Micron closed Wednesday at $904.28, down 8% on the day. By 10:40 AM ET Thursday, the stock had fallen further to $845.98, down $58.30 or 6.45% on the session. The $850 support level that options market participants had been defending with large put sales is now broken on an intraday basis. From the all-time high of $1,255, MU has now declined approximately 32.6% on a closing basis and 33% intraday. The two-day combined loss is approaching 14%, and the stock is trading at levels not seen since early May.
The catalyst driving today's specific leg lower is a development that, in isolation, sounds like good news for the semiconductor industry: ASML reported strong earnings and revealed that its extreme ultraviolet lithography machines can produce memory chips significantly more efficiently than previously understood. Yesterday, SK Hynix surged roughly 8% in Seoul on that news. Today, it gave it all back, falling 10.95% to 1,883,000 won, dragging the KOSPI into its 37th sidecar halt of 2026. Samsung Electronics fell 7.33% in the same session. The reversal happened despite ASML's results being unchanged. What changed was the market's read of the implication: better manufacturing efficiency means more supply, which means potential pressure on the pricing power that underpins every bullish memory thesis.
Why ASML's Good News Is Bad News for Memory Pricing
ASML's EUV efficiency disclosure matters because it changes the denominator of the memory supply equation. If the same number of machines can produce memory chips at higher volumes, then the effective supply capacity of the existing installed base increases without requiring additional capex. SK Hynix and Samsung, which are both planning to adopt ASML's next-generation EUV machines, would be able to expand memory chip output without proportionally increasing their capital spending.
Barron's framed this precisely: "Good News on the Memory Shortage Is Bad News for Chip Stocks." The logic is direct. Memory stocks trade at premiums because supply is structurally constrained relative to AI-driven demand. The entire thesis rests on sustained scarcity maintaining HBM pricing above historical averages. Anything that credibly eases that scarcity, including manufacturing efficiency improvements, weakens the pricing power argument even if the underlying demand remains strong.
The market is not saying demand is collapsing. It is saying the margin at which demand exceeds supply may be narrower in 2027 and 2028 than the most bullish models assumed, because supply can scale more efficiently than the capex forecasts implied. That is a multiple compression story, not an earnings collapse story. The distinction matters for understanding both the pace and the likely endpoint of the correction.
The CoreWeave Report and What It Actually Means
The second catalyst weighing on memory names Thursday is a Reuters report that CoreWeave explored ways to financially protect itself against a future drop in memory and storage prices. The article was careful to note that there was no indication CoreWeave actually expects a price drop, and that most industry analysts forecast memory costs will keep rising through at least 2027. CoreWeave declined to comment.
But the market reads headlines, not caveats. A cloud computing company exploring hedging strategies for memory prices is a signal, however indirect, that sophisticated buyers are at least aware of the possibility that the current price elevation is not permanent. That is enough to activate the risk management instinct among holders of memory stocks already sitting on large gains.
JPMorgan analyst Nikolaos Panigirtzoglou provided the structural context for why this particular concern has outsized market impact right now. After peaking in June, the assets under management of leveraged memory stock ETFs have shrunk by 34%, while all leveraged equity ETFs collectively have shrunk by only 13%. The AUM ratio of leveraged ETFs to the underlying stocks' market cap is three times higher for memory stocks than for equity ETFs broadly. This structural leverage concentration means that any credible downward price signal triggers amplified selling through the mandatory daily rebalancing that leveraged ETFs must execute to maintain their stated exposure ratios. The mechanism is mechanical, not fundamental: when the underlying stock falls, leveraged ETFs sell the underlying to reduce exposure, which pushes the price lower, which triggers more selling from other leveraged vehicles.
The Technical Picture at $850 to $870
The market is focusing on specific technical levels that have attracted attention from options market participants and technical analysts simultaneously.
The options flow data shows that institutional positioning is beginning to shift. Net premium in MU options turned positive in early Thursday trading, indicating that buyers of calls are paying more in aggregate than buyers of puts. Large put selling at the $840 and $800 strike prices, with implied volatilities of 301% and 238% respectively, signals that market makers and sophisticated traders are willing to sell downside protection at those levels, which is structurally equivalent to taking a position that those prices represent excessive pessimism relative to the fundamental value.
As of 10:40 AM ET Thursday, the $850 support level has been broken on an intraday basis, with MU trading at $845.98. The prior support cluster of $850 to $870 that had attracted large put selling has now become resistance rather than support. The next meaningful technical reference levels are the $800 to $840 zone, which is where the large put sales at the $800 and $840 strikes were clustered with implied volatilities of 238% to 301%. These strike prices, where sophisticated sellers collected premium, now represent the market's next implied floor estimate.
