Nasdaq plunges 4% overnight, $1.3 trillion evaporates, U.S. stocks face a triple hit

Author: Xiao Bing, Chao Xiang Research

On June 5, U.S. stocks suffered the worst day since the April 2025 tariff crisis.

The Nasdaq Composite plunged 4.18%, closing at 25,709 points, with more than 1,121 points wiped out in a single day. The S&P 500 fell 2.64%, closing at 7,383 points, posting its largest one-day drop since October. The Dow Jones Industrial Average dropped 695 points (-1.35%), even though it had just hit a historical high the day before. The VIX fear index surged 34% in a single day, breaking through the 20 level, while CNN’s Fear & Greed Index crashed from “Greed” to “Fear.”

Just 72 hours earlier, on June 2, the S&P 500 closed above 7,600 points for the first time. All three major indexes were at historical highs. The market had been climbing for 9 straight weeks—everything was celebratory and prosperous—but it all reversed within 48 hours.

To understand this selloff, you need to see how three triggers were ignited at the same time.

First: Broadcom’s earnings tore open the first crack in the AI narrative

The story starts after the close on June 3.

Broadcom (Broadcom) released its Q2 FY2026 earnings report. On the surface, it was a strong set of results: revenue of $22.2 billion, beating Wall Street expectations; adjusted earnings per share of $2.44, also above expectations; and AI chip revenue up 143% year over year to $10.8 billion—far beyond the company’s own forecast.

The issue lies in the outlook for the next quarter.

Broadcom projected AI chip revenue of $16 billion for Q3. The analyst consensus expectation was $17.2 billion. This $1.2 billion gap might, in a normal year, only trigger a mild pullback—but 2026 is not a normal year.

Over the past year, the valuation of the entire semiconductor sector has been built on a core assumption: capital expenditures on AI infrastructure are unlimited, and mega cloud computing companies (Google, Microsoft, Amazon, Meta) will buy computing power regardless of cost.

Broadcom’s earnings report did not deny AI’s high growth—143% year-over-year growth is enough to show demand is strong. It only implied a possibility: the slope of the growth rate may not be as steep as the most optimistic expectations.

More deadly details appeared during the earnings call. CEO Hock Tan admitted that Google may introduce more chip suppliers, meaning Broadcom would no longer be the only favored one. He also pointed out that the rapid growth in the AI chip business is diluting the company’s overall gross margin.

In the context of a stock that has risen 88% over the past year and is already “priced for perfection,” these signals were enough to trigger a stampede.

Broadcom’s stock plunged 12.6% on Thursday. By Friday, panic spread through the entire semiconductor supply chain: Micron Technology fell 13.2%, Marvell dropped 16.7%, Intel fell 11.3%, AMD fell about 11%, ARM fell 12.8%, and Qualcomm fell 11%. The Philadelphia Semiconductor Index crashed 10.26% in a single day, and none of its 30 constituent stocks escaped.

On that day, chip companies listed in the U.S. collectively evaporated about $1.3 trillion in market value.

One detail is crucial: none of these companies released their own bad news. Intel, AMD, Micron—these moves were simply because investors were “extrapolating” Broadcom’s signals: if Broadcom’s AI growth slows, does the entire AI supply chain need to be repriced?

That is the opposite of “Narrative Alpha.” When a story is strong enough, all related assets get pulled in the same direction, regardless of their individual fundamentals.

Second: Too-strong employment data became poison for the market

Friday morning at 8:30, the U.S. Department of Labor released the May non-farm payroll report: 172,000 new jobs added, with the unemployment rate holding at 4.3%.

At first glance, the number even looks relatively mild. But in the face of expectations, it becomes a bomb: the Dow consensus forecast was only 80,000, and the Reuters survey median was 88,000. 172,000 is exactly double Wall Street’s expectation.

Even more unsettling is that the data for the previous two months were revised upward significantly: March was revised from 185,000 to 214,000, and April was revised from 115,000 to 179,000, for a total increase of 93,000 jobs. The average monthly jobs added over the past three months was about 188,000, far above the Fed’s internal “break-even” estimate of 150,000. As long as employment stays above this line, there is no reason for rate cuts.

In normal economic logic, strong employment data is good news—it means the economy is resilient, businesses are expanding, and consumers have money to spend.

