The market value of AI chips plummeted by one trillion yuan in a single day: the non-farm payrolls data is just an excuse, while market overcrowding is the real truth.
The drop on June 5th was truly alarming. The Nasdaq Composite Index tumbled 4% in a single day, and the SOXX index, which tracks chip stocks, plunged by over 10% in a day, marking its worst performance since the market crash in March 2020 at the onset of the pandemic. Those holding tech stocks couldn't feel good after checking their accounts at the market close.
When the market opened on Monday (June 8th), instead of continuing the downward trend from Friday, the market rebounded led by the hardest-hit chip stocks. The SOXX semiconductor index, which had dropped 10% on Friday, rallied 5.9% on Monday. Broadcom and NVIDIA also regained 2.8% and 1.7% respectively. The market's panic subsided accordingly. The fact that the hardest-hit stocks rebounded the most already indicates that the drop on June 5th may not necessarily be the start of a market crash.
To be honest, the VIX fear index jumped from around 15 to 21 on June 5th, which is indeed not low. However, when viewed over the past year, it's not considered extreme. The last time it really spiked was in early March this year during the Iran conflict. The United States and Israel launched military actions against Iran, the Strait of Hormuz was temporarily blocked, oil prices soared, and the VIX reached 35 intraday. The entire market tumbled for over a week, with all stocks being sold off indiscriminately. That was a real full-blown panic.
Previously, the S&P 500 had been on a nine-week winning streak, and market volatility had been at a low level for a long time. Market expectations had become overly concentrated. This drop disrupted that concentration. Although the single-day decline was the deepest for the Nasdaq in the past year, in terms of intensity, it was more like a sharp pullback after a long period of one-sided gains. The drop was indeed rapid, but it was not out of control.
This drop happened right at the Friday market close. Thanks to the two-day weekend break, investors could take the time to figure out the most important thing: why it dropped. Only by understanding the reason can one decide whether to hold steady or take action. Otherwise, one will simply follow the market sentiment, chasing gains and cutting losses. Chasing and selling are the easiest things to do in a panic, and also the most likely to be wrong. By the time one realizes it, good stocks are often sold at the bottom.
From the perspective of mainstream financial media, the reason for this sharp drop is relatively consistent: the non-farm payrolls data. On June 5th, the U.S. employment data was extremely strong. The market worried that the Federal Reserve would resume raising interest rates, causing tech stocks to lead the broader market down. This logic makes sense. Once interest rate expectations rise, tech stocks, which have the longest duration and are most sensitive to liquidity, are the first to be hit. Moreover, on that day alone, the probability of an interest rate hike by the end of the year jumped from around 50% to over 70%.
However, the interest rate factor can explain the general weakness of tech stocks but fails to explain the truly abnormal aspect of the market on June 5th. The hardest-hit were not just high-valued tech stocks in general, but a specific group of AI chips. The semiconductor sector tumbled by over 10% in a day, two to three times worse than the broader tech market. Moreover, the weakness in the chip sector started two days before the non-farm payrolls data. After the market closed on June 3rd, Broadcom's guidance for AI chips fell short of expectations, and chip stocks crashed on June 4th.
So, the non-farm payrolls data was at most a spark. The real cause of the sell-off was the crowded positions in the AI industry chain. There's also an interesting signal: on June 5th, a batch of funds withdrew from the tech sector and moved into defensive sectors in a rather calm manner. In a real market crash, no one would move their investments so calmly. Therefore, this drop may point to something deeper, and it's worth analyzing the market step by step.
01
It's Not a Crash, It's a Pause
Analyzing the market on June 5th, we can find that the decline was far from even.
Let's start with a direct comparison. Among the 500 components of the S&P 500, when weighted equally, the index dropped only 1.4% on that day. However, when weighted by market capitalization, the decline doubled to 2.6%, almost twice as much. Such a large difference between the two calculation methods indicates that the majority of the decline was concentrated in a few of the largest stocks, that is, the high-weighted large-cap tech stocks.
Looking at other indices: the blue-chip Dow Jones Industrial Average dropped only 1.4%, the high-volatility small-cap Russell 2000 fell 3.5%, and the tech-heavy Nasdaq Composite Index tumbled 4.2%. The more growth-oriented and crowded the sector, the deeper the decline. The more evenly distributed and unremarkable areas were relatively unaffected. A real indiscriminate full-blown panic should have affected all stocks equally, without such a significant difference. This clear gradient shows that the selling pressure was precisely targeted at a small group of large and crowded stocks.
