Maintaining Outperform: Q1 GM 530bps Above Guide, FQ2 Guide of $18.7B / 68% GM / $8.42 EPS Resets the Model, CY26 HBM Fully Sold Out on Volume AND Pricing, $100B HBM TAM Pulled Forward Two Years

Rating Action: Maintaining Outperform

This is the fourth report in our MU coverage arc. We are maintaining Outperform on the back of the Q1 FY2026 print and FQ2 FY26 guide. The arc:

• Q2 FY2025 (Initiating at Hold, constructive bias) — March 21, 2025. Opened at Hold on three concerns: consolidated GM compression (37.9% Q2 actual, 36.5% Q3 guide); NAND under-absorption flowing through inventory; DIO at 158 days vs. the 120-day target. The HBM ramp and leading-edge DRAM positioning were real, but the consolidated P&L still reflected the legacy NAND cycle.
• Q3 FY2025 (Upgrading to Outperform) — June 26, 2025. Every Hold-gating concern resolved the right direction. Q3 GM 39.0% beat the 36.5% guide; FQ4 guide $10.7B / 42% GM / $2.50 EPS comfortably cleared the Street; HBM run-rate ~$6B; 12-high crossover pulled into FQ4; AMD MI355X added as a fourth HBM customer; DIO compressed to 139. Moved FY26 EPS range to $11-13. Downside triggers: HBM 12-high yield slip, tariff broadening.
• Q4 FY2025 (Maintaining Outperform) — September 24, 2025. Q4 revenue $11.3B beat $10.7B guide by ~$600M; GM 45.7% beat 42% guide by 370bps; EPS $3.03 beat $2.50 guide by $0.53. FQ1 FY26 guide of $12.5B / 51.5% GM / $3.75 EPS extended the trajectory and cleared 50% GM a quarter early. HBM customer base widened from four to six; HBM run-rate ~$8B; CY26 HBM3E pricing largely locked; DIO compressed to 124. Widened FY26 EPS range to $16-19. Downside triggers moved from HBM execution to HBM4 production-ramp slip, hyperscaler capex pause, NAND price reversal.
• Q1 FY2026 (Today — Maintaining Outperform). The thesis compounds in a way the prior bridges did not anticipate. Revenue $13.6B beat the $12.5B guide by ~$1.1B; GM 56.8% beat the 51.5% guide by 530bps; EPS $4.78 beat the $3.75 guide by $1.03. The FQ2 guide of $18.7B / 68% GM / $8.42 EPS is +38% sequential on revenue, +1,120bps sequential on GM, and +76% sequential on EPS — a step-change of an order beyond what the FY26 mid-single-digit-quarterly-step-up bridge implied. CY26 HBM is fully sold out on volume AND pricing (vs. "vast majority of HBM3E" plus "HBM4 in the coming months" three months ago). The $100B HBM TAM was pulled forward two years to 2028. Idaho fab #1 timeline pulled in by half a year. CapEx stepped up to ~$20B (vs. ~$18B implicit). Free cash flow set a quarterly record at $3.9B with $2.7B of debt redeemed in-quarter.

The case for Maintaining Outperform: (1) The "tight supply" framing has hardened from base case into structural fact. Sanjay's prepared-remarks language — "aggregate industry supply will remain substantially short of the demand for the foreseeable future" and management can only meet "half to two-thirds" of the demand from several key customers in the medium term — goes substantially further than the Q4 "tight supply" framing. The 3:1 HBM-to-DDR5 trade ratio expanding with each HBM generation is now management's load-bearing supply argument, and it is mechanical rather than cyclical. (2) CY26 HBM is fully sold out on price and volume, removing the largest demand-visibility unknown going into the year. (3) The customer-contract structure is changing: multi-year LTAs with "specific commitments" and "much stronger contract structure" are under negotiation across DRAM and NAND (not just HBM), which changes the cycle-cyclicality calculus that has historically capped MU's terminal multiple. (4) Net cash with $3.9B quarterly FCF — the balance sheet is no longer a constraint on the CapEx step-up to ~$20B, and the accelerated Idaho ramp can be funded out of FCF. (5) The CY26 bit-shipment cap is set at +20% for the industry and for Micron individually, making this a price-driven cycle from here, with content/mix doing the rest.

Maintain-Outperform triggers met from the Q4 framing: "FQ1 GM at/above 51.5% with FQ2 sequentially up." — FQ1 actual 56.8%; FQ2 guide 68%. "HBM CY26 supply largely sold out before CY26 begins." — explicit on the call: fully sold out, volume and price. "HBM4 production ramp on track for CQ2 26." — reiterated, with HBM4 yield ramp expected to be faster than HBM3E. "DIO continues to compress; FY26 framework intact." — DIO 126 (vs. 124 prior; a one-day uptick on inventory build into a tightening market is constructive); DRAM still below 120 days. All four anchors met cleanly; the FQ2 GM guide of 68% is the datapoint that decisively settles whether the FY26 GM bridge can bear the FY26 EPS upside we now think probable.

