Micron Technology (MU) Q2 FY2026 Earnings Preview

• Street consensus Q2 revenue of $19.2B sits $500M above guide midpoint of $18.7B — the market is pricing in another beat of the high end.
• Street Q2 EPS of $8.59 vs. guide midpoint of $8.42 — modest delta, not a heroic assumption, consistent with recent beat pattern.
• Gross margin consensus at 68.6% vs. guide of 68% — slight outperformance expected on pricing and HBM mix.
• Q1 FY2026 was the template: revenue of $13.6B beat the high end of guide, GM expanded +11pp QoQ to 56.8%, EPS of $4.78 surged +167% YoY.
• HBM is the thesis anchor: all CY2026 supply is contracted (price and volume), HBM4 on track for Q2 CY2026 ramp with industry-leading >11 Gbps speeds.
• CapEx raised to ~$20B for FY2026 (from ~$18B) — demand is pulling investment forward, not the other way around.
• NAND improvements broaden the story: data center SSD exceeded $1B in Q1, first-to-market PCIe Gen6 SSD on shelf.
• Risk variables: tariff exposure (explicitly excluded from Q2 guidance), CapEx ramp execution, and HBM4 yield scaling.

1. Executive Summary

Micron Technology enters Q2 FY2026 reporting as the premier beneficiary of the structural shift in memory demand driven by AI infrastructure spending. The investment thesis has three reinforcing legs: (1) HBM is an oligopolistic, supply-constrained product where Micron is gaining share against SK Hynix with a competitive HBM3E architecture and an HBM4 ramp that management says will yield faster than prior generations; (2) conventional DRAM is simultaneously tight, with no segment able to satisfy demand according to management — a signal that the memory up-cycle is broader than just HBM; and (3) NAND is no longer a drag, with data center SSD revenue surpassing $1B in Q1 FY2026 and Gen6 PCIe SSD giving Micron a first-mover advantage in the highest-margin NAND segment.

The consensus backdrop is exceptional. Of 52 analysts covering MU, 47 rate it Buy and zero rate it Sell — one of the most uniformly bullish analyst skews in the semiconductor sector. Mean price target of $431.52 implies only modest upside to the current price of $426.13, suggesting the street views the stock as fairly priced rather than cheap, which places the burden of upside squarely on earnings execution. The stock has already delivered ~52% YTD performance and trades above both its 50-day ($387.87) and 200-day ($222.99) moving averages by wide margins. RSI-14 at 57.1 suggests momentum without being technically overbought.

The critical near-term debate centers on the magnitude of the forward ramp. Street models for Q3 FY26 imply revenue of $22.1B ( +15% QoQ from Q2 consensus) and Q4 FY26 at $24.6B (+11% QoQ from Q3). The implied full-year FY26 revenue of $78.2B represents 109% growth over FY2025’s $37.4B. If management guides Q3 near the Street’s $22.1B — a near-18% sequential jump from the $18.7B Q2 guide — it would validate the most aggressive interpretation of the HBM ramp. Any guide that trails that level, even if Q2 beats, is likely to compress the multiple.

The options market is pricing ~$32 move (7.5%) around the print, reflecting elevated uncertainty about guidance magnitude rather than the headline Q2 beat itself. Put/Call OI ratio of 1.34 indicates modestly defensive positioning despite the overwhelmingly bullish rating consensus, suggesting institutional hedges are being maintained into a potentially volatile print.

Core investment question: Is Micron executing an AI-driven structural re-rating of memory economics — in which case the NTM P/E of 9.8x is deeply undervalued — or is the street extrapolating a cyclical peak that will mean-revert as HBM competition intensifies and CapEx economics deteriorate? The Q2 print and, more critically, the Q3 guide will be the most important data point in resolving that debate.

