Micron Reclaims $1,000 as Memory Ripped 6.10% Into a Closed Market — $50B Guide Lands September 30

Micron Reclaims $1,000 as Memory Ripped 6.10% Into a Closed Market — $50B Guide Lands September 30

DRAM contract prices tripled across two quarters and took gross margin from 39.0% to 84.9% | That's TradingNEWs

Itai Smidt 9/7/2026 12:12:18 PM

Key Points

  • Micron closed at $1,016.59, up $58.43 or 6.10%, in a 52-week range of $118.52 to $1,255.00.
  • Fiscal Q3 revenue hit $41.46 billion, up 346%, with an 84.9% non-GAAP gross margin.
  • Fiscal Q4 guidance of $50 billion and $31.00 EPS puts the stock at 8.3x forward earnings.

Micron Technology closed Friday, September 4, 2026 at $1,016.59, up $58.43 or 6.10% from $958.16, the third consecutive session of gains. The stock traded a $969.00 low and a $1,017.77 high — a 5.03% intraday swing — and settled at $1,014.95 in after-hours dealing, down $1.64 or 0.16%.

That move happened on a day the S&P 500 fell 0.38% and the Dow shed 271.86 points. Memory was the only place in the market with a bid: SanDisk ripped 11.9%, SK Hynix climbed 8.1%, Western Digital rose 5.86% to $467.46, and Seagate advanced with a year-to-date gain running above 200%. The Roundhill Memory ETF jumped 5% while SPY fell 0.4%, confirming a rotation into storage rather than a broad market rally.

US markets are closed Monday for Labor Day, so Friday's print stands as the live quote heading into a shortened four-session week.

The thesis here is that Micron has stopped being a cyclical semiconductor stock and started being priced as one anyway, and the gap between those two facts is the entire opportunity and the entire risk. Fiscal third-quarter revenue hit $41.46 billion, up 346% year over year with an 84.9% non-GAAP gross margin. Fourth-quarter guidance calls for $50.0 billion and $31.00 in EPS. On that run rate the stock trades near 8x forward earnings while the S&P 500 trades above 22x.

The market is applying a severe cyclical discount to a business carrying $100 billion in Strategic Customer Agreements, HBM3E and HBM4 fully booked through calendar 2027, and seven consecutive quarters of beating its own guidance.

The discount is not irrational — memory has destroyed investors at cycle peaks before, most recently in 2022 and 2023 when Micron and SK Hynix lost billions after overestimating how long pandemic demand would last. What has changed is contract structure. Multi-year fixed-price agreements convert a spot-price business into something closer to recurring revenue, and that is the argument the bulls have to win.

Fiscal fourth-quarter results land September 30, 2026. That report will either confirm the $50 billion guide or break the thesis in a single session. Twenty-three days of positioning separate now from then, and the stock sits 19.0% below its 52-week high of $1,255.00 with a 52-week low of $118.52 behind it.

The Tape: A 697% Year and a $1,255 Ceiling Overhead

The one-year chart is not a chart most large-cap equities produce.

Micron shares are up 697% over the past twelve months. The 52-week range runs from $118.52 to $1,255.00 — a spread of more than ten times. At $1,016.59 the stock sits 757% above its low and 19.0% below its high.

Peer performance frames it further. SanDisk has gained roughly 3,400% over the same stretch. SK Hynix's Korean-listed shares have surged 600%. The three DRAM incumbents have all repriced by multiples, driven by surging contract prices, ballooning gross margins, and free cash flow that arrived faster than any model anticipated.

The path within 2026 has been anything but linear. Micron traded near $880 in early August and around $911 by mid-month before the recent leg. It spiked 15.98% to $1,213.69 in the session after fiscal Q3 results in late June, then spent much of the summer correcting from that peak. The past two weeks have delivered a 5.15% gain with the stock rising in 6 of the last 10 sessions.

Recent momentum readings turned positive on Friday's close. One technical scoring system upgraded Micron from Hold to Buy Candidate with a score of 4.94 on a -10 to +10 scale based on data through September 4. The same system carries a three-month trend projection of a 9.69% decline, which captures the tension cleanly: near-term momentum is constructive, intermediate-term mean reversion is not.