Downside triggers from Outperform: HBM4 yield ramp execution slip in CQ2 2026 (the planned first-shipment window); CY26 hyperscaler capex pause that softens HBM/HC-DIMM pull; non-HBM DRAM softening if 1-gamma ramps faster than expected at one or more competitors (the supply-discipline question shifts from "MU's discipline" to "the industry's discipline" once everyone is at the leading edge); a tariff regime that materially impacts customer pull (management explicitly excluded tariff impact from FQ2 guidance); and the secondary risk that the multi-year LTA discussions stall on length/specificity/penalties — if a major customer balks at the new contract structure, the visibility premium that anchors the Outperform compresses.

Results vs. Consensus

The print is a beat on every line that matters — revenue, GM, EPS, segment composition, cash flow, balance sheet, and a forward guide that rebases the consensus model. The defining datapoints are the GM swing (56.8% vs. 51.5% guide is +530bps upside) and the FQ2 guide (68% GM is +1,120bps sequential, an unprecedented step-up at this revenue scale). Three consecutive quarters of triple-digit-bps GM upside above guide is no longer a pattern; it is the new operating cadence.

Quality of Beat

• The GM beat decomposition is clean and supply-driven. 1,110bps QoQ GM expansion against a guide of +580bps is a 530bps upside surprise. Mark Murphy attributed the swing to "higher pricing with strong cost execution and favorable mix" — the weight is on pricing, given DRAM ASP +20% and NAND ASP +mid-teens% sequentially with bit shipments roughly flat to up modestly. This is the inverse of the Q3 print, where volume did the work. The sustainability question shifts to whether non-HBM DRAM pricing can hold through CY26 — management's answer is yes, on the supply-side rigidities described in the prepared remarks.
• Revenue mix is heavily DRAM-led. DRAM $10.8B (+20% QoQ) at 79% of total is now the dominant top-line driver. NAND $2.7B (+22% QoQ) at 20% of total is mix-led on data center NAND strength — data center NAND revenue exceeded $1B in the quarter for the first time. The Q4 mobile-managed-NAND-exit signal is reflected in the cleaner segment numbers; portfolio discipline is showing up in the mix.
• Segment GM dispersion has compressed massively. The Q4 segment GM range was 31% (AEBU) to 59% (CMBU), a 2,800bps spread. Q1 dispersion is 45% (AEBU) to 66% (CMBU), a 2,100bps spread. Every BU expanded GM materially (CMBU +620bps, CDBU +990bps, MCBU +1,700bps, AEBU +1,400bps), with MCBU's 17pt GM expansion the largest single-segment swing in the print. The lift across non-CMBU segments is the datapoint that argues this is a market-wide supply story, not a CMBU/HBM-only story.
• Cash flow inflection at unprecedented scale. $3.9B quarterly FCF (vs. $0.8B at Q4) on $8.4B OCF and $4.5B CapEx. The prior FCF record (FQ4 2018, ~$3.0B) was beaten by >20% even with elevated CapEx. Net cash positive with $250M+ at quarter-end after $2.7B of in-quarter debt redemption. This shifts the FY26 capital-allocation conversation: the company can fund the ~$20B CapEx step-up entirely from FCF and still take down debt or repurchase shares ($300M repurchase already executed under CHIPS-permitted terms).
• DIO uptick is constructive context. Inventory $8.2B at 126 days (vs. 124 days at Q4) reflects a $150M sequential decrease in absolute inventory dollars but a 2-day DIO uptick on the cost-of-revenue base. DRAM days remain below the 120-day target. Read: management is not building inventory; the days math is moving on the cost denominator. With supply structurally short, days will trend higher mechanically as Micron lengthens lead times to customers; this is not the legacy demand-air-pocket signal.

Segment Performance

All four business units printed quarterly records on revenue. The new BU disclosure (introduced at Q4) is now serving its purpose: segment GM is the cleanest read on which parts of the portfolio are pricing-rich versus volume-rich, and the dispersion has compressed sharply on the across-the-board pricing strength.

CMBU: 66% Segment GM, $100B HBM TAM Pulled to 2028

Cloud Memory revenue of $5.3B (+16% QoQ, 39% of total) is the segment that prints first and matters most. Segment GM of 66% (+620bps QoQ) is sustained by HBM mix and cost execution, with bit shipments and pricing both contributing this quarter (vs. mostly cost+mix at Q4). The HBM ramp continues at a sequential record, although management declined to break out the dollar figure ("HBM revenue was a record" without quantification — a step back in disclosure granularity that is the price of the CY26-sold-out posture).