But the U.S. economy in June 2026 is not operating under “normal economic logic.”

Since the Iran war broke out at the end of February, the effective blockade of the Strait of Hormuz has pushed up global oil prices. Even on June 5, WTI crude oil was still above $92 per barrel, and Brent crude was above $94. High oil prices have pushed everything up: from transportation costs to food prices, inflation pressure has seeped from the supply side into the economy’s capillaries.

Against this backdrop, the signal sent by an employment report that came in above expectations changed flavor: the economy is too hot—so hot that the Fed may not only refrain from cutting rates, but could even be forced to raise them.

The bond market reacted faster and more honestly than the stock market. The yield on the 10-year U.S. Treasury jumped from 4.47% to 4.54%, reaching the highest level since late May. The CME FedWatch tool’s data was even more striking: just a day earlier, the market had priced the probability of a rate hike by year-end at around 50%, a fifty-fifty split; after the report, that number leapt to 73%, and after the close it broke through 80%. Expectations for rate cuts are effectively zero.

This hit to tech stocks is twofold.

The first layer is valuation compression. Tech stocks—especially high-growth AI-related stocks—depend heavily on discounted future cash flows. When risk-free rates rise, the value of every dollar of future profit today becomes smaller. If interest rates rise by 1 percentage point, the theoretical valuation of a growth stock priced at 40 times earnings could shrink by more than 10%.

The second layer is capital rotation. When bond yields rise above 4.5%, you do not need to take any risk to earn decent returns. For investors who have already made a fortune in AI stocks, selling overvalued tech and moving into Treasuries to lock in yields becomes a simple math problem.

An interesting counterexample is that the Russell 2000 small-cap index rose 1.45% that day. Money flowed out of overvalued large-cap tech stocks and into mid- and small-cap stocks with more reasonable valuations and lower sensitivity to interest rates. This kind of divergence itself shows that the market is not panicking and dumping everything indiscriminately; it is simply repricing the most extreme parts of the AI story.

And beneath the surface of the big number 172,000, the quality of employment is also sending uneasy signals. The figures are supported by hotel service workers (plus 70,000), government employees (local government plus 55,000), and nurses (medical plus 35,000). But the industries that truly reflect whether the economy is hot or cold are shrinking: finance jobs fell by 22,000, and employment in information services has declined 11% since its peak in November 2022.

The wage data also cannot hold up under close scrutiny. In May, average hourly earnings rose 3.4% year over year, which sounds good, but April CPI was already 3.8%. Do the elementary subtraction: real wage growth is negative. Nominal wages are rising, but purchasing power in people’s pockets is shrinking. This is not economic prosperity—it is “the harder you work, the poorer you get.”

Third: The inflation shadow of the Iran war lingers

The third clue is more like an undertow. It may not trigger a crash on its own, but it multiplies the destructive power of the first two triggers.

On February 28, 2026, the U.S. and Israel launched military action against Iran. Iran immediately blocked the Strait of Hormuz, cutting off about 20% of the global oil supply routes. The International Energy Agency described it as “the largest scale supply disruption in the global oil market ever.”

Three months have passed, and the war is still not over. Although the U.S. and Iran reached a framework for a temporary ceasefire last week, new developments in Lebanon have caused the final agreement to stall. Oil prices have retreated from March’s $110 peak, but WTI is still above $90, far higher than pre-war levels.

This continued high oil price puts the Fed in a dilemma. On the one hand, supply-side inflation caused by the war is not a problem that monetary policy can solve; raising rates will not reopen the Strait of Hormuz. On the other hand, if inflation expectations become unanchored due to high oil prices, the Fed will have to respond.

The upcoming June FOMC meeting is approaching. The Fed’s latest Summary of Economic Projections (SEP) still suggests that the next step would be rate cuts, maintaining a dovish bias. But the market is no longer buying it. Federal funds futures are pricing in rate hikes, not cuts. If the Fed is forced to turn hawkish at the June meeting, that would be the formal end of the “soft landing” narrative of the past two years.

On June 5, Citi analysts issued a warning: the bubble level in global equities has reached the highest point since 2008.

