The Framework Update Landed, and It Was Bigger Than the Quarter: WFE to Low $150B, Long-Term Model to Mid-40s Operating Margin, September Guided $1B Above the Street

Results vs. Consensus

Lam guides four lines each quarter and beat all four, three of them above the top of the range. The company's own characterization was unusually precise about which was which.

"Our revenue was above the midpoint of guidance. With gross margin, operating margin and earnings per share all exceeded the high end of our guided range." — Douglas Bettinger, CFO

Guide vs. Actual

Actual vs. Street

The revenue beat against the Street was modest at 0.9%. The EPS beat at 7.7% was substantially larger, which tells you the quarter was won on margin rather than on volume. And the September guide beat, at roughly a billion dollars, was larger in dollar terms than the entire year-over-year revenue increase Lam posted in fiscal 2023. That is where the 18% move came from.

Quarter-over-Quarter Comparison

Year-over-Year Comparison

Fiscal Year 2026 Summary

All income-statement, balance-sheet and cash-flow figures above are as reported in the Item 2.02 Exhibit 99.1 furnished July 29, 2026. FY2026 revenue of $23,232.7M equals the sum of the four reported quarters ($10,669.0M first half plus $5,841.5M March plus $6,722.2M June). Segment and regional dollar figures elsewhere in this report are implied from disclosed percentages applied to reported revenue totals and are labeled as such; the company discloses those splits in percentage terms only.

Revenue Assessment

June revenue of $6.722B is the fourth consecutive quarterly record and the twelfth consecutive quarter of sequential growth. The 15.1% sequential and 30.0% year-over-year rates are both accelerating from March's 9.4% and 24%, which is the more important observation: this is not a business at a plateau being held up by a strong comparison, it is a business whose growth rate is still rising three years into the up-cycle.

The composition of the sequential increase is what changes the analytical picture. Of the $880.8M added, systems contributed $519.3M and CSBG $361.5M. Within systems, the swing was almost entirely NAND: non-volatile memory went from 12% to 23% of systems revenue, implying roughly $978M against $448M, an increase of about $530M that by itself exceeds the total sequential systems increase. Foundry fell to 44% from 54%, implying about $1,870M against $2,015M, as mature-node spending by domestic Chinese customers declined and was largely offset by leading-edge 2nm and 3nm investment plus advanced packaging. DRAM held roughly flat in dollars at an implied $978M while falling to 23% of the mix from 27%, which is a mix effect rather than a demand signal.

For the full fiscal year, revenue of $23.233B was up 26.0%, landing slightly above the roughly $23.1B trajectory we modeled in April. Lam's calendar-2026 revenue is tracking toward roughly $29B against roughly $20.6B in calendar 2025, growth of about 42% against WFE growth of about 36% from $110B to the low $150B range. That is the third consecutive year of outgrowing WFE, which management stated as an expectation rather than a hope, and it is the single cleanest validation of the SAM-expansion pillar.

Margin Assessment

Non-GAAP gross margin of 52.0% is the highest in two decades and 150bps above the guide midpoint. Management attributed it to three factors without ranking them: pricing actions, operational and scale efficiencies, and favorable product mix. The order matters, because pricing was listed first and was named explicitly, which is a departure from prior quarters where the emphasis fell on manufacturing footprint and tool maturity. Under direct questioning twice, management declined to characterize like-for-like pricing, which leaves the durability of that particular contributor unresolved.

Operating margin of 38.4% is a record and 190bps above the top of guide. The leverage math is stark: operating expenses rose $50.3M sequentially against a revenue increase of $880.8M, an incremental operating margin of 60.6%. Some of that is unsustainable, since the expense line includes variable compensation that scales with profitability and headcount that was still being added through the quarter (up 1,800 people, or 8.7%, to roughly 22,400). The September guide normalizes this to a 44.9% incremental operating margin, which is the number to carry forward.