"We have completed agreements on price and volume for our entire calendar 2026 HBM supply, including Micron's industry-leading HBM4. We forecast an HBM TAM CAGR of approximately 40% through calendar 2028, from approximately $35 billion in 2025 to around $100 billion in 2028. This $100 billion HBM TAM milestone is now projected to arrive two years earlier than in our prior outlook. Remarkably, this 2028 HBM TAM projection is larger than the size of the entire DRAM market in calendar 2024." — Sanjay Mehrotra, CEO

HBM datapoints from the call. (1) CY26 supply fully sold out on volume AND pricing — explicitly including HBM4 (vs. Q4's "vast majority of HBM3E" + "HBM4 in coming months"). (2) HBM4 first production ramp on track for CQ2 2026; yield ramp expected to be faster than HBM3E (vs. Q4's "in line with customer demand"). (3) HBM4 specs: industry-leading speed over 11 Gb/s; advanced CMOS and metallization on internally manufactured base logic die. (4) HBM CY26 mix will be HBM3E + HBM4, both contributing strong YoY growth. (5) HBM4E custom-base-die customer engagements progressing; manufactured by TSMC; "higher gross margins than standard HBM4E" reiterated. (6) Customer base no longer disclosed by count — the Q4 "six customers" framing has been retired in favor of "the entire ecosystem of HBM customers."

The TAM revision is the headline. $100B HBM TAM by 2028 vs. the prior $100B-by-2030 framing — pulled forward by two full calendar years. The 2028 number alone is larger than the entire CY24 DRAM market (~$80B). The 40% CAGR through 2028 implies a 2026 HBM TAM of ~$50B and 2027 HBM TAM of ~$70B (geometric interpolation). Against this, Micron's CY26 HBM is fully sold out on the current supply trajectory, with the share commentary now retired in favor of "managing the mix" — the implication being that supply, not share, is the binding constraint on CY26 HBM revenue.

LPDRAM-for-server momentum continues. 192GB LP SOCAM2 product sampled, enabling 50% capacity-per-module increase and rack-scale LPDRAM density >50 TB. Assessment: CMBU is now a $21B-annualized, 66%-GM segment — alone, this segment generates ~$14B of gross profit annually, more than the entire FY24 Micron consolidated GP. The structural anchor that we flagged at Q4 has compounded.

CDBU: +990bps GM Step on the Strongest Sequential Revenue Growth

Core Data Center revenue of $2.4B (+51% QoQ, 17% of total) is the largest sequential growth segment in the quarter, and it carries enterprise/government DRAM and the data-center SSD business. Segment GM of 51% (+990bps QoQ) reflects pricing leverage on the strong bit-shipment ramp. Data center NAND revenue exceeded $1B for the first time, driven by enterprise SSD strength (third in market share, with HSur from JPMorgan flagging ~25% sequential growth in third-party data); the world's-first PCIe Gen6 SSD on G9 NAND has rapidly increasing hyperscaler-qual commitments; QLC-based 122/245TB G9 SSDs in qualification at multiple hyperscalers. Assessment: CDBU has emerged as the second-most-economically-meaningful segment (CMBU + CDBU together = 56% of revenue at blended ~61% GM, vs. ~54% / 54% blended at Q4). The data-center HDD-shortage tailwind into NAND is now showing up in the numbers.

MCBU: 54% Segment GM Up 17 Points QoQ — The Standout

Mobile & Client revenue of $4.3B (+13% QoQ, 31% of total) on lower bit shipments offset by sharply higher pricing. Segment GM expanded 17 points to 54% — the largest single-segment GM swing in the print, in a segment that historically anchored the lower end of MU's mix. Drivers: tight DRAM supply driving pricing power into PC and smartphone, with content growth (12GB+ DRAM in 59% of CQ3 flagship smartphones, more than double a year ago) layering on. PC unit growth raised again to high-single-digits CY25 (vs. mid-single-digits at Q4) on Windows-10 EOL and AI PC adoption, with management flagging that 2026 unit growth could be capped by memory supply rather than demand. The 1-gamma 16Gb LPDDR6 sampled to leading OEMs in Q1 — ">50% higher performance and improved power efficiency for flagship smartphones and AI PCs." Assessment: MCBU has gone from a margin drag to a margin contributor in three quarters — 36% Q4 → 54% Q1 is the second-largest single-segment swing in MU's recent history (only AEBU is larger, on a smaller base). The "memory pricing impacts elasticity" question Vivek Arya raised is real but, by management's read, already reflected in the unit-growth caps in their forecast.