When the foundation of the narrative starts to loosen

If you separate these three triggers, you can see that they attack different dimensions of market confidence:

Broadcom’s earnings attack the narrative that “AI growth is limitless.” It is not saying AI is bad; it is only implying that growth might not remain exponential forever. But when the valuation of an entire sector is built on the assumption of “exponential growth,” even the slightest hint of slowdown is enough to trigger a collective repricing of valuations.

The non-farm payroll data challenges the expectation that “the Fed is about to cut rates.” Over the past year, another pillar supporting the stock market has been expectations for liquidity. If the Fed not only refrains from cutting but also raises rates, then both pillars supporting lofty valuations—(1) the growth narrative and (2) liquidity expectations—will wobble at the same time.

The Iran war challenges the consensus that “inflation has been tamed.” When oil stays above $90 and the Strait has not fully reopened to shipping, the specter of inflation continues to hover over the market, making every Fed decision harder.

Together, these create a dangerous feedback loop: AI growth slows, tech stock valuations come under pressure, rate-hike expectations rise, the cost of capital increases, overvalued stocks face even more pressure, and selling spreads.

The plunge in U.S. stocks quickly transmitted worldwide.

South Korea’s KOSPI fell 5.54% on Friday; Samsung Electronics fell 6.4%; and SK Hynix plunged 9.9%. Tokyo’s stock market also dropped sharply. In Europe, ASML fell 3.8%, while Germany’s Infineon plunged by more than 6%.

The crypto market was not spared either. Bitcoin fell about 4% to around $60,000; Coinbase stock fell 7.1%; and Strategy (formerly MicroStrategy) fell 6.9%. When risk assets retreat across the board, crypto’s “digital gold” narrative is once again tested by reality.

Gold futures edged down 0.35% to $4,489 per ounce and failed to play its traditional safe-haven role. In an environment where rate-hike expectations are heating up, the appeal of non-yielding assets is also declining.

Is this the beginning of the AI bubble bursting?

That is the question everyone cares about the most, but the answer is not as simple as it looks.

The bearish case is clear: the Philadelphia Semiconductor Index plunged 10% in a single day. Such a level of selloff typically means the market is fundamentally questioning the sector’s growth assumptions. Marvell fell more than 16% over two days, and Micron fell 17% over two days—these are signs that faith is wavering.

But the bullish case also carries weight. Broadcom’s AI chip revenue grew 143% year over year, and its full-year AI semiconductor revenue guidance still exceeds $56 billion. These are not the kinds of numbers an industry that is bursting out of a bubble should be producing. The issue lies in the slope of growth: AI demand is still real and massive, but can the growth rate keep up with Wall Street’s wildest imaginings?

A more accurate characterization might be this: it is a “valuation repricing,” not a “narrative collapse.” The market is waking up from the frenzy of “AI can make everything rise to the sky,” and is starting to look more calmly at which companies can truly make money from AI—and which are just riding the tailwind.

After the plunge, the S&P 500 still remains close to historical highs. It has pulled back about 5% from this week’s peak, which historically falls within the range of normal technical corrections. The real test is whether this pullback stops at 5% or slides toward 10%—or even deeper.

Over the next two weeks, three key milestones will determine the market’s direction.

First, the June FOMC meeting. Will the Fed continue to hold the position that the next move is rate cuts, or will it formally turn hawkish? If the Fed acknowledges the possibility of rate hikes, the market could face another round of valuation compression.

Second, earnings reports and guidance from more AI companies. Broadcom has opened Pandora’s box. The market needs other AI winners (especially Nvidia) to prove that the AI growth story is not over. The next earnings season will be a critical validation window.

Third, how the Iran situation evolves. If the ceasefire agreement ultimately takes hold, oil prices fall to below $80, inflation pressure eases, and the Fed’s policy room will expand significantly—then the market could rebound quickly. If the war continues to drag on, everything will become more complicated.

The selloff on June 5 was a warning, not a verdict. The underlying logic of the AI revolution has not changed; demand for chips is still real. What has changed is the market’s expectations for growth—and the price investors are willing to pay for those expectations.

When the tide starts to recede, you can see who has been swimming naked.

On June 5, the tide itself was still there—just the pace of the rise was one step slower. But that one beat was already enough to soak the clothes of those fully loaded with positions—like the poor little writer, for example.