2. Quarterly Performance Trend

Micron’s Q1 FY2026 print was transformational in scale and margin. Revenue of $13.6B was 57% above the prior-year quarter and 21% above Q4 FY2025 sequentially, driven by explosive HBM ramp, continued AI-driven data center DRAM demand, and improving conventional DRAM pricing. Gross margin expanded to 56.8% from 46.0% in Q4 FY2025 — an 11-percentage-point sequential improvement that reflected HBM product mix, better DRAM pricing, and cost structure benefits from 1-gamma DRAM yield maturation. EPS of $4.78 was up 167% YoY, and free cash flow of $3.9B was a company record.

The full-year FY2025 reference frame is essential context: $37.4B in revenue represented ~50% YoY growth, yet even that elevated base is now the starting point for a consensus forecast of $78.2B in FY2026 — another 109% step up. The fiscal year revenue cadence below shows the trajectory that underpins this forecast.

Note: Q1–Q4 FY2025 revenue figures are estimates based on FY2025 total of $37.4B and reported quarterly cadence. HBM revenue is management-disclosed or analyst estimates where not separately reported. Q2 FY2026 row reflects Street consensus as of March 14, 2026.

Business Unit Breakdown — Q1 FY2026 (Reported)

Key Sequential Drivers: Q4 FY2025 → Q1 FY2026

• Revenue: +21% QoQ ($11.2B → $13.6B) — HBM volume ramp + conventional DRAM pricing improvement
• Gross Margin: +11pp QoQ (46.0% → 56.8%) — HBM product mix shift, 1-gamma DRAM cost improvements, better pricing across segments
• Operating Margin: +12pp QoQ (~35% → 47%) — operating leverage on sharply higher revenue
• EPS: +58% QoQ ($3.01 → $4.78) — margin expansion and revenue growth compounding
• DRAM Revenue: +27% QoQ — $10.8B / 79% of total; DRAM up 69% YoY
• NAND Revenue: +13% QoQ — $2.7B / 20% of total; data center SSD crossing $1B milestone
• FCF: Record $3.9B — reflects high operating margin and controlled working capital
• Inventory: 126 days — modest build from prior quarter, management characterizes as healthy for demand environment

3. Full Year Estimates

Consensus forecasts for FY2026 and FY2027 reflect an unprecedented rerating of Micron’s earnings power, driven almost entirely by the HBM revenue ramp, structural DRAM pricing improvement, and operating leverage at scale. The Street is modeling FY2026 revenue of $78.2B — more than double FY2025 — and FY2026 EPS of $34.79, which represents a 226% increase over FY2025. These are extraordinary numbers by any historical standard for the memory industry and reflect how completely the AI infrastructure buildout has restructured the supply/demand balance.

FY2026 Quarterly Revenue Ramp (Consensus)

The implied sequential acceleration from Q1’s $13.6B to Q2 consensus of $19.2B (+41%) is the single largest quarterly step-up in Micron’s history. The Street is embedding roughly $5.5B of incremental revenue in a single quarter, the vast majority of which is attributable to HBM volume ramp and pricing normalization in conventional DRAM. Management’s guidance midpoint of $18.7B already implies a +37% sequential step. If management guides Q3 near the $22.1B level, it would confirm the trajectory and likely re-rate the stock; a guide materially below that level would pressure valuation even on a Q2 beat.

4. Consensus Expectations — Q2 FY2026

The guide/consensus comparison below is the most critical framework for interpreting the Q2 print. Management has now beaten the high end of its guidance range for multiple consecutive quarters, which is why the Street has anchored $500M above the midpoint. The real question is not whether Q2 beats but whether the Q3 guidance validates the aggressive ramp implied in consensus.