Trailing valuation reflects how fast earnings have caught up to price. The trailing P/E sits at 23.01, and the dividend yield is 0.05% — effectively a rounding error and a signal that capital return has not been the priority during the build phase.

With roughly 1.15 billion shares outstanding, the market capitalization at $1,016.59 sits near $1.17 trillion. Micron has crossed into trillion-dollar territory on a memory cycle, which has no precedent in the company's history and few in the sector's.

Friday's daily range of $969.00 to $1,017.77 defines the immediate battleground. The close at the high end of that range, on the third straight up day, into a holiday weekend, with the sector broadly bid, is a constructive setup heading into Tuesday's reopen.

The gap risk cuts the other way too: SK Hynix rose another 8.26% in Seoul on Monday while US markets were shut.

Fiscal Q3: $41.46 Billion, 346% Growth, and an 84.9% Gross Margin

The June 24 report is the reason this stock is where it is, and the numbers still look implausible three months later.

Revenue for the third quarter of fiscal 2026, ended May 28, came in at $41.46 billion against $23.86 billion the prior quarter and $9.30 billion in the year-ago period. That is 346% year-over-year growth and 74% sequential growth, and it marked the fifth consecutive quarterly revenue record.

The margin expansion is the more remarkable line. Non-GAAP gross margin reached 84.9%, up from 39.0% a year earlier — a 4,590-basis-point improvement in four quarters. GAAP gross margin hit 84.6% against 37.7%. Operating margin printed 81.2%. Both gross and operating margins were company records.

Earnings followed. GAAP net income was $28.24 billion, or $24.67 per diluted share. Non-GAAP net income reached $28.86 billion, or $25.11 per diluted share.

The beats were not marginal. Revenue topped Micron's own guidance by $7.96 billion, a 23.8% overshoot, and beat Street consensus of $35.82 billion by $5.64 billion. Non-GAAP EPS of $25.11 exceeded the $20.71 consensus by $4.40, a 21% surprise. Against pre-quarter models that had penciled in figures a fraction of these totals, the report ranks as the largest quarterly outperformance a major-cap semiconductor company has delivered in recent memory.

Cash generation matched. Operating cash flow reached $25.39 billion against $11.90 billion the prior quarter and $4.61 billion a year earlier. Adjusted free cash flow hit $18.3 billion — both quarterly records.

The segment detail matters more than the headline. Core data center revenue more than doubled sequentially to $11.5 billion at an 87% gross margin. A greater-than-2x quarter-over-quarter move in the enterprise and hyperscale AI segment indicates the infrastructure buildout was still accelerating rather than plateauing as of late May.

Micron's DRAM market share expanded to 24%, with NAND share at 15%.

The full release and supporting schedules are filed at investors.micron.com.

For the September 30 report, the comparison base is now $41.46 billion. Beating it requires $50 billion.

The $50 Billion Guide and What September 30 Has to Confirm

Management's fiscal fourth-quarter guidance is the single most consequential number on the board.

Micron guided fiscal Q4 revenue to $50.0 billion plus or minus $1.0 billion, gross margin of approximately 86.0%, operating expenses of approximately $1.65 billion, and EPS of $31.00 plus or minus $1.00 on a share count near 1.15 billion. The tax rate for the quarter and the fiscal year was set at around 15.0%.

That guide exceeded pre-print Street consensus for the quarter by roughly $6.5 billion, or 15%, and implies 21% sequential growth off an already record base. Sequential acceleration of that magnitude at $41 billion in quarterly revenue is not normal behavior for any semiconductor business.

The line inside the guidance that deserves the most attention is management's own qualifier: the fiscal Q4 gross margin outlook reflects a meaningful moderation in the rate of price increases. Margin is still guided higher, from 84.9% to 86.0%, but the second derivative is turning. Prices are still rising; they are rising more slowly.

That is the first crack in the pricing story, and it came from the company rather than from a bear.

What September 30 has to establish, in order of importance: whether $50 billion was hit or exceeded, what fiscal Q1 2027 guidance looks like, and whether the gross margin trajectory flattens or reverses. A print at $50 billion with Q1 guided to $52 billion and margin held near 86% validates the durability argument. A print at $50 billion with Q1 guided flat and margin guided to 82% tells the market the peak is visible, and an 8x multiple will not protect the stock from a de-rating.