The framework revision reframes both. The prior long-term model, set at the 2025 Investor Day, was above 50% gross margin and above 35% operating margin. Lam is now printing 52.0% and 38.4% and guiding 52.0% and 39.5%, so the old model was not merely reached but comprehensively passed. The new target is mid-50% gross margin and mid-40% operating margin over the next several years. In April we forecast that this update would land near 51% to 52% gross margin and 37% operating margin. Management came in roughly 300bps higher on gross margin and roughly 800bps higher on operating margin than we projected, and we were not the low estimate on the Street.

EPS and Cash Flow Assessment

Non-GAAP EPS of $1.82 beat the guide high end by $0.02 and the Street by $0.13. Fiscal-year non-GAAP EPS of $5.82 was up 41%, against GAAP diluted EPS of $5.76 up 38.8%. Our April model called for FY26 non-GAAP EPS of $5.85 to $6.10, so the year finished marginally below our range despite revenue landing slightly above it. The gap traces to the first half, not to this quarter, and it is a useful reminder that our margin ramp assumptions were front-loaded relative to how the year actually unfolded.

Cash conversion is the one line in this print that got materially worse, and it received one analyst question about inventory and none about receivables. Free cash flow for fiscal 2026 was $4.891B, down 9.7%, while net income rose 35.6%. Conversion fell from 101% to 67%. The driver is working capital: the change in operating assets and liabilities was a $1.914B use of cash in FY26 against a $442M source in FY25, a $2.36B swing. Receivables ended the year at $5.340B, up 58.1% year-over-year against revenue up 30.0%, pushing days sales outstanding to 72 from 59.

This is a growth signal rather than a quality-of-earnings signal, and the distinction matters. Revenue accelerating 15% sequentially with tools shipping late in the quarter mechanically inflates receivables, and inventory turns actually improved to 3.0x, the best in nearly five years, so the balance sheet is not accumulating unsold product. But it does mean reported earnings are running ahead of cash, and with a stated policy of returning at least 85% of free cash flow, a lower free cash flow base is a lower capital-return base. In fiscal 2026 Lam returned $5.122B against $4.891B of free cash flow, or 105%, funded partly by drawing cash down from $6.391B to $5.579B while also retiring $755M of debt.

The quarter's capital return tells its own story. Buybacks fell 79% to $247M from $1,163M in March, taking total return to 45% of free cash flow from 184% (both measured on the cash-flow-statement basis, where repurchases include excise tax). Management framed this as building balance-sheet capacity for capital spending and working capital, which is accurate. It is also, in hindsight, well timed: the March quarter's repurchases averaged $211 per share, the stock traded above $400 through much of June, and Lam conspicuously did not chase it. Whether by design or by working-capital necessity, the effect was to preserve $4B of remaining authorization for a stock now 31.3% below its 52-week closing high.

Segment Performance

NAND: The Quarter's Engine, Doubling Sequentially

Non-volatile memory went from 12% to 23% of systems revenue, an implied move from roughly $448M to roughly $978M. This is the single largest sequential segment swing in the four quarters we have covered, and it is the delivery of the pull-forward management telegraphed in April, when it said the majority of the $40B installed-base conversion program would occur before the end of calendar 2027.

"Sequential top line growth was led by a doubling of NAND revenue from the prior quarter underscoring the growing importance of storage to AI system performance." — Timothy Archer, CEO

The technical driver disclosed this quarter is more specific than April's framing. Conversions are targeting 256-layer-and-above devices, and the end application named is enterprise SSDs rather than the broader AI-data-center-bits argument. The demand mechanism the CEO gave is expanding context windows and persistent memory requirements in AI inference workloads, which is a more durable claim than a general bit-growth argument because it ties NAND consumption to a specific and growing model-architecture requirement.