From a historical correction standpoint, MU's two prior corrections in this cycle were 27% from the October 2025 high and 37% from the February 2026 war-period high. A 37% correction from the June $1,255 all-time high implies a floor around $790. The current intraday low of $845 represents a 32.7% correction, meaning the stock is approaching but has not yet reached the deepest correction seen in this cycle. The 50-day moving average, which has historically served as a recovery target rather than a support level in prior MU corrections, currently sits near $923, now approximately 9% above the current price.
The 5.5x forward earnings multiple at current prices is the valuation anchor that analysts are referencing. Barron's has publicly argued that Micron could double from current levels as AI-driven memory demand moves the business beyond its historical boom-bust cycle, and the average Wall Street analyst price target sits around $1,579 according to FactSet data. Whether those targets are achievable depends entirely on whether the contracted revenue base and sold-out HBM supply through 2027 translate into the earnings trajectory the models project.
TSMC This Morning: The Demand Confirmation
TSMC reported Q2 earnings early Thursday, delivering exactly the kind of fundamental anchor that the memory sector needs to establish a floor. AI chips now account for approximately 61% of TSMC's total revenue. The company's own Q2 guidance of $39 to $40.2 billion in revenue, with gross margins of 65.5% to 67.5%, came in at or above the guided range. TSMC's trailing 12-month revenue through March 2026 was $131.7 billion, up 38.8% year over year. Management reiterated that demand for AI-related chips remains "extremely robust" and reaffirmed that the transition from generative AI to agentic AI is increasing computing requirements across hyperscale data centers.
The key management signal to watch from the conference call is any update to the full-year 2026 revenue guidance and the capital expenditure plan. If TSMC raises its full-year outlook above the existing "above 30% in USD terms" language and maintains or increases its $52 to $56 billion capex budget, it constitutes a company that manufactures the physical chips telling the market directly that the AI infrastructure buildout has sufficient multi-year visibility to justify record capital commitment. That is a more credible demand statement than any analyst note or sell-side forecast.
TSMC's results do not directly address the ASML efficiency concern or the CoreWeave hedging question. But they do address the more fundamental question underlying both of those: is AI chip demand growing fast enough to absorb whatever incremental supply improvements ASML's technology enables? Given that TSMC's June monthly revenue showed 67.9% year-over-year growth and AI chips account for 61% of revenue, the evidence from the physical manufacturing layer remains strongly affirmative.
When Does the Correction Find a Floor?
The historical, fundamental, and options market evidence converge on a framework for evaluating where the floor forms, without offering a precise price target.
From a historical correction standpoint, the current drawdown of approximately 35% is within the range of prior MU corrections in this cycle. If the current correction mirrors the largest prior one at 37%, the floor would be in the low $790s. If it stops at the current level, the floor is approximately $850 to $870 based on the support cluster attracting options activity.
From a fundamental standpoint, three events remain as potential catalysts for stabilization. SK Hynix reports July 29 and is expected to deliver the strongest quarter in company history. The big tech earnings cycle, specifically Microsoft and Alphabet in late July, will provide the most direct available read on whether hyperscaler AI capex commitments are holding or being revised. And the July CPI data in mid-August will determine whether the Fed's September rate hike probability continues to compress multiples or begins to ease.
The options market positioning shift from net negative to net positive premium in MU Thursday morning is the most current leading indicator available. When sophisticated options sellers are willing to collect premium at put strikes of $800 and $840 with implied volatilities above 200%, they are expressing the view that the stock is unlikely to reach those levels. That is not a guarantee. It is the market's collective current read on the probability distribution of outcomes.
The leveraged ETF AUM shrinkage from 34% suggests that the most mechanical source of forced selling is already partially exhausted. When leveraged ETF AUM falls from peak by that magnitude, the marginal selling pressure from daily rebalancing diminishes, because the vehicles have less total exposure to unwind on any given down day.
The summary of where the evidence points: the fundamental case for memory has not deteriorated, the momentum unwind and leverage-driven selling are measurably diminishing, technical support exists in the $850 to $870 zone, and the $840 and $800 put strikes are attracting significant selling from sophisticated participants. None of these observations guarantee a specific floor. All of them are consistent with a market that is in the late stages of a sentiment-driven correction rather than the early stages of a fundamental breakdown.
Explore more exclusive insights at nextfin.ai.