Micron has beaten its own guidance every quarter for seven consecutive quarters. That track record is the strongest single argument for the upside case and the reason the stock has held a trillion-dollar valuation through a summer of chop.

Operating expenses are set to rise by approximately $1 billion in fiscal 2027 as R&D expands. On a $200 billion annualized revenue base that is immaterial to margin, but it signals management is spending into the opportunity rather than harvesting it.

Confirmation of the report date came August 26.

HBM4 in Volume Production While Samsung Is Still Qualifying

The product position is where Micron's competitive case actually lives.

HBM4 is in high-volume production for the lead customer, with qualification samples sent to multiple end customers. HBM4E is already in development for calendar 2027 volume production, and HBM4E qualifications are underway ahead of schedule. Samsung, by contrast, is still qualifying HBM4 at some customers.

That gap — one generation ahead in qualification timing against the largest memory manufacturer on the planet — is the technical moat. High-bandwidth memory is not a commodity in the way conventional DRAM is. It qualifies into specific accelerator designs, the qualification process runs 12 to 18 months, and once a supplier is designed in, switching costs are substantial through the product life.

The order book reflects that. HBM3E and HBM4 are fully booked through calendar 2027 with demand visibility extending into calendar 2028. Micron has reportedly exhausted allocation for its most advanced memory production capacity through the end of 2026, which hands the leading manufacturers substantial pricing leverage for the remainder of the year.

Sold-out capacity plus a generation lead is the combination that produced an 87% gross margin in core data center.

The demand side keeps validating it. UBS raised its HBM pricing growth projection to 79% annually from a prior 67% estimate. Dell reported a $95 billion AI-server backlog. Nebius disclosed remaining performance obligations of $37.49 billion alongside first-half capital expenditures of $8.13 billion. Every one of those figures represents contracted demand that has to be filled with memory.

The capacity roadmap extends the position. Singapore is developing as a second center of excellence for advanced packaging, with HBM capacity expected to begin in the first half of calendar 2027. The New York fab cluster broke ground in January 2026 with Bechtel as construction partner.

The risk embedded in the lead is that it is temporal rather than structural. Samsung will qualify HBM4. SK Hynix already competes at the top of the stack and holds a comparable share. A twelve-month technology advantage in a market growing this fast is worth enormous margin — right up until it closes.

Watch the September 30 call for HBM4E qualification updates and any commentary on calendar 2028 bookings.

Strategic Customer Agreements: $100 Billion Booked, $22 Billion Deposited

The structural change management has been pushing is the one the market has not fully priced.

Micron has secured roughly $100 billion in Strategic Customer Agreements and holds $22 billion in incoming customer deposits. Mehrotra framed the SCA structure as a transformation of the business model rather than a commercial enhancement, describing it as a shift that will make revenue more predictable and durable over a multi-year horizon.

The mechanics matter. Traditional memory contracts price quarterly against spot, which is precisely why the sector has historically delivered 40% gross margins at the top of a cycle and negative margins at the bottom. A multi-year agreement with contracted pricing floors and prepaid customer deposits inverts that. It converts the supplier from a price-taker into a counterparty with committed volume and a known revenue floor.

Twenty-two billion dollars of customer deposits is the part that carries the most information. Customers do not prepay for commodities. They prepay for scarce capacity they cannot afford to lose, and the deposit itself is a form of collateral that makes cancellation expensive.

If the SCA framework holds through a downcycle, Micron's trough earnings look nothing like 2023's, and an 8x multiple on peak earnings becomes indefensible on the downside rather than the upside. That is the bull case in its cleanest form.

The unresolved question is what happens at the far end of those agreements. Contracted pricing floors protect against a price collapse for the duration of the contract. They do not protect against a renewal negotiated at market rates in a market that has normalized, and they do not extend beyond the terms already signed.

A capital return commitment is scheduled to take effect after December 9, 2026, which would mark the first meaningful shift from build-phase reinvestment toward shareholder distribution. At a 0.05% dividend yield and roughly $18 billion of quarterly free cash flow, the capacity for buybacks is enormous and entirely untapped.