From a timeline perspective, the non-farm payrolls data was released at 8:30 a.m., an hour before the market opened. However, from 8:30 to 9:00 a.m., large-cap ETFs like SPY barely moved, with fluctuations of only 0.1% to 0.2%. The real sharp drop didn't occur until after the market opened at 9:30 a.m. This isn't conclusive evidence, as pre-market trading is usually thin. But it at least shows that the market didn't rush to sell off as soon as the data was released.
What's really worth pondering is not which stocks were sold, but what kind of stocks they were. Even after the over 10% drop on June 5th, the SOXX semiconductor index has still gained nearly 80% this year. Marvell, the hardest-hit stock, tumbled 16% in a day, Micron dropped 13%, and the leading NVIDIA also fell by over 6%, with its market capitalization dropping below the 5 trillion mark. The stocks that were sold the most aggressively were precisely those that had risen the most and had the thickest paper profits in the past six months.
Moreover, this selling pressure isn't just a U.S. phenomenon. In fact, the sequence was reversed in terms of time: Asia led the decline, followed by Europe, and finally the United States. Samsung, SK Hynix, and ASML were all affected. The crucial factor is this time difference: by the time the U.S. non-farm payrolls data was released at 8:30 a.m., the Korean market had already closed, and the decline was a foregone conclusion. When the decline spread in Asia, the non-farm payrolls data hadn't even been released. It's hard to say that a U.S. data that Asian chip stocks didn't even have a chance to react to was the root cause of this global decline. The real common factor is the label they all share: AI.
The more crowded a trade is and the thicker the paper profits are, the more likely investors are to have the idea of "taking profits" at a certain point, especially when the index has just hit a new high and almost everyone is making money. It doesn't need a huge negative news to trigger a sell-off. As long as some investors start to take profits, other funds in the same trade will panic and follow suit, leading to a stampede.
The money sold off has to go somewhere, and the capital flow on June 5th is exactly what we should understand. First, look at the defensive sectors that bucked the trend and rose: the consumer staples sector rose 1.7%, the strongest on that day. Procter & Gamble, Coca-Cola, and other consumer goods stocks rose 3% to 5% against the trend, and the healthcare sector also closed in the green. Zooming out, more than half of the stocks in the S&P 500 rose on that day. The index closed down only because it was dragged down by the large market capitalization of a few giant tech stocks at the top.
But the real key lies elsewhere: the money that withdrew from AI and chip stocks didn't go to chase a new leader. Small-cap and cyclical stocks, which usually charge ahead with tech stocks, didn't take over either. The Russell 2000 still fell 3.5%. If the funds were really determined to abandon AI and switch to other sectors, there should have been a new leading direction in the market. However, the market only reduced the positions in the most overheated and crowded sector.
Therefore, instead of saying that June 5th marked a peak or a rotation, it's more like an AI bull market that had risen too rapidly needed to take a breather. At least from an earnings perspective, AI remains the strongest fundamental factor in this market. The expected earnings per share growth rate of the "Magnificent Seven" next year is still more than twice that of the other hundreds of companies. However, after this drop, market differentiation and concentration are likely to become more pronounced, and the market's consensus on AI may even reach an unprecedented level.
02
Can It Sustain Such Intense Growth?
Over the past weekend, there were numerous AI-related commentaries. Jensen Huang interacted with Samsung and SK Hynix in South Korea and signed deals, reigniting the enthusiasm of many. However, so much money being bet on AI ultimately boils down to a bet on whether this business can keep growing at such a rapid pace. To sustain this growth, how much money needs to be invested, who will provide it, and whether they can afford it were hardly considered during the period of the most rapid growth. This drop didn't shake the bet on AI, but for the first time, it brought this overlooked question to the forefront.
From a macro perspective, a significant part of the U.S. economic growth in the past two years can be attributed to AI investment. According to official U.S. data, in the first quarter of 2026, the annualized economic growth rate was 2%. Among them, AI-related investments, namely computer equipment and software, contributed approximately 1.1 percentage points, almost on par with the contribution of total consumer spending, even though the scale of consumer spending is about twenty times that of AI investment. In other words, without this wave of AI investment, the growth rate in that quarter would have been cut in half from 2% to around 1%.