AEBU: D4/LP4 Allocation Drives 14-Point GM Expansion

Auto & Embedded revenue of $1.7B (+20% QoQ, 13% of total) at 45% segment GM (+14pts QoQ) reflects robust auto/industrial demand against allocation-priced legacy DRAM. ASIL-rated LPDDR5X and UFS 4.1 NAND products with billions of dollars of automotive/robotics design wins. Industrial demand strengthening on autonomous-systems adoption (factory automation, aerospace/defense, humanoid robotics, edge networking, video surveillance). Manassas/Virginia fab investment continues to anchor long-term D4/LP4 supply. Assessment: AEBU has more than doubled its segment GM in two quarters (31% Q4 → 45% Q1) — the cleanest expression of the "D4/LP4 in allocation" pricing dynamic on the leading-edge node migration constraint described in the prepared remarks.

Key Topics & Management Commentary

Overall management tone: Forceful and explicit on the demand-supply asymmetry. Sanjay's lead framing — "best competitive position in [Micron's] history" and "one of the semiconductor industry's biggest enablers of AI" — is another step up from the Q4 "positioned better than ever" formulation. The "half to two-thirds of demand from several key customers" datapoint is one of the strongest demand-supply gap statements management has ever made. Mark Murphy's GM commentary moved from describing the FQ2 step-up to framing margin expansion as continuing through the year, but "more gradual" past FQ2 because of the math at high GM levels — this is the first time management has framed the bridge as bounded rather than open-ended.

The CY26 HBM Sold-Out Update: Full Sold Out, HBM4 Included

The Q4 framing was: "vast majority of CY26 HBM3E supply" locked, plus "HBM4 pricing/volume agreements expected to close out total CY26 HBM supply in the coming months." Q1 closes that loop fully. Sanjay's explicit language: "We have completed agreements on price and volume for our entire calendar 2026 HBM supply, including Micron's industry-leading HBM4." The visibility implication is meaningful: with pricing locked alongside volume, the HBM portion of CY26 revenue becomes a near-deterministic figure rather than a price-sensitive estimate, with the remaining variables being yield ramp on HBM4 (which management called "faster than HBM3E") and the sequencing of HBM3E→HBM4 transition through CY26.

Sanjay also retired the customer-count disclosure (Q4 had six). The stated reason — "engaged with the entire ecosystem of HBM customers" — is consistent with the full-sold-out posture: granular customer-by-customer disclosure is no longer informationally useful when the aggregate supply is fully committed. The signal: HBM allocation conversations are now competitive across the entire AI-accelerator ecosystem, with the customer mix decided by Micron's allocation discipline rather than by demand availability.

The $100B HBM TAM Pulled Forward Two Years

Prior framing (set several quarters ago): $100B HBM TAM by 2030. New framing: $100B by 2028 at ~40% CAGR from ~$35B in 2025. This is the single largest forecast revision Sanjay has issued in the cycle. The arithmetic implication: 2026 HBM TAM ~$50B, 2027 ~$70B, 2028 ~$100B (geometric interpolation). Against this, Micron's CY26 HBM is fully sold out at the supply Micron has, suggesting Micron's CY26 HBM revenue lands in the $13-18B range under reasonable share assumptions (~25-35% of the ~$50B TAM, depending on Micron's mix share within the leading-edge HBM3E/HBM4 supply). Sanjay declined to provide share guidance, but the supply-constrained share posture (vs. capacity-led share gain) means upside surprises in CY26 are unlikely to come from share — they would come from pricing and from HBM4 volumes ramping faster than the implicit mix assumption.

Multi-Year Customer LTAs: New Contract Structure Across the Portfolio

Tim Arcuri (UBS) opened the Q&A on long-term agreements. Sanjay's response is the most structurally interesting commentary of the call. Key points: (1) under discussion with "several key customers" across DRAM and NAND; (2) "very different from prior LTAs" with "specific commitments" and "much stronger contract structure"; (3) multi-year in nature; (4) span beyond data center customers to "multiple customers across our market segments"; (5) data center SSDs are part of the discussion. The pattern suggests Micron is using the structural-tightness window to repaper customer relationships on terms more favorable to the supplier, with specific commitments (volume floors, pricing structures) that historically were one-sided in the customer's favor.

Read: if these LTAs land at the contract specificity Sanjay is describing, MU's terminal cyclicality discount compresses materially. The cycle-cyclicality argument has historically been the primary reason MU trades at a structural discount to NVDA/AVGO in the AI-stack composite. Multi-year LTAs with hard volume/pricing commitments would partially defang that argument by materially derisking the trough-to-peak earnings amplitude.

Capital Allocation: ~$20B FY26 CapEx, "Roughly Double" Construction Spend, Idaho Pulled In

FY26 CapEx raised to ~$20B (vs. ~$18B implicit at Q4), weighted to the second half of the fiscal year, with construction "roughly double" the FY25 figure. Composition: HBM supply capability + 1-gamma DRAM nodes. CJ Muse pressed on whether the CapEx growth was conservative against the supply gap; Murphy's response was that clean room build-out has irreducible lead times and Micron is doing everything possible (pulling in tools, accelerating construction). The trade-off: capacity-add cannot be accelerated meaningfully beyond what is already planned; the gap-closing happens through node transitions (1-gamma majority of CY26 bit growth) rather than greenfield.