Micron also established a $250 million Micron Ventures Paradigm Fund on August 13 to finance next-generation computing, and committed $10 billion on August 20 to build Micron Research Labs in Boise, Idaho, with construction phases starting in 2027.

Those are the capital allocation decisions of a company that believes the demand is structural.

DRAM and NAND Contract Prices: The Engine Behind the Margin

The margin expansion has a single explanation, and it is pricing rather than cost.

Conventional DRAM contract prices rose 93% to 98% quarter over quarter in the first quarter of 2026, followed by a further 58% to 63% in the second quarter. NAND flash contract prices climbed 85% to 90% and then 55% to 60% across the same two quarters. Worldwide DRAM revenue increased 57% sequentially in the second quarter while NAND revenue surged 70%.

Compounding those two quarters puts conventional DRAM contract prices roughly three times higher than where they started the year. That is the mechanism that took Micron's gross margin from 39.0% to 84.9% in four quarters. In a business where fixed costs dominate and incremental volume carries near-zero marginal cost, every dollar of price increase drops almost entirely to gross profit.

The supply-demand imbalance driving it is straightforward. AI infrastructure buildout requires HBM for accelerators and enterprise SSDs for training data storage. Both draw on the same wafer capacity that would otherwise produce conventional DRAM and NAND. Capacity redirected to HBM creates a shortage in everything else, and the shortage prices through the entire market.

The bull argument for why this cycle differs from prior ones rests on capital discipline. Samsung, SK Hynix, and Micron — the three companies that have long dominated DRAM — have shown more restraint than in past cycles, resisting the reflexive rush to add capacity that has historically converted every shortage into a glut within eight quarters.

The bear argument is that this has been said at the top of every memory cycle. Prices spike when demand outruns supply, manufacturers add capacity, that capacity arrives simultaneously, and prices collapse. The pattern has held for two decades.

The evidence that the second derivative is already turning comes from management. The fiscal Q4 gross margin guide of 86.0% explicitly reflects a meaningful moderation in the rate of price increases. Deceleration from a 60% quarterly increase to something lower is still growth, but the direction has changed.

Watch third-party contract pricing surveys through September. A quarterly increase that comes in below 30% would confirm normalization is underway and would compress Micron's multiple regardless of what the reported quarter delivers.

Balance Sheet: $24.4 Billion Net Cash Against a Record Capex Cycle

The financial position gives Micron optionality that prior cycles never afforded it.

Net cash stood at $24.4 billion at the end of fiscal Q3, a record liquidity position for the company. Operating cash flow for the quarter was $25.39 billion and adjusted free cash flow was $18.3 billion. Add the $22 billion in customer deposits and the funding structure for the current capacity build is effectively self-financing.

That matters because the build is enormous. Micron is investing at record levels in technology, products, and supply, per Mehrotra's own framing. The New York fab cluster, the Singapore advanced packaging center, and the $10 billion Boise research campus represent multi-year commitments running well past the current pricing environment.

The contrast with 2022 is the point. Micron entered the last downcycle having levered into capacity, then absorbed billions in losses when pricing collapsed. It enters this one with net cash exceeding $24 billion, prepaid customer deposits, and free cash flow of roughly $18 billion per quarter. If contract prices halved tomorrow, the company would still generate cash.

Annualizing the current run rate: $50 billion of quarterly revenue at 86% gross margin against $1.65 billion of quarterly operating expenses produces roughly $41.4 billion of quarterly operating income before tax, or approximately $35 billion after the guided 15% rate. That is a $140 billion annual earnings power figure against a $1.17 trillion market capitalization.

The capital return question is the open one. A capital return commitment takes effect after December 9, 2026, and with a 0.05% dividend yield the current distribution is negligible. A buyback authorization sized against $18 billion of quarterly free cash flow would be one of the largest in the sector and would put a floor under the stock during any post-peak de-rating.

Operating expenses are set to rise approximately $1 billion in fiscal 2027 to support R&D expansion. Against $200 billion of annualized revenue that is a rounding error, and it signals reinvestment rather than harvest.

