Just yesterday, they were saying that memory is a strategic resource, and today the circuit breaker has been triggered.

Just one day ago, the market was still buzzing about the memory story: Micron signed a strategic partnership with Anthropic, a memory manufacturer became a shareholder in a large AI model company, and memory was redefined as a strategic resource that must be secured through advance contracts. The conclusion seemed almost irrefutable. Memory is no longer a cyclical stock.
One day later, the market put that conclusion to the test.

Today, global AI and chip stocks fell across the board. South Korea’s KOSPI plunged nearly 10% in a single day, triggering circuit breakers twice during the session; Samsung Electronics and SK Hynix both fell by about 12%, while Kioxia dropped more than 15%. In the US, the Nasdaq fell about 1.9% in early trading, while Alphabet had already dropped 6% in the previous trading session. The memory sector took the brunt of the fall: Micron plunged about 10% today. Yet it had just hit an all time high on June 18 and surged nearly 7% yesterday on the back of its partnership with Anthropic. In less than a day, the narrative shifted from euphoria to panic.

Many people’s first reaction was to blame a price cut by a certain AI model, which triggered a token price war. But that wasn’t today’s trigger. What really set things off was a convergence of three factors.

First, capital expenditures no longer align with free cash flow. Alphabet raised its 2026 capital expenditure guidance to 180 to 190 billion dollars, six times the 2022 level. This is double the 2025 figure, and the company explicitly stated that it would increase spending significantly again in 2027. At the same time, its first quarter free cash flow plummeted 47 percent year over year. What rattled the market even more was that, to finance this expansion, Alphabet completed an 85 billion dollar equity offering, the largest on record. This raised questions in the market: If even one of the companies with the most abundant cash reserves has to rely on the largest ever financing round to support its AI investments, when exactly will the returns materialize?

Second, talent drain. One of the co leads of Gemini and a co author of the seminal paper on the Transformer architecture defected to OpenAI; a few days later, the head of DeepMind’s AlphaFold team, a Nobel laureate, joined Anthropic. These two high profile departures have shaken market confidence in Google’s AI competitiveness.

Third, expectations for interest rate hikes are intensifying. Driven by rising oil prices due to the situation in the Middle East, inflation expectations have rebounded, and market bets on the Federal Reserve raising rates this year have jumped from about 57 percent a week ago to nearly 90 percent. Rising interest rates directly weigh on high valuation growth stocks.

As for the widely misreported token price war, it actually represents a deeper and more long term concern: what if AI inference continues to get cheaper? If high computing power is no longer a critical need for enterprises, can the hundreds of billions of dollars in annual capital expenditures still be sustained? If not, will downstream memory and hardware manufacturers still have as many orders? This line of reasoning certainly exists, and analysts have cited improvements in AI efficiency as one of Micron’s risks, but it serves as the underlying narrative, not the immediate trigger for today’s sell off.

So, why has memory taken the hardest hit?

Because, across the entire AI supply chain, memory is the most cyclical component and also the one with the most perfectly calibrated pricing. When the very premise underpinning its valuation, the massive capital expenditures by tech giants, begins to be called into question, the perpetual revaluation of memory built on that premise is naturally the first to be doubted. This is precisely the flip side of yesterday’s memory analysis: the narrative of the memory bull market and the panic driving this downturn are, at their core, two sides of the same coin.

It’s worth emphasizing that this decline isn’t a collapse of fundamentals, but rather a reconciliation between pricing and reality. HBM is still selling out, and gross margins remain at historic highs, these facts haven’t changed. What has changed is that stock prices had already priced in years of good news in advance. When a stock is priced for perpetual exponential growth, even merely maintaining guidance rather than raising it is interpreted as growth peaking. This is precisely what happened following Broadcom’s earnings report earlier this month.

The South Korean market also offers an intriguing detail: during this plunge, foreign investors net sold approximately 3.8 billion dollars in a single day, while local retail investors bucked the trend by net buying a record amount. Smart money is pulling out, while retail investors are stepping in to catch the falling knife. This in itself does not constitute a buy or sell signal, but who is catching the falling knife for whom is a question everyone following this industry chain should carefully consider.

The real verdict will come with Micron’s earnings report, scheduled for after the market closes on June 24.

The market expects Micron’s revenue for this quarter to grow by approximately 283 percent year over year, with a gross margin of about 81 percent; if realized, this would be the highest in the company’s history. However, the options market has already priced in approximately 17 percent one day implied volatility for this earnings report. With a stock that has just hit an all time high and boasts such high implied volatility, there is virtually no room for the market to be disappointed. More noteworthy than revenue are several other figures. First, whether the gross margin can hold at 81 percent. Once supply eases, gross margins are the first to decline, and they reflect more clearly than revenue whether the cycle is in the middle or at its peak. Second, whether HBM’s share of revenue continues to expand, this is the core indicator of the shift from cycle driven pricing to strategy driven pricing in the memory market. Third, the guidance for the next quarter and order visibility for 2027, as well as management’s wording regarding how long supply tightness will last; any ambiguity is a signal. Fourth, the contract prices for HBM, DRAM, and NAND, selling out is only meaningful if prices hold up. The market’s divergence is now out in the open: on one hand, investment banks have significantly raised their target prices for Micron and advised ignoring this sell off; on the other hand, some voices still insist that memory is ultimately a cyclical stock, and market traders are beginning to take a bearish stance on chip prices. While sell side analysts are calling for buys and raising target prices, the market is voting with its feet, this divergence is precisely the most noteworthy tension in this earnings report. Finally, let’s take a longer term perspective. In pursuit of high profits, memory manufacturers have shifted capacity toward data centers, resulting in a sharp rise in memory prices for consumer electronics. Some institutions predict that global smartphone shipments will see a record decline this year. While this shift in supply benefits memory manufacturers’ profits in the short term, it may backfire on end user demand in the medium term, which in turn could undermine overall demand for chips. The seeds of a boom often contain the seeds of a correction. Has the memory market truly broken free from its cyclical pattern, or is this merely a carefully packaged cycle? The market may find a preliminary answer to this question in tomorrow’s earnings reports.