On the stock market side, Goldman Sachs estimates that about half of the earnings growth of the S&P 500 in 2026 will come from AI-related investments. The contribution from the remaining hundreds of companies is very limited, leaving little room for error. Without AI, the U.S. economy won't collapse immediately, but it will be impossible to maintain the current growth rate.
The situation is similar globally. It goes without saying how much the stock markets in South Korea and Japan have benefited from this wave of AI semiconductor boom. Looking at the A-share market in China, shifting our focus from stock price movements to corporate earnings, the differentiation is also striking. In the first-quarter earnings reports of 2026, the net profit of all A-share companies increased by only 6.6% year-on-year. However, in industries strongly related to AI and computing power, such as semiconductors and software, the net profit growth rate reached one hundred to two hundred percentage points. There were even a number of companies in the optical module and storage sectors whose revenues and profits doubled. Currently, the pattern of the global stock market is clearly divided: companies related to carbon-based life are facing deflation, while those related to silicon-based life are booming.
It's precisely at this time that several events have come together, making this business seem a bit tricky for the first time. These are more forward-looking risk signals rather than the cause of the drop on June 5th. First, Broadcom released a record-breaking earnings report after the market closed on June 3rd, with its AI chip revenue surging 143% year-on-year. However, the CEO didn't raise the full-year target as the market expected, but only maintained the $56 billion forecast, and the guidance for the next quarter was slightly lower than expected. Due to this small gap, the stock price plunged 12.6% the next day, its worst performance in over a year.
Then there's Google. It's reported that even a cash-rich company like Google plans to raise about $85 billion in the market to invest in AI data centers. To be honest, this amount of money isn't a significant negative factor for a company of Google's size. However, interest rates have remained high, with the yield on the 10-year U.S. Treasury note rising back above 4.5%, increasing the cost of borrowing for construction.
There's also a quieter but more worthy signal to note: this trend shows no sign of stopping and is even accelerating. Several large companies plan to invest over 30% more in 2026 than last year. The problem is that the economy and the stock market are becoming increasingly reliant on this single force. However, no investment can keep growing indefinitely. Sooner or later, the growth will slow down. No one can predict which quarter it will happen. But once that day comes, the force that has been driving half of the growth will turn into a drag.
However, these are still just distant questions for now, not immediate problems. Broadcom's AI revenue is still growing, and Google is still increasing its investment. This business won't stop in the short term. Therefore, with the current strong consensus, the real thing to do isn't to guess when the bubble will burst or to follow the non-farm payrolls data to chase gains and cut losses.
What was sold off on June 5th was the overheated positions, not the fundamentals of good companies. Instead of panicking and selling good stocks at rock-bottom prices, it's better to focus on the upstream spending in the AI industry chain and manage the profitable positions well.
03
Conclusion
Due to well-known reasons, it has become more difficult for investors on the Chinese mainland to directly trade U.S. stocks than before. However, this doesn't mean that the U.S. stock market and other markets highly correlated with it aren't worth paying attention to and studying.
Since the beginning of this year, the A-share and H-share markets in China have become increasingly similar to the U.S. stock market. The companies with strong earnings and rapid growth in the A-share and H-share markets follow a similar pattern to those in the U.S. stock market. Moreover, the semiconductor industry is a global industry chain. Any movements of NVIDIA, Broadcom, and other U.S. semiconductor companies will directly affect A-share companies related to AI and chips, causing fluctuations in their earnings and stock prices.
What happens in the U.S. stock market isn't just a matter of the U.S. market itself. It also affects the direction of the global capital market, including the A-share market.
Since we can't avoid it, we need to understand it better. What puzzles me the most about the current market isn't how much it has dropped, but the abnormal calmness after the drop. A market that can remain so calm after a 4% single-day drop and where investors rush to buy the dip immediately after the drop - is this calmness a sign of real confidence or just a form of numbness after a long period of growth?
Could this calmness itself be a risk when no one is really afraid even after a drop?
This article is from the WeChat official account "Lying Flat Index", author: Sister Tang. Republished with permission from 36Kr.