The screen is all red. The M7, semis, the KOSPI, and even memory and SK Hynix, the lone outperformers, all fell together.
One thing is clear here. Everything falling together is not rotation. If money had moved from A to B, one side would rise. When everything falls instead, people are not picking sectors, they are simply cutting risk itself and reaching for cash. If you sensed "are they raising cash?", that instinct is right.
This piece is the follow-up to the structure edition, Memory's Great Power Reversal. It is the "event edition," about why the share prices all collapsed even though the structure is intact. So why now?
Five Shocks Hit in a Single Week
Bad news that would normally have stayed separate stacked up within a few days.
Five Triggers Stacked in One Week
- 1. Broadcom guidance shock
- Q3 AI-chip revenue $16B vs $17.2B expected, stock -14%
- 2. Strong jobs data
- Rate-cut hopes vanish, Dec hike odds 0% → ~40%
- 3. Record-size IPO
- Drained cash from the market
- 4. Geopolitical shock
- Iran's Revolutionary Guard strikes a US base
- 5. Technical froth
- Already overextended, just waiting for an excuse to fall
It started with Broadcom. It guided Q3 AI-chip revenue to $16 billion, short of the $17.2 billion the market expected. The stock fell 14 percent that day, and in a single session $1.3 trillion of market cap evaporated from the semiconductor sector. The Philadelphia Semiconductor Index broke more than 10 percent in one day. More important than the number is the meaning: the belief that "AI demand is infinite" cracked for the first time.
Almost simultaneously, jobs data came in stronger than expected. Rate-cut hopes vanished, and the odds of a hike within the year suddenly appeared. The probability of a December hike, near 0 percent only a month earlier, spiked into the 40s. The newly arrived Kevin Warsh signaling a hike at his first FOMC added to it. On top of that, a record-size IPO drained cash from the market, and a geopolitical shock, Iran's Revolutionary Guard striking a US base, piled on. Finally, the market was already in a state of technical froth, overextended and just waiting for an excuse to fall. With five of these crowded into one week, assets that are normally unrelated were swept down in the same direction together.
So Why "Everything"? Three Mechanisms
A lot of bad news does not automatically mean everything falls together. For everything to move the same way, there has to be plumbing that connects them. There are three pieces.
First, rates are the master switch. The number-one cause of everything falling at once is almost always rates. When the risk-free rate rises, the discount rate that pulls future cash flows back to present value rises, and that marks down the value of every asset regardless of type. The hardest hit are high-valuation growth stocks that lean on far-future profits, namely AI and semis. On top of that, once cash starts paying more interest, you start to wonder, "is there any reason to take on risk at all?" That is exactly the rotation into cash, the source of that cash grab you felt.
Second, forced selling, and the winners get sold first. This is the most counterintuitive part. AI and semis were a crowded trade loaded with leverage. When the core names crack, the side that bet with debt is pushed into margin calls and redemptions. At that point you do not sell what you want to sell, you sell what you can. Losing junk does not raise cash when sold, so they sell the memory and semi names that had run up the most, the ones with the fattest gains. The best names become the ATM. That is why even SK Hynix and Samsung, the ones we had called the new bosses, fall too. This is what it means that correlations between names converge to 1 in a crisis.
Third, the narrative cracked. A stock that rose on momentum collapses the moment the last marginal buyer disappears. Once Broadcom cast doubt on the "infinite AI demand" story, a market that had been selling even on good news now started selling all at once on bad news.
Why Korea Hurt More
The KOSPI and KOSDAQ swung more than the US because two extra layers piled on. One is the currency. When the won weakens, foreigners dump Korean stocks first to avoid FX losses. The other is that Korean tech had run up the most. When liquidity is being withdrawn, the asset that rose the most becomes the first ATM. That is how the KOSPI's 8,000 line broke.
So, Is the Power-Reversal Thesis Dead?
This is where you need to stay cool. Do not mix two things. One is structure: memory's supply shortage and pricing power, a story that spans years and did not break this week. The other is liquidity and positioning: rates, leverage, risk aversion. What made this crash is the latter.
| Category | Structure (power reversal) | Liquidity / positioning (this crash) |
|---|---|---|
| Time frame | A multi-year story | A few days |
| Driver | Memory supply shortage, pricing power | Rates, leverage, risk aversion |
| Broke this week? | No | This is what made the crash |
| Signal to watch | Leading indicators (capacity, book-to-bill, spot) | The share price |
The more crowded a structural winner is, the harder it falls in a liquidity shock, even when the fundamentals are intact. If anything, it gets sold more precisely because it rose more. So concluding from this one drop that "the power reversal was wrong" is premature.
One thing does deserve real caution, though: the funding squeeze flagged as a risk in the structure edition is moving closer to reality. When rates rise, the cost of the private credit, SPVs, and lease structures that propped up AI-chip investment rises with them. The burden Big Tech had pushed off its balance sheet can come back on.
So how do you tell whether this is a brief liquidity fit or the real tail end of the cycle? The answer was already written in the structure edition: watch the leading indicators. If only the share price is falling, that is, it is macro and liquidity, while leading indicators like capacity guidance, book-to-bill, and spot prices stay intact, this is closer to a correction within a bull market. If those leading indicators actually start to roll over, that is the real beginning of the re-reversal. Which indicators to read, and in what order, is laid out head-to-tail in the structure edition's three-stage leading-indicator checklist.
In One Line
This crash is a cash grab, not rebalancing. The trigger is a deleveraging in which a rate pivot, a cracked AI narrative, a leverage unwind, and geopolitics all overlapped at once. To break the structural thesis, the leading indicators have to roll over, not the share price, and we are not there yet.
Disclaimer
This article is for informational purposes only and is not investment advice or a recommendation to buy, sell, or hold any security. The companies mentioned (Broadcom, SK Hynix, Samsung Electronics, and others) are analytical examples, and the author may hold some of these positions. All investing carries the risk of loss of principal, and past performance does not guarantee future results. Investment decisions are your own responsibility, and you should consult a qualified financial professional where appropriate.