Institutional positioning has split on exactly this question. Philippe Laffont bought Micron in the second quarter. David Tepper sold Micron and SanDisk and is hedging against one of their largest customers.

Two credible investors, opposite conclusions, same data.

Valuation: 8.3x Forward Earnings Against a 23x Trailing Multiple

The multiple is the most contested number in this entire analysis.

At $1,016.59 against FY2027 consensus EPS of $121.77, Micron trades at 8.3 times forward earnings. Using the fiscal Q4 guide of $31.00 as a run rate, annualized EPS approaches $124, putting the stock near 8.2x on a run-rate basis. The trailing P/E sits at 23.01, reflecting the four quarters before the margin explosion fully flowed through.

An 8x forward multiple on a trillion-dollar company with 86% gross margins is not a valuation the market applies to a business it believes is durable. It is the valuation applied to peak cyclical earnings, and it embeds an assumption that EPS falls sharply within two to three years.

The bear case is legible in that number. If the market believed $124 of annual EPS were sustainable, the stock would trade at a market multiple and be worth several times its current price. It does not, so the discount is doing the work of a forecast: consensus positioning expects normalization.

The sell-side disagrees. Forty-eight analysts polled carry a Strong Buy consensus with an average price target of $1,513, implying 48.84% upside from Friday's close. The dispersion is extreme — the low target sits at $361, a 64.5% decline, while the high reaches $2,200, a 116.4% gain. A separate compilation puts the average at $1,554.67, and another panel at $1,295.63. Lynx Equity set a $1,325 target for Micron alongside $2,450 for SanDisk, forecasting extended supply constraints.

A $361 low against a $2,200 high on the same stock is not analyst sloppiness. It is an honest representation of a business whose 2028 earnings could plausibly be $150 per share or $15 per share depending entirely on whether the SCA structure holds.

For context on how the market prices adjacent names, Western Digital's forward P/E has more than doubled over the past year from around 12x to about 29x, with Seagate showing a similar expansion. Hard-drive makers are being re-rated upward while the DRAM leader trades at 8x.

That divergence is either an inefficiency or a judgment that HBM earnings are less durable than nearline storage earnings. The September 30 guide will inform which.

Competition: SK Hynix, Samsung, SanDisk, and CXMT's Half-Trillion Debut

Micron holds the number-three position in a three-player oligopoly, and the competitive dynamics are shifting on two fronts.

Samsung, SK Hynix, and Micron cut across DRAM, NAND, and HBM, while nearly every other participant specializes in a single memory technology. Micron's DRAM share stands at 24% with NAND at 15%. The Roundhill Memory ETF's concentration illustrates the structure: Samsung Electronics at 25%, SK Hynix at 24.2%, and Micron at 23.8% of the fund, meaning three names drive most of the theme.

SK Hynix has been the strongest performer of the group operationally, with Korean shares up 600% over the past year and a US listing now trading. RBC initiated coverage at Outperform with a $200 target, citing a memory upcycle running for years. SK Hynix rose 8.1% Friday and another 8.26% in Seoul on Monday.

SanDisk has been the momentum leader, up roughly 3,400% over twelve months and closing Friday at $1,728.61 after an 11.17% gain. SanDisk and SK Hynix jointly released the first industry standard for High Bandwidth Flash through the Open Compute Project — an open UCIe-based blueprint that could expand the AI memory addressable market across chip vendors, and one Micron is not party to.

The genuinely new competitive variable is Chinese. CXMT completed an IPO in July at a market capitalization approaching half a trillion dollars and produces DRAM at scale. Mainstream Chinese DRAM competition is one of the explicit risks attached to Micron's margin structure, because CXMT competes in conventional DRAM — the segment where pricing has tripled and where a determined, subsidized entrant can do the most damage fastest.

CXMT does not compete at the HBM4 node today. It does not need to. Conventional DRAM prices are what took Micron's gross margin from 39% to 85%, and adding meaningful low-end supply compresses that spread without ever touching the high-bandwidth tier.

On the litigation front, an appeals court upheld Micron's patent suit win against Netlist on September 2, removing one overhang.

The read across for Micron: technology leadership in HBM4 is real and defensible through calendar 2027. Pricing leadership in conventional DRAM is neither, and roughly three quarters of revenue still comes from outside the highest-margin data center tier.

