What caused this week’s market turmoil? Goldman Sachs: It’s not Warsh—it's AI rebalancing.

The recent turbulence in global stock markets this week is not rooted in macro concerns triggered by the change in Fed Chair, but rather in a structural rebalancing within the AI investment boom—the cyclical nature of the memory cycle, the extreme overcrowding of hedge fund positions, and a shifting market perception of winners and losers in the AI supply chain collectively constitute the real driving forces behind this turmoil.

Mark Wilson, a partner in Goldman Sachs' Sales & Trading division, stated clearly in his latest report that although macro signals such as a sharp drop in breakeven inflation rates, rising real interest rates, and a strong U.S. dollar have emerged early in the tenure of new Fed Chair Kevin Warsh, he does not believe these are the root cause of this week's market volatility. He maintains his consistent view: "Equities are still driving macro, not the other way around."

A vivid snapshot of this week's market turmoil: South Korea's Kospi index triggered circuit breakers twice, Microsoft hit a 52-week low, Amazon fell below its 200-day moving average; the "Magnificent Seven" (Mag7) have declined over 5% year-to-date, with Apple and Dell each dropping more than 5% in a single day. Meanwhile, Micron recorded a record gross margin—the divergence between beneficiaries and those under pressure in the AI supply chain is now being presented to the market with unprecedented clarity.

Wilson's core conclusion is that this volatility is not a signal of the end of the AI investment cycle—"We are still in the midst of a historic investment boom"—but rather a deep reassessment by the market of who the net winners and net losers are in this boom.

South Korea as the Epicenter: Leverage Structures Amplify AI Theme Vulnerability

The South Korean market became the most concentrated epicenter of global AI trading volatility this week. The Kospi index, which had risen about 100% year-to-date at its peak, triggered daily price limit circuit breakers twice this week. According to Wilson's statistics, half of all circuit breaker days in the South Korean market this century occurred in 2026, with this week alone accounting for 20% of them.

This extreme volatility has structural roots: about 60% of the Kospi's weight is concentrated in just two stocks, and behind them are hundreds of billions of dollars in leveraged ETF replica products, amplifying any significant moves in either direction. More critically, the latest expansion plans from Samsung and SK Hynix have made the market realize that the current historically high profit margins in the memory industry will eventually face pressure from new supply.

Wilson points out that this is a net positive for the overall supply chain, but for the memory stock beneficiaries, where positions are currently most crowded, it creates direct downward pressure.

Winners and Losers: Structural Repricing of the AI Supply Chain

The differentiation logic in this week's market is particularly clear: companies that are heavily investing capital or facing cost pressures were punished, while supply chain beneficiaries continued to attract demand.

Microsoft not only hit a 52-week low but also fell below its 2021 highs, meaning zero return over the past five years; Amazon dropped below its 200-day moving average. The Mag7 as a whole are down over 5% year-to-date.

On the other end, terminal manufacturers like Apple and Dell have been forced to announce broad price increases due to rising memory costs, with consumers directly bearing the cost corresponding to Micron's record gross margins. Wilson summarizes this logic: "Companies that are investing in AI (with uncertain returns) and those facing margin compression have been punished by the stock market; while those benefiting from capital expenditure and price increases have continued to appreciate in value."

The Memory Cycle: Hidden Risks of Reversal Behind Record Margins

Wilson believes that the evolution of Micron's profit margins could become a classic business school case in the future, but the cyclical risks behind it should not be underestimated.

After the global financial crisis, the memory industry accelerated consolidation, with profit margins systematically rising through the troughs of each down cycle. This evolution is well-documented, strategically supported, and forms a clear investable logic.

However, with Micron's gross margins at an all-time high and SK Hynix's expansion plan announced, Wilson reminds investors: this industry is still fundamentally cyclical. As Samsung and SK Hynix gradually bring new capacity online, current supply bottlenecks will ease—this is a positive signal for the overall supply chain, but for memory stocks that have already benefited significantly and are the most crowded positions, it implies tangible valuation pressure.

Hedge Funds: High Leverage and Extreme Crowding, Risk Accumulated to a Tipping Point

The latest position analysis from Goldman Sachs' Prime Brokerage strategy team reveals another layer of deep risk underlying this week's market volatility.

The data shows multiple extreme signals occurring simultaneously: hedge fund gross leverage has risen to an all-time high; year-to-date, hedge funds have been significantly net sellers of U.S. assets while heavily increasing positions in Asia (mainly Japan and South Korea); within AI-themed holdings, the growth rate of net long positions in memory stocks is twice that of semiconductor or power sector stocks.

From a performance structure perspective, fundamental, systematic, and multi-strategy hedge funds have generally delivered year-to-date returns between 14% and 18%, impressive results, but differentiation is also notable—the excess returns of systematic and multi-strategy funds mainly come from non-AI names; while the returns of fundamental funds are almost entirely dependent on AI themes, with non-AI holdings contributing almost zero alpha. The performance correlation between U.S. and Asian AI positions has now risen to the 99th percentile historically, an extreme level; Europe has largely missed this rally.

Wilson emphasizes that crowded positions alone are not sufficient reason to short, but once fundamentals change, extremely crowded positions will amplify downside risk multiple times—the violent swings in the South Korean market this week are a direct manifestation of this mechanism.

The AI Boom Isn't Over, but the Landscape Is Quietly Reshaping

Wilson explicitly rejects interpreting this week's volatility as an inflection point for the AI investment cycle. He believes that the increased volatility is sending a potentially significant signal to the market: the pattern of net winners and net losers in the AI investment boom is undergoing a deep transformation.

He also points out that political pressure on the AI narrative is rising. As the U.S. midterm elections approach, AI's impact on employment will receive more attention. Relevant data shows that the proportion of U.S. tech industry employment has been on a trend decline for more than five consecutive years, with job losses attributed to AI accelerating, and the industry as a whole has yet to form a compelling positive narrative to counter this.

For investors, the core message from the current turmoil is: one can no longer simply call everything an "AI beneficiary"; a more nuanced differentiation is needed between capital spenders, margin-compressed entities, and supply chain beneficiaries—this will be a key variable for investment decisions in the second half of the year.