Forward Quarter Consensus (Key Variables for Stock Reaction)

The consensus forward arc assumes gross margin expansion of ~210bps from Q2 to Q3 and a further 90bps from Q3 to Q4. The primary driver is expected to be HBM4 revenue contribution in H2, which carries higher gross margins than HBM3E, combined with continued yield improvements on 1-gamma DRAM. If management’s Q3 guide comes in near $20.5–$21B (below the $22.1B bar), multiple compression is likely even on a Q2 beat, as investors will extrapolate a shallower FY26 trajectory and question the FY2027 $101.7B consensus. If management guides above $22.1B — which would be unprecedented for management conservatism — the stock likely re-rates meaningfully higher.

5. Analyst Views Prior to Earnings

The analyst community is as uniformly bullish on MU as it has been on any major semiconductor name in recent memory. With 47 Buy ratings, 5 Hold, and zero Sell ratings out of 52 analysts, the absence of skepticism is itself a data point. Mean price target of $431.52 implies only ~1% upside from the current price of $426.13, while median PT of $450 implies ~6% upside. The clustering of price targets near current levels suggests analysts view the stock as fairly priced for the cycle they can underwrite rather than cheap, which means execution on the HBM4 ramp and forward guidance are required to generate further target increases.

Bull Thesis (Consensus)

• HBM market leadership: Micron has closed the gap with SK Hynix on HBM3E, all CY2026 supply is contracted, and HBM4 is expected to ramp in Q2 CY2026 with superior bandwidth (>11 Gbps). Long-term HBM TAM of $100B by 2028 (CAGR ~40%) is a structurally new revenue pool for what was historically a commodity business.
• Structural supply tightness: Management said no segment of DRAM can meet current demand. This is not a one-product tightness story — it is a system-wide supply/demand imbalance driven by AI CapEx, on-device AI, and continued cloud buildout.
• Multi-year AI CapEx cycle: Hyperscalers have committed to multi-year infrastructure investment plans. NVIDIA’s Blackwell and Rubin platforms both increase HBM content per GPU, expanding Micron’s addressable opportunity per unit as the installed base scales.
• Valuation discount to earnings power: At 9.8x NTM P/E, MU trades well below historical mid-cycle multiples, and the argument is that HBM structurally raises the earnings floor through the next cycle — reducing the traditional trough-to-peak amplitude that has historically justified peak multiple discounts.

Bear Concern (Minority View)

• Cycle peak valuation risk: Memory investors are conditioned to sell into peak margins. Even if HBM changes the cycle dynamics, multiple expansion from current levels requires fresh conviction on FY27+ earnings sustainability.
• CapEx ramp execution: $20B of FY2026 CapEx is a major commitment. Equipment delivery, fab construction timelines, and simultaneous HBM4 yield ramp across multiple facilities creates execution risk that could compress FCF below expectations.
• Tariff exposure: Excluded from guidance explicitly. Any China-specific semiconductor tariff action or broad Section 232/301 escalation would be material and unmodeled.

6. Strategic News Flow (Last 120 Days)

Since the Q1 FY2026 earnings call on December 17, 2025, Micron has communicated a series of strategic updates that collectively reinforce the bull thesis while introducing some execution risk around timing and CapEx scale. The key themes: HBM4 ramp confirmation, supply contract coverage, and fab investment decisions.