The forward framing extends further than the current conversion wave. Management now states that Lam's served available market per wafer in NAND doubles from the 128-layer node to 500-plus-layer devices, a function of longer process times for taller stacks plus additional tool count to manage stacking complexity. Pressed on whether the $40B program is a one-time opportunity, the CEO confirmed it is not, and that the next conversion cycle from 200 to 300 and 400-plus layers will run against a larger installed base because greenfield additions are happening concurrently. He also volunteered that the second wave has not been quantified.

Assessment: NAND is now the highest-conviction near-term driver and delivered exactly the acceleration we modeled in April, faster. The risk that has appeared in its place is repeatability: a segment that doubles sequentially cannot double again, and a 23% systems mix is at the high end of the historical range. Management guided 2027 growth order as DRAM first, foundry/logic second, NAND third, which implies NAND decelerates from here even as it grows. We model NAND at roughly $4.0B to $4.5B in FY27 systems revenue.

DRAM: Flat Dollars, but the Franchise Broadened Beyond HBM

DRAM fell to 23% of systems from 27%, with dollars roughly flat at an implied $978M against $1,007M. Read alone that is a pause. Read alongside the prepared remarks it is more interesting, because this is the quarter where Lam's DRAM story stopped being a high-bandwidth-memory story.

Three specific conventional-DRAM disclosures landed. Customers are adopting the VECTOR hard mask deposition platform for low-k film patterning, with management claiming co-optimization of hard mask film properties with Lam's conductor etch delivers better transistor performance and yield plus over 20% cost savings against traditional approaches. VECTOR diffusion barrier systems are being adopted for next-generation requirements, with a claimed 5% capacitance reduction versus competing technologies. And the Akara conductor etch platform, first adopted for 2nm-and-below gate-all-around in foundry/logic, is now gaining momentum in advanced DRAM with several strategic tool-of-record positions including the most challenging gate etch applications.

"we are winning in conventional DRAM, but it's related to the strength of the device, the patterning of the device, the materials that are being introduced and I think that continues. And then you layer on top of that, if you do HBM, it's even better for us." — Timothy Archer, CEO

The underlying argument is that DRAM performance scaling is converging on leading-edge foundry/logic requirements: more EUV layers pulling in patterning etch, RC performance concerns pulling in diffusion barriers and etch stops, and vertical scaling pulling in high-aspect-ratio capability. Lam already holds strong positions in each of those categories in foundry/logic and is porting them across.

Assessment: The flat dollar quarter obscures a genuine widening of the DRAM opportunity from a TSV-and-electroplating HBM position into the conventional front end. Management named DRAM as the fastest-growing segment for 2027. Combined with the disclosed capacity constraints in conventional DRAM, we treat this pillar as intact and broadening rather than stalling, and model DRAM at roughly $4.5B to $5.5B in FY27 systems revenue.

Foundry: Down on China Mature Node, Offset at the Leading Edge

Foundry fell to 44% of systems from 54%, an implied decline from roughly $2,015M to roughly $1,870M. Management gave a clean attribution: mature-node spending with Chinese customers declined sequentially, largely offset by strength in leading-edge 2nm and 3nm capability plus advanced packaging. The offset was not complete, hence the 7% implied dollar decline.

Taiwan reaching a record dollar level at 27% of total revenue, up from 23%, corroborates the leading-edge half of that attribution. Asked directly about foundry share gains at 3nm and 2nm, the CEO declined specifics on customer-concentration grounds but pointed to the categories: low-k spacers, patterning etch as features become smaller and taller under EUV, and metallization resistance and dielectric capacitance work where the new ALD tools and Akara are placed. The Argos selective etch system was framed against a doubling of surface-treatment applications in the FinFET-to-gate-all-around transition, winning at 2nm and below.

Advanced packaging sits inside foundry and was raised for the third consecutive quarter, now above 70% year-over-year growth against above 50% in April and above 40% in January. The forward architecture disclosure was the most concrete Lam has given: future AI packages exceeding 9x reticle size, roughly 3x larger than today's mainstream designs, driving the industry toward panel-level approaches. Lam has shipped 510x515mm panel systems into development programs across multiple geographies and will ship its first 310x310mm panel tool this year.

