Hey there, bargain hunter.

Something broke in the chip market today. Not the business. The sentiment.

This morning, Samsung Electronics fell 13.4%, its worst single session in almost two decades. SK Hynix dropped 14.7%. South Korea’s KOSPI index lost 10.8% and triggered circuit breakers. The selling followed through to Europe, then to U.S. premarket, where Micron was down nearly 10%, AMD off more than 3%, and Nvidia sliding again.

That is not a sector dip. That is a global rout.

The question worth asking right now is the one most people are too rattled to ask: did anything actually change in the underlying businesses?

Because if the answer is no, then what we are witnessing is not a fundamental collapse. It is fear recycling itself across time zones.

What Happened

The current wave of selling has multiple triggers, none of them individually decisive, all of them landing at once.

• Broadcom’s June earnings showed record quarterly revenue of $22.2 billion, up 48% year over year, with AI chip revenue surging 143% to $10.8 billion. The market punished it anyway because forward guidance disappointed, and the fiscal 2027 target of $100 billion was left unchanged rather than raised.
• A strong May jobs report added macro pressure, with 172,000 jobs added, more than double analyst expectations, reigniting fears of a hawkish Federal Reserve pivot.
• CXMT, a Chinese memory chipmaker, surged 465% on its Shanghai IPO debut today, sharpening investor fears about long-term competitive pressure from Chinese producers.
• Broker reports began circulating about memory prices peaking around 2027, rattling holders of high-flying HBM names.
• Positioning was simply overextended. SOXX had run 102% off its February low before the selling started.

The result: since the late-June peak, the sector has lost roughly $2.1 trillion in market value, with a median decline across major chip names of around 21%. SOXX has fallen approximately 13% to 15% from that high in a matter of weeks, its sharpest drawdown since April 2025.

The long-short ratio on Nasdaq 100 futures dropped 63% over the past year and recently hit a 17-year low. When positioning is that stretched, almost any catalyst becomes a selling catalyst. The fundamentals did not have to change. The math of crowded positioning did the work instead.

The Case That Interests Me: Micron

Here is where it gets interesting.

Micron reported fiscal Q3 2026 results that were, by any honest measure, extraordinary. Revenue rose 346% year over year to $41 billion. Non-GAAP gross margin expanded to 85%, compared to 39% a year earlier. Operating income hit $33.3 billion. Diluted EPS came in at $24.67. Adjusted free cash flow for the quarter was $18.3 billion.

Read that again. $18.3 billion in free cash flow in a single quarter.

The company’s trailing twelve-month free cash flow now stands at approximately $26 billion. The balance sheet carries $26 billion in cash against $6.4 billion in debt, a net cash position of roughly $23.75 billion. Return on invested capital is running at 67.6%.

And yet, as of this morning, MU was trading near $812, down from an all-time closing high of $1,213 on June 25. That is a 33% decline in about a month, on a business that just reported the best quarter in its history and guided for further growth in Q4.

The forward P/E on Micron right now is approximately 6.3x. The PEG ratio is 0.04. The EV/EBITDA sits near 12.8x. For context, the broader SOXX ETF trades at a weighted forward P/E of around 22.8x.

Wall Street’s consensus price target is approximately $1,507. The average analyst rating is Strong Buy. Even after everything that happened today, nobody who covers the company professionally has walked away from the fundamental thesis.

Is This Cheap, or Is It a Trap?

This is the exercise that actually matters.

The bear case is real and should not be dismissed. Micron is a cyclical business with a high fixed cost base. As Morningstar’s analyst notes, that structure leaves it exposed to severe margin compression when memory markets turn. As recently as fiscal year 2023, revenue fell 49% in a single year. The current boom in high-bandwidth memory for AI data centers is real, but broker reports are already projecting that memory prices could peak around 2027. China’s CXMT just IPO’d to enormous fanfare, signaling that the competitive landscape is widening. And Micron’s share of the HBM market remains well below that of SK Hynix.

Those are legitimate risks. Anyone pretending otherwise is selling something.

But here is the part people are skipping.

Micron has signed 16 long-term strategic customer agreements totaling more than $22 billion in committed HBM revenue, with current supply reportedly sold out through 2026. The company’s CEO Sanjay Mehrotra said the board was confident enough to raise the quarterly dividend 30% after Q2, and Q3 results came in 24% above revenue guidance and 31% above EPS guidance. Operating cash flow in Q3 was $25.4 billion, up from $11.9 billion the prior quarter and $4.6 billion a year ago. That is not a company that is quietly deteriorating.

The hyperscaler spending cycle underpinning all of this is not evaporating either. JPMorgan raised its estimate for global AI-related capital expenditure through 2030 to $5.5 trillion in its June midyear outlook. Analysts forecast hyperscaler capex reaching $650 billion in 2026 alone, surpassing $1.1 trillion in 2027. Memory is a required input at every layer of that build-out.

What the Market Believes vs. What the Evidence Says

The market right now is pricing in one of two things: either the AI capex cycle is about to stall, or memory competition from China will compress Micron’s margins faster than anyone expects. At a forward P/E below 7x on a business earning $18 billion in quarterly free cash flow, that is essentially what is being implied.

The evidence says something different. Hyperscaler capex is not contracting. It is accelerating. Consensus estimates see U.S. semiconductor company earnings growing 98% in 2026, more than five times the rate of hyperscaler earnings growth. Micron’s multi-year supply agreements provide a durability the stock’s current valuation does not reflect. And the Standard Chartered CIO covering equities said this morning that