• HBM4 ramp on schedule: Management confirmed HBM4 is on track for Q2 CY2026 volume ramp, featuring industry-leading bandwidth of >11 Gbps — a meaningful performance advantage over competing products. HBM4 is critical because it carries higher ASPs and likely better gross margins than HBM3E, supporting the H2 FY2026 margin expansion the Street models.
• CY2026 HBM fully contracted: All CY2026 HBM supply — price and volume — has been contracted with customers. This is an unusual level of visibility for a semiconductor product and effectively de-risks the HBM revenue line for the next 12+ months. The contracts include HBM4 production, which begins ramping in Q2 CY2026.
• HBM TAM expansion: Management revised the HBM TAM CAGR estimate to approximately 40% through CY2028, with the TAM reaching $100B by 2028 — two years earlier than the prior outlook. This accelerated TAM expansion reflects faster adoption of HBM-intensive GPU architectures across both training and inference infrastructure.
• CapEx raised to ~$20B for FY2026: FY2026 CapEx was raised from the initial ~$18B framework to approximately $20B, weighted to H2 FY2026. The increase is demand-driven — specifically, HBM assembly/test capacity and 1-gamma DRAM conversion investment — rather than speculative capacity addition. Management characterized the increase as a response to customer demand pull, not a supply push.
• 1-gamma DRAM yield maturation: Micron’s 1-gamma DRAM node achieved mature yields in record time — 50% faster than the previous generation. This is critical for cost structure in conventional DRAM and for HBM cost competitiveness, as HBM4 will be built on 1-gamma.
• Data center SSD >$1B milestone: The data center SSD business exceeded $1B in revenue for the first time in Q1 FY2026, and Micron was first to market with a PCIe Gen6 SSD. QLC NAND mix reached a record high, improving cost economics. This validates the NAND turnaround narrative that had been a drag on the investment thesis for prior quarters.
• Supply-demand tightness extending beyond CY2026: Management explicitly stated that tight supply/demand conditions in DRAM are expected to persist beyond CY2026. This comment is significant because it extends the earnings visibility horizon and reduces the probability of a sharp cyclical downturn in the FY2027 time frame.
• LPDDR6 sampling: LPDDR6 products are being sampled to leading OEMs. LPDDR5X 24GB configurations have also been sampled. These products target on-device AI use cases in premium smartphones, a segment that drives both DRAM ASP improvement and content growth per device.
• Singapore HBM facility: Singapore HBM assembly/test facility is on track for its first meaningful contribution in CY2027. This facility diversifies Micron’s geographic manufacturing footprint and adds capacity for the post-FY2026 HBM demand cycle.
• Idaho new fab: First wafer output from the new Idaho fab is expected in H2 CY2027. This is a multi-year CapEx investment that supports the long-term capacity roadmap without impacting near-term supply or cost structure.
• Multi-year customer contracts in discussion: Management confirmed they are in discussions for multi-year contracts with specific price and volume commitments. No signed agreements have been disclosed, but the structure would significantly enhance earnings visibility and margin predictability beyond the current contracted CY2026 HBM book.
• Tariff exclusion from guidance: Management explicitly excluded any tariff impact from Q2 FY2026 guidance. This is a notable hedge — it signals internal uncertainty about tariff outcomes, and any negative development would require either a mid-quarter update or a Q2 guide haircut at the print.

7. Analyst Questions and Concerns from Q1 FY2026 Earnings Call

The December 17, 2025 Q1 FY2026 earnings call and subsequent post-earnings analyst call surfaced several themes that will dominate Q2 follow-up. Key analyst names, the substance of their questions, and why each matters for the investment thesis are catalogued below.