Assessment: The sequential foundry decline is mix noise inside a segment where the leading-edge and advanced-packaging components are both accelerating and the China mature-node component is deliberately shrinking. That is the right composition change even though it costs revenue this quarter. Advanced packaging raised three times in three quarters is not a forecasting process converging on truth, it is a business inflecting faster than management can model, and it is the cleanest evidence in the print that Lam's SAM is expanding rather than merely its share. We model foundry at roughly $8.5B to $9.5B in FY27 systems revenue.

China: Down to 26% as Guided, and the Composition Improved

China fell to 26% of total revenue from 34%, an implied decline of about 12% in dollars. Management guided precisely this in April, so the decline itself is delivery rather than surprise. The more useful disclosure is the split within it.

"I would mention that within China, the global multinational customers grew sequentially, while the domestic customer base declined." — Douglas Bettinger, CFO

Management expects that pattern to repeat through the latter part of the year and maintains that China WFE overall is flat to slightly up, with quarter-to-quarter lumpiness. The composition point matters for the bear case: the portion of China revenue exposed to export-control escalation and to domestic-champion substitution is the indigenous customer base, and that is the part that shrank.

Assessment: China concentration risk continues to resolve favorably and faster than we modeled, and it did so while gross margin expanded 210bps, which retires the old argument that China mix was propping up margins. The residual risk is no longer concentration but substitution, and it sits outside this print: China began mass-producing domestic immersion DUV lithography systems in July. That is ASML's category rather than Lam's, so the direct read-through is limited, but it raises the probability that domestication broadens into etch and deposition over a multi-year horizon. We model China at 22% to 26% of revenue through FY27.

CSBG: Third Consecutive Record on Record Upgrade Revenue

The Customer Support Business Group posted $2.472B, up 17.1% sequentially and 42.6% year-over-year, its third consecutive record and now 36.8% of total revenue. This materially exceeded the informal guidance given in April, when management said CSBG would run roughly at the $2.1B level through the remaining calendar-year quarters, "maybe up a little bit." It came in 17% higher instead.

The composition explains both the beat and its repeatability. The sequential increase was primarily record upgrade revenue, with smaller increases in Reliant and services, while spare-part purchases stayed consistent with the March quarter's strong level. Upgrades are the component directly geared to the NAND conversion wave, so CSBG's beat and the systems NAND doubling are the same event measured twice. That is a coherent story, and it also means CSBG's growth rate is now tied to the conversion cycle rather than to the utilization-driven spares baseline that management described as unable to go higher.

The Equipment Intelligence and Dextro layer continues to build underneath. Management has doubled the number of preventative-maintenance tasks that can be automated since the start of 2026, and the solutions initially proven in NAND are now expanding into DRAM, which management explicitly tied to additional service revenue in the second half. Asked about the September quarter, management declined specific numbers but said to expect a similar profile with upgrades continuing to be strong.

Assessment: CSBG at a $9.9B annualized run rate is now larger than Lam's entire company revenue was in fiscal 2017. The quality caveat is that the upgrade component is cyclical in a way the spares and services base is not: upgrades are capex by another name, purchased out of the same customer budget as systems. The recurring floor is the spares and services base, not the whole $2.5B. We model CSBG at $9.5B to $10.5B in FY27.

Key Topics & Management Commentary

Overall Management Tone: Confident and, for the first time in this coverage arc, willing to reset the ceiling rather than defend it, with the framework revision delivered as a scripted closing item rather than in response to questioning. The posture on 2027 was the mirror image: management used superlatives freely while refusing quantification through four separate attempts, and the repeated fallback to "too soon" against a demand backdrop it described as extraordinary was the one place the call was less convincing than it needed to be. Pricing drew the same treatment, named as a gross margin driver in the prepared remarks and then declined twice under direct questioning.