Idaho fab #1 first-wafer-out pulled to mid-CY27 (vs. 2H CY27 prior — 6+ months earlier). Idaho fab #2 begins construction in 2026, operational by 2028. New York fab groundbreaking in early 2026, supply in 2030+. Japan: technology and manufacturing investments coordinated with Boise R&D; clean room added at Hiroshima for advanced nodes. Singapore HBM advanced-package facility on track to contribute meaningfully to HBM supply in CY27. India assembly/test in pilot production, ramping in 2026.

FY27 CapEx framed as "up." Capital intensity dropping vs. revenue, even at $20B FY26, because revenue is stepping faster than CapEx; the ~25-30% CI Tim Arcuri observed is below the 35% historical metric, with management framing this as a feature of demand-led-revenue rather than supply-led-CapEx.

FY26 Industry Outlook: Bit Shipments +20% (Industry AND Micron)

CY25 DRAM bit demand growth raised to low-20% range (vs. high-teens at Q4). CY25 NAND bit demand growth raised to high-teens (vs. low-to-mid-teens at Q4). For CY26: industry DRAM and NAND bit shipments both projected at ~+20% YoY — the industry's bit-shipment ceiling is set by the supply constraint, not by demand. Micron's own bit shipments also expected at ~+20%, in line with industry. Server unit demand expectation raised to high-teens for CY25 (vs. ~10% at Q4). PC unit demand raised to high-single-digits for CY25 (vs. mid-single-digits at Q4) on Windows-10 EOL and AI PCs.

The capping of CY26 bit shipments at ~+20% is the single most important supply-side datapoint in the print. With demand growing faster than supply across DRAM and NAND, CY26 becomes a price-and-mix story by construction. The bit cap also makes the multi-year LTA discussion structural: if customers want supply visibility past CY26, they have to negotiate it now, not optimize on spot.

1-Gamma DRAM and G9 NAND: Technology Cadence Continues

1-gamma DRAM ramping well; will be primary driver of CY26 bit growth and majority of bit output in 2H CY26. Development underway on 1-delta and 1-epsilon nodes. G9 NAND ramping with robust yield ramps across data center and client SSDs; QLC mix at a record high; G9 expected to become the largest NAND node later in FY26. CY25 was a record year for Micron on internal AND customer quality measures. Read: the technology pipeline supporting the GM bridge through CY26 and into CY27 is intact and arguably ahead of plan.

Analyst Q&A: Selected Exchanges

Six questions in the queue (with one technical glitch when management's lines were muted briefly mid-CapEx-answer to Tim Arcuri). Notable feature: every question accepted the print as a given and pushed on either contract structure (LTAs, HBM mix, share), supply discipline (CapEx, clean room), or out-quarter margin trajectory. No question pushed back on the supply-tightness narrative, on FQ2 sandbagging, or on the durability of the pricing leverage. The Street is now substantially aligned with management's structural-tightness framing.

Tim Arcuri (UBS): Multi-Year LTAs and Capital Intensity

Arcuri opened on the nature of customer LTAs — how the bundling of DDR5 with HBM and NAND is structured, what timeframe the agreements stretch to (he flagged hearing of 2027/2028 contracts in the channel). Sanjay confirmed: multi-year, specific commitments, much stronger contract structure than prior LTAs. Declined to name customers or quantify volumes. Arcuri's follow-up to Murphy on CapEx: at $20B net CapEx, capital intensity is ~25-30% (below the 35% historical metric); is the gap because of fab-space constraints, and does the catch-up roll into FY27? Murphy: yes, brick-and-mortar construction "roughly doubles" FY25-to-FY26; FY27 expected to be "up"; Micron will remain disciplined on CapEx growth. Sanjay added: capital intensity is dropping because the market-condition setup is so constructive; Micron is being efficient with the spend.

CJ Muse (Cantor Fitzgerald): CapEx Conservatism; FY26 Cost-Down Trajectory

Muse pressed on whether the CapEx step-up was conservative given the demand backdrop. Murphy's response was the most candid statement of supply-discipline reality on the call: clean room space takes time, there is no near-term solution, the entire industry is short to demand and Micron is no different. Sanjay added the "half to two-thirds of demand from several key customers" datapoint — an unusually direct framing of unmet demand. On cost-down through CY26: Murphy declined out-quarter cost guidance but reiterated 1-gamma DRAM and G9 NAND ramps as cost tailwinds; some startup costs from new fab construction will land in CY26-27 but be relatively small at the operating scale. On the HBM3E→HBM4 transition cost question: Sanjay said HBM4 yield ramp is expected to be faster than HBM3E, with strong yield progression already.