• Gross margin trajectory to 70%+ — Vivek Arya (BofA): Arya pressed management on the path from the 68% Q2 guide to the 70–75% range that many analysts model for late FY2026/FY2027. CFO Mark Murphy responded that favorable factors — AI-driven demand, Micron’s technology leadership position, and structural supply constraints — support further gross margin expansion beyond Q2’s guided level. He declined to give specific quarterly targets beyond Q2 but characterized the direction as clearly expansionary. Why it matters: The Street already models 70.7% gross margin in Q3 FY2026 and 71.6% in Q4. If management implies the path to 70%+ is more gradual or conditional on HBM4 timing, numbers will come down. Any reaffirmation of the directional trajectory is a positive data point.
• HBM capacity allocation vs. conventional DRAM — Vijay Rakesh (Mizuho): Rakesh asked how Micron is balancing HBM wafer allocation against conventional server DRAM given the supply/demand tension in both markets. President Sanjay Mehrotra (and Chief Business Officer Sumit Sadana) responded that all DRAM segments are short — the company cannot meet demand anywhere across HBM, non-HBM server DRAM, client DRAM, or mobile DRAM. Why it matters: Conventional DRAM profitability has historically been margin-dilutive vs. HBM. If both segments are oversubscribed, pricing discipline across the conventional book becomes an earnings tailwind independent of the HBM story, broadening the margin expansion thesis.
• HBM4 yield ramp and qualification cadence — Multiple analysts: Multiple sell-side analysts pressed for specificity on the HBM4 yield ramp timeline and customer qualification schedule. Management stated that HBM4 yield ramp is expected to proceed faster than the HBM3E 12-high ramp, citing manufacturing process improvements and learning curve advantages on the 1-gamma node. Why it matters: HBM4 revenue contribution timing directly drives H2 FY2026 EPS estimates. A delay of even one quarter in HBM4 volume qualification would meaningfully impact the Q3 and Q4 revenue ramp that consensus has embedded.
• CapEx trajectory and FCF implications — Aaron Rakers (Wells Fargo): Rakers asked about CapEx phasing within FY2026 and the OpEx trajectory. Murphy guided Q2 OpEx flat vs. Q1 (~$1.38B), Q3 flat, and Q4 up (primarily driven by an extra fiscal week in Q4). On CapEx, management confirmed the ~$20B framework for FY2026 is weighted to H2, driven by HBM assembly/test equipment deliveries and 1-gamma conversion investment. Why it matters: Front-loaded CapEx would compress near-term FCF. H2 weighting is more favorable for FCF in H1 FY2026, but H2 CapEx intensity will be substantial and should be monitored for any increase beyond the $20B guide.
• NAND/SSD momentum and QLC mix — Multiple analysts: Analysts noted the SSD business crossing $1B and asked about sustainability and the competitive advantage from Gen6 PCIe SSD leadership. Management highlighted QLC mix at record highs (improving unit economics), first-to-market position on Gen6 SSD, and growing share in data center storage. They characterized NAND supply/demand as improving but not as tight as DRAM. Why it matters: NAND has historically been the margin drag in memory cycles. Structural improvement in NAND — particularly in high-value data center SSD vs. commodity NAND — broadens the profitability story and reduces downside risk to blended gross margin if DRAM pricing softens.
• Multi-year customer contract structure — Multiple analysts: Analysts pushed on when discussions for multi-year contracts with specific commitments would convert to signed agreements. Management confirmed active discussions but offered no timeline for announcement. They described the contract structure as covering both price and volume commitments, which would be a significant departure from the historically spot-market-driven memory pricing dynamic. Why it matters: Signed multi-year contracts would meaningfully de-risk the FY2027 revenue and margin outlook, which is currently the primary source of earnings uncertainty. Any concrete announcement would be a major positive catalyst.

8. What to Listen For at Earnings

The Q2 FY2026 print has a specific hierarchy of importance: Q3 guidance matters most, followed by HBM4 timing updates, gross margin bridge commentary, and tariff/contract developments. A Q2 beat with weak forward guidance would be a negative outcome; a Q2 in-line or slight miss with a strong Q3 guide could still be a positive outcome. Investors should listen for the following in order of importance:

• Q3 FY2026 guidance — the primary event: Street is at $22.1B revenue, $10.50 EPS, 70.7% GM for Q3. Management must guide near this level to validate the full-year consensus of $78.2B. The sequential ramp from Q2’s $18.7B guide to $22.1B implied is ~18% — an extraordinary quarter-over-quarter step that requires HBM4 to be contributing material revenue by Q3, not just sampling. If management guides $20–$21B, the $78.2B FY26 revenue consensus becomes very difficult to achieve and estimates will need to come down, compressing the multiple. Any guide above $22.1B would be a significant upside catalyst.
• HBM4 revenue contribution timing: All CY2026 HBM supply is contracted, but the quarterly revenue phasing of HBM4 vs. HBM3E mix will determine H2 gross margin. HBM4 carries higher bandwidth, higher ASPs, and likely superior margins. Management has guided Q2 CY2026 (Apr–Jun 2026, i.e., Micron’s Q3 FY2026) as the initial volume ramp. Confirmation of customer qualification progress, yield improvement benchmarks vs. HBM3E, and early signs of revenue contribution in the Q2 print would be incrementally positive.
• Gross margin bridge to 70%+: Management guided 68% for Q2 and Street models 70.7% by Q3. The ~270bps bridge in a single quarter requires a specific explanation: Is it entirely HBM4 mix? Is it conventional DRAM pricing expansion? Is it 1-gamma yield-driven cost reduction? Management’s quantitative breakdown of the GM expansion path will determine whether analysts can underwrite 70%+ with confidence or whether consensus needs to be trimmed.
• Tariff update: Q2 guidance excluded tariffs entirely. Analysts will press on exposure specificity — how much of COGS is tariff-exposed, what mitigation actions have been taken (customer pass-throughs, supply chain restructuring), and whether any customer behavior changes are visible in the demand picture. A concrete framework for tariff quantification would reduce a key overhang.
• Multi-year contract progress: Has management signed any multi-year agreements with specific price and volume commitments since the Q1 call? Any announcement would be a major positive signal and would directly support the FY2027 earnings visibility argument. If still “in discussions,” the lack of progress after three months warrants a follow-up on what the key obstacles are.
• CapEx update: Is the $20B FY2026 framework intact, or has demand pull-forward created pressure to raise it further? Where are the marginal dollars going — HBM assembly/test (demand-driven), 1-gamma conversion (yield-driven), or greenfield DRAM fab (speculative)? Any CapEx increase beyond $20B would raise FCF concerns but could also signal further demand visibility.
• CY2026 supply-demand update: Management’s prior language was that supply tightness “extends beyond CY2026.” Any update to that characterization — tighter, looser, or time-bounded — would be an important data point for the FY2027 margin sustainability debate. Specifically, listen for commentary on DRAM bit growth outlook (guided ~20% for CY2026) and whether supply is expected to remain below demand through CY2026.
• Customer diversification and HBM base: Management disclosed HBM customer base expanded to six customers as of Q1. Any expansion of that number, qualification of new architectures, or movement from sampling to volume with additional customers would signal a deepening of the demand base beyond the top two hyperscalers.

9. Management Guidance (Last 120 Days)

All explicit Q2 FY2026 guidance was provided on the December 17, 2025 Q1 FY2026 earnings call. Management did not provide an investor day or formal mid-quarter update between December 2025 and the Q2 report, so the guidance framework below is the most recent official company communication. The tariff exclusion caveat is particularly important for the gross margin and EPS lines.

Explicit Q2 FY2026 Guidance (Provided December 17, 2025)

Full Year FY2026 Framework (Non-Guidance Language from Dec 17 Call)

10. Catalysts to Watch — Next 90 Days

The 90-day catalyst window is unusually dense given the combination of an imminent earnings print, a major product ramp (HBM4), and a macro/policy backdrop that includes active tariff policy uncertainty. The hierarchy below reflects investment relevance rather than probability.

11. Key Risks and Considerations

The risk profile for MU at current levels is asymmetric toward execution risk rather than fundamental demand risk. The demand environment is strong, the supply is oversubscribed, and management has under-promised and over-delivered for multiple consecutive quarters. The risks below are not near-term demand risks; they are valuation, execution, and policy risks that could impair the multiple and forward estimate trajectory.