1. The Long-Term Framework Revision: Mid-50s Gross Margin, Mid-40s Operating Margin

This is the most consequential disclosure of the print and the delivery of the commitment management made in April, when the CFO acknowledged the company was above its stated model and promised an update "later in the year." It arrived as the final scripted item, not as a response to a question.

"Now within this framework, we intend to drive gross margins to the mid-50% level and operating margins to the mid-40% level over the next several years as the AI transformation drives continued greater investment in global semiconductor capacity." — Douglas Bettinger, CFO

The prior framework, set at the 2025 Investor Day, was above 50% gross margin and above 35% operating margin. Against a September guide of 52.0% and 39.5%, the new target implies roughly another 300bps of gross margin and roughly 550bps of operating margin still to come. On the September guide's revenue base, 550bps of operating margin is about $445M per quarter of incremental operating income, or roughly $1.42 per share annualized before tax effects.

Two constraints were placed on it immediately. Asked how long it takes, the CFO said several years and described the drivers as a mix of scale, new product introduction and pricing for value. Asked separately whether 50%-plus gross margin is the right baseline at current volumes, he said Lam is in the 51% to 52% range and can continue delivering that in the near term. So the mid-50s is a multi-year destination, not a next-year number, and the near-term modeling anchor stays at 52%.

Assessment: We predicted this update would land at 51% to 52% gross margin and 37% operating margin. It landed roughly 300bps and 800bps higher. That is a large forecasting miss on our part in the company's favor, and it is the primary reason our FY27 and FY28 earnings estimates move up by roughly a third. The absence of a timeline is the honest caveat: "several years" spanning three to five years produces materially different discounted values, and management chose not to narrow it.

2. Calendar 2026 WFE Raised to the Low $150B Range

"As we enter the second half of the year, we expect calendar 2026 wafer fab equipment spending or WFE to be in the low $150 billion range up from our prior outlook of $140 billion with upside bias." — Timothy Archer, CEO

This is the second upward revision inside calendar 2026, following January's $135B and April's $140B-with-upside-bias. The CEO clarified under questioning that the upside bias attached to the April number is what produced the $150B figure, and that no further upside bias attaches to the new outlook. That clarification is worth registering: the raise consumed the previously flagged optionality rather than adding new optionality on top.

The mechanism deserves attention because it directly addresses the constraint management itself had identified as the binding limit on 2026.

"People find ways and the demand is very strong. And so people have squeezed out a little bit of extra space. They've resolved bottleneck tools." — Timothy Archer, CEO

The CEO added that Lam works with customers to resolve throughput bottlenecks where its own tools are the constraint, which sometimes frees customers to spend on other bottleneck tools. He also cautioned that further in-year upside is now harder to find, and that the discussions have shifted to 2027 and beyond as new fabs open.

Assessment: A $40B increase in calendar-2026 WFE across three revisions, from $110B in 2025 to the low $150B range, is a 36% industry growth rate. The clean-room constraint that framed the bear case for two quarters turned out to be elastic under sufficient demand pressure, which is informative about 2027: the constraint is real but it bends. The offsetting caution is that squeezing existing space is a one-time lever, and management said as much.

3. Calendar 2027: Superlatives Without a Number

"Looking into 2027, we see an extraordinary setup for WFE growth. AI is driving record revenue and profitability for our customers who have signaled unprecedented long-term demand visibility. They've also announced multiyear time lines for new fab projects and are working with us to secure equipment orders to fill the incremental clean room space as it comes online." — Timothy Archer, CEO

Four separate analysts pressed for quantification and none obtained it. The CFO's substantive contributions were that the industry is meaningfully undersupplied this year and that condition rolls into next year; that there are probably 8, 9 or 10 new fabs coming online between now and the end of next year at the larger customers; and that he feels incrementally good about each successive quarter. On the segment ordering, management said 2027 growth is led by DRAM, then foundry/logic, then NAND, which is the same ordering as 2026.