Harlan Sur (JPMorgan): ASIC XPU Demand and Enterprise SSD LTAs

Sur asked about the upward revisions to ASIC XPU volume forecasts (Google TPU, AWS Trainium) all using HBM3E, and how Micron manages the 3E demand upside alongside the HBM4 ramp given CY26 is fully contracted. Sanjay: the CY26 HBM mix will be HBM3E + HBM4; engaged with the entire ecosystem of HBM customers; mix will be managed based on customer requirements; flexibility from front-end fungibility (1-beta wafer base shared across DRAM products). On enterprise SSDs (Sur cited ~25% sequential SSD growth from third-party data): Sanjay confirmed strong share gains, Micron is also negotiating multi-year LTAs on enterprise SSDs, and noted that storage intensity rises with inferencing/video AI workloads — not just training.

Thomas O'Malley (Barclays): HBM-as-% of DRAM and Competitive Positioning into CY26

O'Malley pushed on quantifying HBM as a % of total DRAM revenue and on competitive positioning into CY26 (with one large competitor publicly seeking re-entry on HBM3 qualification). Sanjay declined to break out HBM as % of DRAM dollars; reiterated industry-leading HBM4 specs (>11 Gb/s, 30% lower power than competitors on HBM3E); reiterated that share is no longer the framing — the framing now is mix management between HBM and non-HBM, with both sides of the portfolio in tight supply and both at "strong profitability." On the CY3E share question, declined to comment.

Krish Sankar (TD Cowen): GM Trajectory Past FQ2; HBM Share Vector to 2028

Sankar asked whether GMs continue to expand past FQ2 (May quarter); how to think about the trajectory. Murphy's framing: not guiding past FQ2, but margins should be up for both DRAM and NAND, supported by tight market conditions and continued cost execution; expansion past FQ2 will be "more gradual" than the FQ1/FQ2 steps because at high GM levels the math compresses (every $1 of incremental price contributes less in GM percentage). On the HBM share question for 2028 (against the new $100B TAM): Sanjay declined to specify share, framing the answer as portfolio-mix management rather than a share target, but reiterated industry-leading product positioning.

Chris Danely (Citi): LTA Timing and HBM Pricing Float

Danely pushed on when LTAs will be signed and what the holdup is (length? size?), and whether AI customers are contributing to fab build-outs. Sanjay declined specifics on signing timing or customer co-investment, but reiterated the differentiated contract structure and specific commitments. On HBM pricing for CY26: confirmed pricing is locked alongside volume for full CY26 supply; no float mechanism (vs. DDR5 spot exposure). HBM at strong profitability; non-HBM at healthy profitability.

Vivek Arya (Bank of America): Memory Price Elasticity Outside Data Center

Arya asked at what point rising memory prices start to affect demand for consumer electronics — smartphones, PCs, traditional enterprise. Sanjay acknowledged some unit-demand impact and customer mix adjustments are already in the forecast (smartphones/PCs may see modest demand pressure, content mix may shift). The forecast already accounts for elasticity effects, and even with that, the supply gap is large. Counterpoint: AI experiences across edge devices require more memory, so content per device continues rising even if unit demand softens at the margin. The call ended on Arya's question without time for follow-ups or for Stacy Rasgon (Bernstein), who was in queue.

The Q&A had two notable absences: nobody pushed materially on the FY26 OpEx/depreciation re-base from the $20B CapEx step-up (the FY27 GM bridge), and nobody asked about non-HBM DRAM pricing reversal risk if 1-gamma ramps faster than expected at Samsung or SK Hynix. The Street's risk-frame is concentrated on HBM4 execution and customer contract specifics, not on the traditional commodity-cycle reversal vectors.

What They're NOT Saying

What management chose not to address — or addressed only obliquely — is informative.