The Nvidia Commitment Cliff After Fiscal 2029

The clearest dated risk on the calendar comes from Micron's largest end customer.

Nvidia's second-quarter disclosure showed total supply and capacity commitments rising from $119 billion to $279 billion, described by its CFO as primarily related to the procurement of memory. The schedule breaks down as $92 billion for the remainder of fiscal 2027, $87 billion in fiscal 2028, and $88 billion in fiscal 2029 — $267 billion earmarked through a fiscal year that ends in January 2030.

Then it stops. Fiscal 2030 drops to $6 billion. Fiscal 2031 falls to $5 billion. Fiscal 2032 and beyond totals $1 billion.

An order book that goes from $88 billion to $6 billion in a single year is the most concrete forward-looking data point available on this stock, and it is not favorable.

The timing translation matters. Nvidia's fiscal years end in January while Micron's end in August. Nvidia's last clearly funded year of heavy memory spend runs February 2028 to January 2029, straddling the back of Micron's fiscal 2028 and the start of its fiscal 2029.

Two qualifications belong alongside that. Nvidia's purchase orders do not all convert to Micron revenue — some funds go to other components, and Nvidia sources HBM and DRAM from SK Hynix and Samsung as well. Shipments also lag commitments, so Micron can fill orders placed earlier well into 2029 regardless of what the forward table shows.

The structural read is more favorable than the headline. What Nvidia's schedule actually demonstrates is that memory suppliers now command multi-year visibility and binding commitments from the largest buyers in technology. Three years of contracted demand from a single customer is a level of forward booking the memory industry has never previously extracted.

The market implication is about timing rather than fundamentals. Markets discount forward, and a thin fiscal 2030 order book enters the pricing window roughly twelve months before it arrives. That places the de-rating risk somewhere between December 2028 and January 2029, not now.

A lean commitment table beyond 2029 is not evidence that demand collapses. Aging AI processors need replacement, and replacements need HBM. But absent explicit Nvidia commitments extending into the next decade, the discount currently embedded in an 8x multiple has a defensible rationale.

That is 27 months away. It is not this quarter's problem, and it is not this quarter's trade.

Technical Levels: $1,017.77 Overhead, $958.16 Pivot, $1,255 Target

The chart is cleaner than the fundamentals.

Friday's session established the immediate structure: a low of $969.00, a high of $1,017.77, and a close at $1,016.59 — within 12 cents of the session high. Closing at the top of a 5% intraday range on the third consecutive up day is momentum behavior, not distribution.

The first level to clear is $1,017.77, Friday's high and the round-number $1,000 psychological zone that the stock has just reclaimed. Above that, the path opens toward $1,213.69, the post-earnings spike close from late June, and then $1,255.00, the 52-week high. From $1,016.59, the distance to the June spike is 19.4% and to the 52-week high is 23.4%.

Downside references start at $969.00, Friday's low, followed by $958.16, Thursday's close and the pivot from which this three-day advance began. Losing $958.16 negates the current sequence and puts the stock back into the $880 to $911 zone it occupied through early and mid-August — a decline of 10% to 13% from spot.

The stock has found support near its 50-day moving average through a summer of elevated volatility across the broader chip sector, and that average has functioned as the operative floor for the entire recovery leg.

The broader semiconductor backdrop is mixed. The PHLX Semiconductor Index remains far higher for the year but well below its June peak, while software names have rebounded since July from sharp losses earlier in 2026. Memory has been decoupling from the broader complex, rallying on days the SOX and the S&P 500 both fall — Friday being the clearest example.

The catalyst calendar shapes the next three weeks. There is no company-specific event before September 30 earnings, which means the stock trades on sector flow, DRAM pricing data, and macro. Friday's US August CPI and the September 15-16 FOMC decision, with hike odds near 60%, are the two macro events capable of moving a long-duration growth stock 5% in a session.

The setup into earnings: constructive momentum, a defined pivot at $958.16, and a 23.4% gap to the 52-week high with 23 days of positioning ahead of a binary print. Options positioning into September 30 will get expensive.

What Would Break the Thesis

Three specific developments would invalidate the constructive case, and each is observable before the stock reprices.