Management also updated a framework it introduced last year linking data-center capex to WFE, moving the ratio from roughly $8B of WFE per $100B of data-center capex to roughly $9B or $10B, and characterizing the original as a low estimate.

Assessment: The language escalated from April's "compelling growth" to "extraordinary setup," and the supporting facts (8 to 10 fabs, customers securing orders at or beyond lead time, an undersupplied industry rolling forward) are concrete enough to carry weight. But the refusal to quantify across four attempts, in a print where management was otherwise willing to raise a multi-year margin framework, is a real asymmetry. The most defensible reading is that management is confident in the direction and genuinely uncertain about magnitude, which for a business now guiding 20% sequential growth is a fair position. We model calendar 2027 WFE at $170B to $185B.

4. The September Guide: $8.10B and 20.5% Sequential

The September revenue guide of $8.10B plus or minus $400M is the largest single-quarter guidance beat in this coverage arc, roughly $1.0B and 14% above a Street sitting near $7.1B. It implies 20.5% sequential growth and roughly 52% year-over-year growth against a September 2025 quarter of approximately $5.32B.

The margin structure inside the guide is more informative than the headline. Gross margin holds at 52.0% while operating margin rises 110bps to 39.5%, which means the entire incremental margin comes from operating leverage rather than further gross margin expansion. Implied operating expenses of roughly $1,013M represent 10.5% sequential growth against 20.5% revenue growth.

"I would just mention, we are growing spending in the September quarter albeit at a much slower rate than the growth in revenue." — Douglas Bettinger, CFO

The implied incremental operating margin is 44.9%, down from the 60.6% Lam actually delivered in June but well above the 38.4% current operating margin, which is the arithmetic that drives the margin expansion. EPS of $2.15 represents 18.1% sequential growth, less than the 20.5% revenue growth, entirely because of tax.

Assessment: A 20.5% sequential step in a business this size is extraordinary and it is the single strongest refutation of the AI-capex-peak thesis that drove the stock down 38.6% into the print. It is also, historically, the kind of print that appears near cycle peaks rather than in mid-cycle, and Lam's own 2022 experience is the cautionary reference. The distinguishing feature this time is that management has forward visibility from named fab openings rather than from order momentum, which is a better-quality signal, but it is not proof.

5. The Tax Rate Steps Up, Permanently

"We do believe the tax rate will be in the mid-teens in the September quarter due to the increase in revenue in higher tax jurisdictions, primarily the United States as we enter the new fiscal year. The U.S. GILTI rate is also higher as we enter the fiscal year. We should expect this uptick in the tax rate to continue for the remainder of 2026 and likely beyond." — Douglas Bettinger, CFO

The June quarter's non-GAAP tax rate was 11%. The September guide implies roughly 16%. Holding the tax rate at 11%, the September guide's operating income would produce roughly $2.28 of EPS rather than $2.15, so the step-up costs about $0.13 per share in the quarter, or roughly $0.52 annualized at current earnings levels and more as earnings grow.

The phrase carrying the most weight is "and likely beyond." This is not a one-quarter mix effect, it is a structural repricing driven by revenue shifting into US jurisdictions plus a higher GILTI rate. Lam has enjoyed a single-digit-to-low-double-digit effective rate for several years (10.9% GAAP this quarter, 9.3% in March, 3.3% in the June 2025 quarter), and that era is ending.

Assessment: This received no analyst question and it should have. A permanent 500bps step in the effective tax rate is worth roughly 6% of net income, and it partially offsets the operating-margin framework raise. Our FY27 and FY28 estimates assume a 16% rate. Anyone modeling Lam off historical tax rates will overstate EPS by roughly 6%.

