Position liquidated, but the story continues: Understanding the underlying logic behind this round of AI stock selloff

Original source: Beyond Research (研外之意)

The U.S. added 172,000 nonfarm payroll jobs in May, more than double market expectations. This should have been a reassuring report—the economy isn’t stalling, and employment is still expanding. But the way it landed felt more like a trigger being pulled.

On Friday, June 5, the Nasdaq Composite closed down 4.18%, its worst single-day performance since April 2025; the Philadelphia Semiconductor Index plunged 10.26% in a single day, with the chip sector collectively wiping out over $1 trillion in market capitalization—the most brutal session since the circuit breakers in March 2020. Three days later, on Monday morning, South Korea’s KOSPI index opened with a drop of over 8%, falling below 7,500 points and triggering a circuit breaker that halted trading for 20 minutes. China’s A-shares opened lower and continued declining, with the Shanghai Composite briefly falling below 4,000 points. Japan’s Nikkei and Taiwan’s stock market each fell nearly 4%.

A jobs number that wasn’t even off the charts managed to upend tech stocks from New York to Seoul within 48 hours. The mismatch between the magnitude of the trigger and the severity of the reaction is precisely what makes this sell-off truly worth examining.

The Real Issue Isn’t Those 172,000 Jobs; It’s the Positions They Hit

Blaming this rout entirely on rising rate hike expectations doesn’t hold up.

After the nonfarm payrolls release, the 10-year Treasury yield did jump to 4.5%, and the 2-year yield rose to 4.17%, hitting its highest level since February 2025. Market bets on rate cuts this year were completely wiped out, and futures even began pricing in rate hikes. Higher interest rates typically hit high-valuation growth stocks first—that logic is sound. However, on the same day, the Dow Jones Industrial Average only fell 1.35%, the small-cap Russell 2000 actually turned positive intraday, and nearly half of the S&P 500 components were still rising. If this were truly a systematic repricing of all assets due to interest rates, the sell-off wouldn’t look like this.

The selling pressure was highly concentrated—focused precisely on the stocks that had risen the most and become the most crowded trades over the past two months.

Just how crowded were these stocks? Wall Street had already sounded the alarm before the crash. Goldman Sachs data shows that global hedge fund net leverage surged from below 70% to over 80% in less than two months, approaching the 85th percentile of the past five years. Net allocation to the information technology sector increased by 853 basis points in a single quarter—the largest quarterly jump on record. The semiconductor industry alone accounted for 19% of total global hedge fund exposure, an all-time high, and this proportion had more than doubled since the start of 2026.

Three days before the crash, Citigroup strategist David Chew laid it out plainly: bullish bets on the Nasdaq 100 had been stretched to extreme levels, such that "any negative catalyst would significantly increase the probability of profit-taking and long position liquidation." Citigroup’s famous "Bear Market Checklist" triggered 11.5 out of 18 items on June 5, the highest reading since the 2008 financial crisis.

When a sector is being bet on by global capital using the same direction and the same leverage, it is no longer a diversified investment; it is a trade. And in any trade, someone always has to press the sell button first.

The nonfarm payrolls report was simply the reason to press it.

This crowding is also self-reinforcing. In recent years, trend-following quantitative funds, risk-parity strategies, and a large volume of zero-days-to-expiration (0DTE) options have become powerful forces in the U.S. stock market. Their common trait is following the trend. Once prices break below key levels and volatility spikes, models mechanically reduce positions. The selling itself then pushes prices lower and volatility higher, feeding the next round of selling. Humans decide whether to sell; machines decide how fast. On June 5, the VIX volatility index surged about 34% in a single day, climbing back above 20—a clear signal that this positive feedback loop had been triggered. As for the exact volume of programmatic liquidations that day, there is no public market data; this is a mechanistic judgment, not a precise amount to be calculated.

Watch How Individual Stocks Fell; This Is Deleveraging

Those who rose the most took the hardest hit this time.