• FY26 full-year revenue. Murphy reiterated declining a full-year revenue guide (Tim Arcuri pressed). With FQ1 actual + FQ2 guide summing to $32.3B (~$33-34B with normal upside surprise), the back half guided "to strengthen" sequentially up implies an FY26 revenue range of $70-80B+ (vs. FY25 $37.4B). The reluctance to anchor publicly on a number that would imply a doubling of FY25 revenue is a deliberate sandbag-vs.-credibility trade-off; management is preserving optionality on a number whose ranges are very wide.
• HBM dollar revenue and share. Q4 disclosed HBM revenue was nearly $2B; Q1 disclosed only "HBM revenue was a record." The customer count (six at Q4) was retired in favor of "the entire ecosystem." This is an active de-disclosure on HBM granularity. Read: with CY26 HBM fully sold out on volume AND price, granular disclosure on quarterly HBM dollars and customer mix is no longer informationally useful and is potentially destabilizing in customer negotiations.
• HBM3E vs. HBM4 mix in CY26 revenue. Sanjay said "both will contribute strong YoY growth." No share/mix split offered. Read: HBM4 yield ramp is the variable; if it is faster than expected, HBM4 mix in CY26 is higher than the current customer-platform sequencing implies. Management is preserving optionality on a number that depends on yield curves not yet fully observed.
• Specific HBM ASP direction. "HBM has strong profitability" + "non-HBM has healthy profitability" + "completed agreements on price and volume for our entire calendar 2026 HBM supply." No statement that HBM4 ASPs are higher than HBM3E. Industry expectation is that HBM4 commands a meaningful premium on the trade-ratio expansion, but management is not anchoring on that publicly. Read: pricing levels are commercially sensitive against ongoing CY27 negotiations.
• FY27 CapEx, FY27 GM bridge. Murphy said FY27 CapEx will be "up" but offered no anchor. With the construction step-up from FY25 to FY26 being "roughly double" and FY27 still up, the implied FY27 net CapEx range is $22-25B+. The FY27 GM bridge depends critically on whether the depreciation re-base (which compounds with each CapEx vintage) outpaces the mix lift to HBM4/HBM4E. No framing offered.
• Multi-year LTA volume / pricing structures. Sanjay confirmed "specific commitments" and "much stronger contract structure" but offered no mechanics. Whether these are take-or-pay, MFN-protected, capacity-floor-with-price-bands, or some hybrid is the most consequential question in the cycle. Read: management is preserving negotiating leverage; details will surface as contracts are signed.
• Tariff scenarios. Murphy explicitly excluded tariff impact from FQ2 guidance. No commentary on what scenarios have been stress-tested. Read: real but bounded macro risk that management is choosing not to quantify.

Market Reaction

Context: MU's pre-print run had already absorbed a meaningful slug of the cycle setup, with shares trading materially higher into the print on the Q4 framework + multiple sell-side upgrades through CY25. The post-print follow-through suggests the FQ2 guide and the HBM TAM pull-forward were not in the pre-print consensus model. With the FQ1 actual + FQ2 guide alone summing to $13.20 of EPS — comfortably within striking distance of the $11-13 FY26 EPS framework we set at the Q3 upgrade in June — the market is now repricing FY26 to a fundamentally different earnings base.

The relative strength of the FQ2 guide is the load-bearing surprise: $18.7B revenue is +38% sequential from a Q1 that was already a record; 68% GM is +1,120bps sequential; $8.42 EPS is +76% sequential. With management framing growth as continuing "more gradual" past FQ2, the implicit FY26 EPS run-rate is in the $28-32 range — a step that consensus had not contemplated as recently as the Q4 print. This is a model reset, not a beat-and-raise.

Street Perspective

The Street view going into the print was that the Q4-print-set FY26 framework ($45-58B revenue, 50-54% GM, $16-19 EPS) was the high-credibility band, with upside on HBM CY26 contract closure and downside on tariff or hyperscaler-capex pause. The print resolved the upside cleanly: HBM CY26 fully sold out volume AND price; the $100B HBM TAM pulled forward two years; FY26 revenue trajectory now points to $70-80B+; FY26 EPS in the $28-32 range. None of these were within the pre-print Street-consensus band.

The bull case being made on the Street: Micron has crossed from "AI-memory beneficiary" to "structurally repriced franchise," with multi-year LTAs, full CY26 HBM sell-out at locked pricing, capacity-led supply discipline, $100B HBM TAM pulled forward two years, and net cash returning. The CY26-CY27 visibility window is now multi-year, with the 1-gamma node ramp + Idaho pull-in + HBM4 yield-leadership positioning supporting a structurally higher GM band than any prior memory cycle has achieved. The terminal-multiple compression argument for memory equities — cyclicality → trough EPS volatility → multiple discount — is now actively challenged by the multi-year LTA structure under negotiation.

The bear case being made on the Street: this is still a memory cycle, and the GM at 68% is a cycle-peak number that historically reverts within 6-8 quarters. The CY26 HBM sold-out commentary doesn't anchor the larger non-HBM DRAM market, where pricing is the variable; if 1-gamma ramps cleanly at both Samsung and SK Hynix, the supply-discipline narrative can soften by mid-CY26. CapEx at ~$20B is a step-up that compounds depreciation into CY27-28; if the GM bridge falters even modestly on HBM4 ramp slip or non-HBM DRAM softness, the depreciation re-base hits harder than the cycle-peak earnings can absorb. And the multi-year LTA structure is unsigned — if a major customer balks on the new contract specificity, the visibility premium deflates.