6. Akara and the Conductor Etch Franchise

Lam disclosed an installed base of more than 40,000 conductor etch chambers worldwide, and framed Akara as the platform extending that leadership. The specific claims were that Akara combines direct-drive plasma technology with high-aspect-ratio patterning capability, that it was first adopted for 2nm-and-below gate-all-around in foundry/logic, and that it is now gaining momentum in advanced DRAM.

"We have secured several strategic tool of record positions, including recent wins for the most challenging gate etch applications. Since its launch, Akara's installed base has doubled every year, and we expect that growth trajectory to continue in 2027." — Timothy Archer, CEO

Tool-of-record positions at the most challenging gate etch applications are the highest-value placements in the category, and winning them at gate-all-around nodes locks in multi-year revenue as those nodes ramp. The cross-over into advanced DRAM is the structurally more interesting half, because it converts a foundry/logic platform investment into a second addressable market without incremental development cost.

Assessment: An installed base doubling annually with a stated expectation of continuing through 2027 is the most quantified product-momentum disclosure in the print. It supports the SAM-expansion pillar with something more concrete than a percentage target. The unquantified part is what Akara contributes in revenue, which management has never disclosed.

7. Advanced Packaging Raised a Third Time, to Above 70%

"Turning to advanced packaging. As an industry leader in TSV etch and electroplating, we are on track to deliver greater than 70% year-on-year growth." — Timothy Archer, CEO

The progression across three quarters is above 40%, then above 50%, then above 70%. Asked whether the raise reflected faster HBM, 2.5D and 3.5D transitions or customers pulling capability forward from next year, the CEO said it was all of them, and added that adoption is occurring broadly enough that the next few years will be hard to forecast.

The forward architecture case was the most specific Lam has made. Management expects each successive generation to require more redistribution layers, denser copper interconnects, taller mega-pillar structures and more complex power delivery, and sees future AI packages exceeding 9x reticle size, roughly 3x today's mainstream designs. That is the argument driving the shift to panel-level packaging, where Lam has shipped 510x515mm systems into development programs across multiple geographies and will ship its first 310x310mm panel tool this year.

Assessment: Three consecutive raises in three quarters is a forecasting process chasing a business that is inflecting faster than it can be modeled, which is the good version of a bad forecast. The panel-level positioning is early-stage optionality that we do not yet value: development-program shipments are not revenue, and management gave no adoption timeline. We treat panel as a free option on the 2028-plus horizon.

8. The Working Capital Build and the Capital Return Cut

Days sales outstanding rose to 72 days from 64 sequentially and 59 a year ago. Receivables ended at $5.340B, up 58.1% year-over-year against revenue up 30.0%. Inventory rose $276M sequentially to $4.276B, though turns improved to 3.0x, the best in nearly five years. Capital expenditures fell to $189M from $332M, and management reiterated a 4% to 5% of revenue framework with emphasis on lab infrastructure and footprint densification.

"As the business grows, we'd like to build a little bit more cash on the balance sheet to support potential liquidity needs like capital spending and working capital." — Douglas Bettinger, CFO

The buyback fell to $247M from $1,163M, a 79% cut, taking capital return to 45% of free cash flow from 184%. Cash rose to $5.579B from $4.751B. Management restated the policy of returning at least 85% of free cash flow over time, with $4B remaining on the authorization.

Deferred revenue rose $213M to $2.43B, and the CFO identified customer down payments as the largest contributor. That is a reversal worth noting: in April, down payments had fallen roughly $300M to a four-year low, and management framed that positively as evidence growth was coming from large multinationals who do not pay deposits. Both framings cannot be right, and neither was challenged on the call.

Assessment: The working capital build is a growth consequence rather than a warning, but it converts a 35.6% net income increase into a 9.7% free cash flow decline, and free cash flow is what funds the return policy. The buyback cut is defensible on both stated and unstated grounds. The unexplained down-payment reversal is the loose thread, and it will matter if it signals customers locking capacity ahead of a supply squeeze, which would be a pull-forward.