The first is a gross margin guide below 82% for fiscal Q1 2027 on the September 30 call. Management already flagged a meaningful moderation in the rate of price increases when guiding Q4 to 86.0%. Moderation is tolerable. Reversal is not. Given that gross margin ran at 39.0% one year ago, the distance the metric can fall is enormous, and an 8x multiple offers no protection if the denominator halves.

The second is contract pricing data showing quarterly DRAM increases falling below 30%. The sequence has run 93-98% in Q1 and 58-63% in Q2. Deceleration to 30% would still be extraordinary in absolute terms and would still signal the inflection. Memory equities have historically peaked when the rate of price change peaks, not when prices peak — the second derivative leads the stock by roughly two quarters.

The third is capacity announcements from Samsung, SK Hynix, or CXMT. The entire bull case for this cycle differing from prior ones rests on capital discipline among the three DRAM incumbents. A large greenfield fab announcement from any of them, or aggressive conventional DRAM expansion from CXMT, restarts the pattern that has governed the sector for two decades: shortage, capacity, glut, collapse.

Two secondary risks deserve monitoring. Broader AI infrastructure financing uncertainty is a live variable — the demand underpinning $50 billion quarters comes from hyperscalers and neoclouds whose own capital structures depend on continued access to cheap funding, and a two-year Treasury yield at 4.37% with the Fed leaning toward a hike raises that cost. Broadcom's fourth-quarter revenue forecast of $34.8 billion coming in below some expectations was the first data point suggesting the AI capex curve may be flattening at the margin.

The other is customer concentration. Nvidia, the hyperscalers, and a handful of neoclouds represent the overwhelming majority of incremental HBM demand. Concentration works beautifully during a shortage and brutally during a normalization.

None of these have triggered. All are observable. Watch them in that order.

Verdict: Constructive Into September 30 at 8.3x Forward — $1,255 Is the Target, $958.16 the Line

The forecast is constructive with a hard event risk attached. Micron at $1,016.59 after a 6.10% Friday advance is trading at 8.3 times FY2027 consensus EPS of $121.77 and roughly 8.2 times the annualized run rate implied by its own $31.00 fiscal Q4 guide — a valuation the market reserves for peak cyclical earnings it expects to disappear. The counterargument is the strongest it has ever been in this company's history: $41.46 billion of fiscal Q3 revenue, up 346% year over year with an 84.9% non-GAAP gross margin and an 81.2% operating margin, $25.39 billion of operating cash flow, $18.3 billion of free cash flow, $24.4 billion of net cash, $100 billion in Strategic Customer Agreements, $22 billion of customer deposits, HBM3E and HBM4 fully booked through calendar 2027, advanced capacity sold out through the end of 2026, and seven consecutive quarters of beating management's own guidance. HBM4 is in high-volume production while Samsung is still qualifying it, and HBM4E qualifications are running ahead of schedule. That is not a commodity business and an 8x multiple does not describe one. What keeps the discount defensible is the shape of the cycle: DRAM contract prices rose 93-98% in Q1 and 58-63% in Q2, management has already flagged a meaningful moderation in the rate of increase, and Nvidia's own commitment table drops from $88 billion in fiscal 2029 to $6 billion in fiscal 2030 — a cliff that enters the market's pricing window around December 2028. CXMT's half-trillion-dollar debut adds a subsidized conventional-DRAM competitor to the segment where the pricing gains were largest. The near-term setup favors the bulls. The stock closed within 12 cents of its session high on the third straight up day while the S&P 500 fell 0.38%, SK Hynix added another 8.26% in Seoul on Monday, and 48 analysts carry a Strong Buy consensus with a $1,513 average target implying 48.84% upside. Base case into fiscal Q4 results on September 30: continued sector-driven grind with $1,017.77 as first resistance, $1,213.69 and the $1,255.00 52-week high as extended targets, and $958.16 as the pivot that invalidates the current sequence. The bull trigger is a Q4 beat on $50 billion with fiscal Q1 2027 guided above $52 billion and margin held near 86%. The bear trigger is a Q1 gross margin guide below 82%. Everything between now and September 30 is positioning.

That's TradingNEWS