May was a banner month for chip stocks. The Philadelphia Semiconductor Index had rallied over 60% year-to-date, rising on 22 of the past 23 trading days. Micron was up as much as 154% this year at its peak, SanDisk had risen nearly fivefold, and AMD alone surged 40% in May to an all-time high. When money floods in one direction, no one cares about valuation.

On June 5, the bill came due. Marvell, which had soared about 25% in a single day earlier after Jensen Huang called it a "potential trillion-dollar company," gave back over 16%, leading the chip stock declines. Micron fell about 13%, while AMD and Intel each dropped around 11%. All were names high on the list of previous gainers.

Nvidia, by contrast, fell only about 6% on that day. Its market cap dipped below $5 trillion, shedding nearly $280 billion—sounds alarming, but relative to the carnage, it held up relatively well. The general stands; it’s the soldiers rushing forward with the heaviest leverage who fall. This pattern of "leaders holding up, peripherals collapsing" is a classic sign of deleveraging, not a fundamental collapse. If AI demand were truly in question, Nvidia would be the first to face scrutiny.

The direct spark was Broadcom. On June 3, it issued a Q3 AI chip sales guidance of $16 billion, below market expectations of $17.2 billion. More critically, it did not raise its full-year AI chip target. In a market accustomed to "upward revisions every quarter," "maintaining the status quo" was interpreted as a negative. A string already stretched to its limit snapped with just that slight pluck.

Why Was South Korea the First to Trigger a Circuit Breaker?

The selling pressure cascaded downward along the path of heaviest holdings, and the first stop was South Korea.

At Monday’s Seoul open, Samsung Electronics and SK Hynix each fell about 10% intraday. These two companies together account for nearly half of the KOSPI’s total market capitalization—illustrating just how concentrated South Korea’s stock market bet on AI is. The KOSPI had been one of the best-performing markets globally this year. It rose the highest, so it naturally has the most room to fall. When global capital needs to cash out and reposition, the most heavily held and liquid South Korean tech stocks become the most convenient "ATMs." Traders at BNY and Lucerne Asset Management used similar language: these are the world’s heaviest-held positions, making them the first to be sold for liquidity.

The amplifier also included local leverage in South Korea. As of June 4, South Korean retail investors’ margin debt was still at an all-time high of 37.74 trillion won. The Korean won fell to near 1,560 per dollar on Monday, and foreign capital outflows were accelerating. Some brokerages suspended margin trading entirely due to exhausted credit limits. South Korea’s finance minister, central bank, and financial regulatory agency issued a joint statement that day, pledging to intervene in the foreign exchange market.

Further down this chain lie Taiwan’s stock market, anchored by TSMC, and China’s A-share computing power chain. They are at varying distances from the epicenter and are traded on different logics.

Which Assets to Watch and How to Monitor Them This Time

Asset reactions are staggered; don’t lump them all together.

The first and most sensitive point is the epicenter itself: HBM (High Bandwidth Memory). HBM is currently the hardest bottleneck in AI computing power. SK Hynix holds over half the global market share, followed by Samsung. These two are the most direct subjects of this trading. Their price action essentially defines the emotional ceiling for the entire chain. Watching their order books is more accurate than tracking any AI concept index.

Moving outward, the next ring includes the heavily crowded U.S. chip stocks in the Philadelphia Semiconductor Index—Micron, Broadcom, Marvell. These are direct reactions, with their decline driven by the pace of position unwinding, not fundamentals. Further out, TSMC is the chokepoint for manufacturing and advanced packaging, making it a node close to the core of the chain.

China’s A-share optical modules, CPO, PCB, and server stocks represent the chain’s diffusion. They have a real performance correlation with global AI capital expenditure, as orders indeed follow NVIDIA’s and Broadcom’s deployment. However, expectations play a larger role in their pricing, so they have greater elasticity. When the epicenter sneezes, these areas catch a cold. As for a batch of high-level thematic stocks and crypto assets that have rallied broadly on the AI concept, they are more followers of sentiment and leverage, lacking direct fundamental support. At these levels, caution is most warranted. In contrast, the pricing of A-share computing power stocks incorporates more local capital flows and policy expectations, so their moves won’t necessarily sync one-to-one with U.S. stocks. But as long as the global AI narrative remains the anchor for pricing, it will be difficult for them to completely escape this unwinding process.

