebounded sharply, with the two offsetting each other so that overall correlation at the index level has not surged explosively.
Large Options Signals: Contrarian Money Still Positioning for an AI Rebound
Despite the overall pressured sentiment, several contrarian positioning signals worth noting have still appeared in the options market, mainly in call options on certain core AI names.
According to McElligott's observations, the following notable call-buying activity occurred over the past week:
CoreWeave paid a total premium of about $69 million, involving notional exposure of roughly $170 million;
Nebius Group paid a total premium of about $25.7 million, involving notional exposure of roughly $80 million;
Micron paid a total premium of about $20 million, involving notional exposure of roughly $85 million.
The above trades involved out-of-the-money call options expiring between March and April 2025, showing that some capital is still maintaining directional bets on a medium-term rebound in the AI sector amid the current turmoil.
As for potential conditions for stabilization, McElligott believes the key lies in whether volatility risk premium (VRP) sellers are willing to follow historical patterns and "short volatility when it rises," especially the stabilizing role of 0DTE gamma-selling strategies in buying dips and selling rallies, which will be an important indicator for judging whether volatility can converge again.
Comments