Our read aligns with the bull case on this print, with the explicit caveat that the FY26 EPS range ($28-32) is wide because the FQ3-FQ4 trajectory depends on factors not yet observable: the HBM4 yield ramp through CQ2-CQ3 26, the 1-gamma node ramp at competitors, and the timing/specificity of the multi-year LTAs. The downside trigger most worth watching is HBM4 production ramp execution into CQ2 2026 — if that timing slips, the FY26 GM bridge tightens and the bull case loses its central anchor.

Model Implications & Forward View

Q1 FY26 actuals: Revenue $13.6B (+57% YoY), non-GAAP GM 56.8% (+24pts YoY), non-GAAP EPS $4.78 (+167% YoY), operating income $6.4B at 47% OM, FCF $3.9B (record), net cash positive.

FY26 framework (revised up from Q4 maintain):

FY27 placeholder: HBM4 ramp through CY26 + HBM4E custom-base-die (TSMC-manufactured, "higher GM than standard HBM4E") supports FY27 EPS in the $32-38 range under base-case assumptions on margin persistence and CapEx absorption. The HBM4E custom mix is the largest single FY27 lever; the Idaho fab #1 first-wafer-out in mid-CY27 is the supply-side incremental. The FY27 risk vector shifts to CapEx-driven depreciation re-base and to whether the multi-year LTAs land in time to anchor pricing through the FY27 ramp.

Thesis Scorecard: Q4 Maintain Anchors vs. Q1 Outcomes

Six of eight anchors closed favorably; one is mixed (DIO uptick is mechanical not demand-air-pocket, but worth monitoring); one open watch item (tariff broadening) is the residual macro risk. The thesis is intact and the Q1 print/guide combination materially raises the FY26 EPS range.

Risks to the Outperform Rating

• HBM4 production ramp execution. First production shipment slated for CQ2 2026. Yield ramp on a new node with internal CMOS base die is the largest single technical execution risk. Management's "yield ramp expected faster than HBM3E" framing is constructive but not yet observed at scale. A one-quarter slip would tighten the FY26 GM bridge and meaningfully soften the CY26 HBM mix.
• Hyperscaler capex pause. CY26 hyperscaler infrastructure spend is at an unprecedented level. A pause from one or more major customers would soften HBM/HC-DIMM pull and could reverse the leading-edge-DRAM tightness narrative even with HBM contracts locked.
• Non-HBM DRAM oversupply. If 1-gamma ramps faster than expected at Samsung or SK Hynix in CY26, non-HBM DRAM pricing could soften even as HBM stays tight. Non-HBM DRAM is the larger absolute revenue pool; non-HBM softness would compress consolidated GM despite HBM strength.
• Tariff regime broadening. Management explicitly excludes tariff impact from FQ2 guidance. Material expansion of tariffs in the customer-pull chain (consumer electronics, server OEMs) could compress demand even as supply remains tight.
• CapEx-driven depreciation re-base. $20B FY26 CapEx + "up" FY27 CapEx is a meaningful step-up. If the GM bridge falters (HBM4 ramp slip, non-HBM softness), the depreciation re-base could compress GM faster than mix can offset.
• Multi-year LTA execution risk. The contract restructuring is, by Sanjay's framing, a meaningful change in commercial terms. If a major customer balks on the specificity or length, the visibility premium that anchors the Outperform compresses.
• Cycle reversion risk. 68% GM is a cycle-peak number historically. The structural-tightness narrative is durable through CY26 in management's view, but DRAM/NAND remain commodity markets at the margin. A multi-quarter pricing softening in 2H CY26 or 1H CY27 would be a downgrade trigger.

Bottom Line

The Q1 FY2026 print extends and accelerates the Q4 thesis. Revenue, GM, and EPS all printed materially above guide, and the FQ2 FY26 guide of $18.7B / 68% GM / $8.42 EPS is so far above the prior trajectory that it resets the consensus model rather than incrementally raising it. CY26 HBM is fully sold out on volume AND pricing (including HBM4); the $100B HBM TAM was pulled forward two full calendar years to 2028; CapEx steps up to ~$20B with Idaho fab #1 first-wafer-out pulled into mid-CY27; FY26 FCF is guided "significantly higher" YoY with $3.9B already in the bank from FQ1 alone. The balance sheet returned to net cash with $2.7B of debt redeemed in-quarter.

We maintain Outperform and update the FY26 EPS range to $28-32 (vs. $16-19 at the Q4 maintain). The structural setup is now a CY26-CY28 visibility window rather than a quarter-by-quarter beat-and-raise narrative. Risks have moved from execution timing (HBM3E sell-out, FY26 GM bridge) to multi-year contract specificity (the LTAs), structural cycle reversion in 2H CY26 / 1H CY27 (the non-HBM pricing question), and event-specific HBM4 yield-ramp execution.