Who will end the AI bull market: position or narrative?

The more the market surges, the harder it becomes to find reasons for a decline — but the risks haven’t disappeared, they’ve just hidden deeper.

On May 14, Bloomberg market analyst Jon-Patrick Barnert pointed out that the current rally in U.S. stocks has clearly lifted prices, but the costs and timing of shorting remain difficult to grasp. What’s more challenging is that even the “reasons to short” are becoming blurred now.

The core contradiction of this rally is: positions are already extremely crowded, yet the fundamental narrative — especially AI — continues to support market sentiment. Who will collapse first between the two?

Positions: The market is approaching “full long”

From a purely price trend perspective, signals of a pullback are already quite obvious.

The S&P 500 has been rising for six consecutive weeks, making it one of the longest continuous rallies in over 70 years, with one of the strongest gains in history. Barnert states, “Taking a breather” is perfectly normal for this market.

Goldman Sachs’ Risk Appetite Indicator has risen back to 1, the first time this year. It is extremely rare for this indicator to exceed 1, and historically, it often signals a potential correction. The last time it broke this threshold was in 2021, after which the market entered a bear phase.

Looking at the hottest sector stocks, Barnert describes this as an “overbought” market, with some of the most popular sectors reaching extreme overbought levels. Coupled with mechanical capital inflows — which currently appear to be at or near maximum long positions — the overall picture is: limited upside, with enormous potential pressure for position resetting.

But shorting isn’t easy. Barnert points out that position adjustments can be completed within a single day, making the timing of entering and exiting short trades extremely difficult. And if the market chooses to “gradually decline,” volatility positions will quietly lose effectiveness in a mild environment. A more likely scenario is that overall sentiment remains bullish, and once shorts are forced to cover, it could trigger a new round of short squeezes, with prices rising faster than anyone expects.

Funds flowing into some popular ETFs have already begun to show subtle changes — tending toward “locking in profits” rather than “chasing highs.” But Barnert admits that this trend has persisted for weeks and has not yet had a substantial impact on market movements.

Narrative: Without AI, the market is nothing

If positions are a technical hidden risk, the narrative currently appears more solid.

Barnert notes that there are no clear signals to trigger a fundamental bear market. Corporate earnings remain strong, inflation expectations have slightly risen but are not extreme. The market has digested the impact of high oil prices and Middle East tensions, and the latest US employment data have eased recession fears. As for rate hike expectations, they are no longer a catalyst suppressing stocks.

But one issue cannot be ignored: the concentration of this rally is now highly focused on “concentration itself.”

Barnert points out that whether comparing indices with and without AI, or breaking down the sources of gains since March, the conclusion points in the same direction: without AI, the market’s performance can only be described as “mediocre.” More notably, the semiconductor sector alone contributed nearly 40% of the gains since March.

The market narrative around AI has once again entered a “greed mode,” rather than a rational pursuit of reasonable returns. The concerns that were hotly debated a few months ago — whether AI computing costs could be offset by layoffs, data center energy supply bottlenecks, AI pricing wars eroding profit margins, new competitors disrupting the status quo at lower costs, capital expenditure soaring while stock buybacks stagnate, AI safety risks — now seem to have been collectively forgotten by the market.

Risks of the “DeepSeek Moment” replaying

Nomura Securities strategist Charlie McElligott issued the most direct warning.

He said, “Considering the current market structure and highly overlapping themes, if another full-scale ‘DeepSeek’-style shock catalyst occurs one day, it could directly trigger a Nasdaq circuit breaker (limit-down) type of trading.”

McElligott further pointed out that in such a scenario, semiconductor ETFs could easily see a 15% drop in a single day — because “the hypothetical reflexive mechanical capital flow reversal would create a large-scale overshoot decline.”

In other words, those mechanical funds that kept adding during the rally (such as CTA strategies, risk parity funds, etc.) could, once triggered by a reversal, become amplifiers of the downward spiral.

This AI bull market faces two major risks: one technical (overcrowded positions), and one narrative-based (whether the AI story can continue). The former could trigger at any time, while the latter, if it breaks, could cause a deeper shock. The combination of both creates the most fragile structural vulnerability in the current market.