Cross-asset, the decline in gold and Bitcoin shares the same logic: real interest rates are rising, putting pressure on non-yielding assets. On June 5, gold fell over $100 in a single day, breaking below $4,370 and wiping out all its year-to-date gains. Silver weakened in tandem, and Bitcoin briefly fell below $60,000. The U.S. dollar index strengthened throughout. A strong dollar combined with rising real interest rates creates consistent headwinds for non-yielding assets. By Monday, Bitcoin had rebounded back above $63,000, showing risk sentiment recovering ahead of equities. This signal is worth noting.

If net leverage begins to fall, the VIX peaks and heads lower, and HBM and computing power orders and prices remain tight, this is likely a violent but temporary unwinding. The pullback, in fact, creates room. However, if the June 10 CPI report surprises to the upside again, combined with the Fed’s dot plot shift towards rate hikes at the June 16-17 FOMC meeting, the nature would change. A systemic upward shift in the interest rate floor would mean the valuation logic for high-P/E AI assets is being completely rewritten. Deleveraging would be just the opening act. At that point, one would have to concede and pivot toward a more cautious stance. Another counter-signal to watch: if core names like Nvidia and SK Hynix begin to lower guidance or loosen capital expenditure plans, that would be the true moment when fundamentals come into question.

It Was Positions That Crashed, Not the AI Story

In the same week the Philadelphia Semiconductor Index lost a trillion dollars in a single day, TSMC Chairman C.C. Wei stated at a shareholder meeting that global chip supply will struggle to meet AI demand "for several years," and that demand for advanced process technology in 2026 will exceed supply by 25% to 30%. Jensen Huang said in Seoul on June 7 that memory shortages "will last for several years." Nvidia and SK Hynix just announced a multi-year memory collaboration, and SK Hynix’s entire HBM capacity for 2026 is already sold out. SK Group Chairman Chey Tae-won even extended the shortage timeline to 2030. DRAM prices rose about 90% quarter-over-quarter in Q1 of this year.

On the demand side, there are no signs of weakening. It wasn’t the story that crashed; it was the way people were betting on it.

Jensen Huang had just one sentence in response to Friday’s plunge, saying to reporters in Seoul: "We’re just getting started. Whatever happens in the stock market, you should be happy, because now you can buy at a discount." His statement naturally has its bias, but it highlights a core division: if you believe AI is infrastructure as fundamental as the internet, then position washing represents a window to load up. If you doubt that the returns on this wave of computing investment will ultimately materialize, then Friday was the beginning of the crack.

Some people don’t even agree with the term "stampede." In their view, Friday’s market action looked more like a rotation: money didn’t leave the market; it just changed seats, moving out of crowded semiconductors and flowing into banks, industrials, and value stocks. The Russell 2000’s rally against the tide is circumstantial evidence. Under this interpretation, it was a healthy rebalancing disguised in a frightening package. Goldman Sachs strategist Müller-Glissmann also said a bit of consolidation is "not necessarily a bad thing," arguing that speculative leverage and options positions were due for a digestion. This voice reminds people not to mistake a deleveraging event for an apocalypse. But it equally fails to answer one question: until positions are truly cleared and volatility truly peaks, no one can guarantee there won’t be a second wave.

Neither judgment can be conclusively drawn yet. What is certain for now is that the crowded trade global capital piled into over the past two months is being forcibly dismantled. This process is violent, chain-reactionary, and unfinished. The fact that South Korea was still triggering circuit breakers on Monday is itself evidence that the unwinding is far from over.

For the coming week, just watch three things: the June 10 CPI report, the first FOMC meeting chaired by (new Fed Chair?) Walsh on June 16-17 along with the dot plot, and whether South Korean stocks can stop their decline. These will tell you whether this is ultimately a healthy deleveraging or the opening act of a longer correction.

Source: Beyond Research (研外之意)