• Valuation at cycle peak debate: MEDIUM
  NTM P/E of 9.8x looks superficially cheap relative to the S&P 500, but memory investors are historically conditioned to sell peak margin cycles as the mean-reversion vehicle approaches. The counter-thesis — that HBM structurally raises Micron’s earnings floor by converting a commodity product into a high-value, supply-constrained differentiated product — is intellectually sound but unproven across a full cycle. The multiple will not expand unless investors gain conviction that FY2027’s $46.10 EPS is achievable and sustainable, which requires both HBM4 execution and continued conventional DRAM pricing discipline.
• Tariff exposure: HIGH
  Management’s explicit exclusion of tariff impacts from Q2 guidance is a significant hedge and a disclosure risk. China accounts for a meaningful portion of Micron’s end-market demand, and any China-specific semiconductor tariff escalation or retaliatory action — in either direction — would directly impact revenue and margin. Section 232 national security tariffs on semiconductors, broader reciprocal tariff structures, or any restriction on DRAM sales to Chinese entities would require a guidance revision. This risk is unmodeled and binary in nature.
• CapEx execution risk: MEDIUM
  Committing ~$20B of CapEx in a single fiscal year is an extraordinary capital allocation event. The H2 FY2026 weighting implies a significant step-up in equipment deliveries and construction activity over a compressed timeframe. Potential failure modes include: equipment delivery delays from ASML/Applied Materials/Lam Research, fab construction timeline slippage in Idaho, HBM4 assembly/test equipment qualification issues, and simultaneous yield ramp on multiple process nodes. Any significant CapEx overshoot or FCF miss would impair investor confidence in the capital discipline narrative.
• HBM competitive dynamics: MEDIUM
  SK Hynix and Samsung are investing aggressively in HBM4 capacity. SK Hynix is currently the market leader by volume and ASP, with an established hyperscaler qualification base that includes NVIDIA. If SK Hynix ramps HBM4 faster than expected, or if Samsung qualifies HBM4 with NVIDIA before Micron, the pricing premium and volume allocation advantage Micron is currently projecting could erode. Micron’s >11 Gbps HBM4 bandwidth claim is a competitive differentiator, but the history of memory technology differentiation suggests ASP parity is typically the endpoint, not sustained premium pricing.
• AI CapEx demand sustainability: MEDIUM
  The entire FY2026/FY2027 revenue and earnings trajectory is predicated on continued hyperscaler AI infrastructure investment at or above current rates. The five largest hyperscalers have communicated multi-year CapEx commitments totaling hundreds of billions of dollars, but those commitments are not binding contracts and have historically been subject to revision when ROI visibility weakens. Any material softening in AWS, Microsoft Azure, Google, or Meta data center CapEx guidance would flow directly through to HBM demand, DRAM pricing, and Micron’s forward estimates.
• Conventional DRAM pricing sustainability: MEDIUM
  Current conventional DRAM pricing is at an extraordinary level, driven by supply/demand imbalance. Management’s claim that all segments — not just HBM — are supply-short is the strongest possible pricing environment. The key question is structural vs. cyclical: if the AI CapEx cycle plateaus or if DRAM supply additions from the industry’s CapEx plans begin to normalize the market in CY2027, conventional DRAM pricing could correct sharply, compressing margins on the non-HBM portion of the business (which remains the majority of DRAM bit volume).
• Customer concentration risk: LOW-MED
  As HBM revenue scales, the top hyperscalers — NVIDIA customers purchasing HBM-equipped GPUs, directly or through system integrators — represent an increasing share of Micron’s total revenue. Micron disclosed HBM customers expanded to six, but the top two likely represent the overwhelming majority of HBM revenue. Customer concentration reduces pricing negotiating power over time and creates single-customer risk if any major AI infrastructure buildout decelerates or shifts technology architecture away from HBM-intensive designs.
• HBM4 yield and qualification timeline risk: MEDIUM
  Management’s confidence in HBM4 yield ramp outpacing HBM3E is based on process improvements and 1-gamma node maturity. However, HBM4 is a significantly more complex product than any prior HBM generation in terms of I/O density and bandwidth requirements. Any yield shortfall or customer qualification delay of even one quarter shifts HBM4 revenue from Q3 FY2026 to Q4 FY2026, directly impairing the Q3 guidance trajectory that the Street is pricing in.