9. SAM Expansion Running Ahead of the Investor Day Plan

"We see the benefit of this technology acceleration in our expanding SAM, and we are moving towards our target of high 30s SAM as a percentage of WFE faster than what we had outlined at our 2025 Investor Day." — Timothy Archer, CEO

Under questioning the CFO put a number on it: roughly 36% to 36.5% this year, against a low-30s starting point at the 2025 Investor Day and a high-30s target. He also corrected an analyst's back-of-envelope share calculation, noting that dividing Lam revenue by WFE conflates CSBG, which is not purely WFE-driven, and therefore understates the systems-level share.

At a low-$150B WFE base, each percentage point of SAM is worth roughly $1.5B of addressable market. Moving from 36% to high-30s over the next few years, on a WFE base that is itself growing, compounds two ways.

Assessment: This pillar is now quantified and running ahead of plan, which is the cleanest confirmation of the structural thesis in the print. The distinction management drew between SAM and revenue matters: SAM expansion is a prerequisite for growth, not growth itself, and Lam's share-within-SAM remains undisclosed.

10. A New Logic Customer in the United States

"Yes. CJ, you're funny. Asking that tricky question about new logic customers. Yes, there's some of the stuff going on. There's clearly dialogue, and I'll let Tim, comment a little bit. We are talking to that new logic customer in the U.S. that is happening." — Douglas Bettinger, CFO

The CEO added the strategic framing: new entrants are not encumbered by existing roadmaps and installed base, which makes them more willing to evaluate innovative approaches, and therefore represent an opportunity to gain share in foundry/logic where Lam's position is weaker than in memory.

Assessment: Confirmed dialogue is not revenue, and no timeline, customer or magnitude was given. But the logic is sound: greenfield logic capacity in the US evaluated without incumbency bias is precisely where a challenger portfolio converts. We carry no revenue for this and treat it as optionality on FY28 and beyond.

Guidance & Outlook

Implied FY27 trajectory: The September guide alone, held flat for four quarters with no further growth, produces FY27 revenue of $32.4B, up 39.4% from FY26's $23.233B. Management stated it expects sequential growth in each successive quarter, so flat is a floor rather than a forecast. Our base case of modest sequential progression through the year puts FY27 revenue at $33B to $35B.

Street positioning: Consensus sat near $7.1B for the September quarter before the print, so the entire Street model requires rebuilding from a base roughly $1B per quarter too low. That repricing was under way through July 30 and is the mechanical driver of the price-target raises that followed.

Guidance style: Lam guided four lines in April and beat three above the high end. Across the four quarters we have covered, the pattern is consistent conservatism on margin and honest midpoints on revenue. We treat the September gross margin guide of 52.0% as carrying upside skew and the revenue midpoint as genuinely centered.

Analyst Q&A Highlights

Timeline to the Mid-50s Gross Margin Target

The opening exchange went straight to the framework revision and asked the question management had not addressed in the prepared remarks: how long. The answer separated the drivers into scale and revenue growth on one hand, and new product introduction and pricing for value on the other, and put the horizon at several years without narrowing it. The question also surfaced a useful benchmark, that Lam's gross margin gap to large foundry customers had widened from roughly 5 points in early 2024 to roughly two to three times that, implying substantial room to close.

Q: "so the new gross margin of mid-50s, that makes perfect sense. That's great. And I guess the question is, how long will it take to get gross margin to that level? I know -- I mean, obviously, if you look at your margins versus, say, the large foundry margins, I mean it used to be within 5 points and now it's between 5 and 7 points as early as the first half of '24. And now you're like the gap is like 2x that. So there's obviously a lot of room for you to move it up. How long will it take for you to move that up? Is it like a revenue thing? Or is it a time thing?"
— Timothy Arcuri, UBS

A: "A little bit of both, Tim, honestly, right? Part of it is scale and scope and revenue growth. Part of it is new product introduction and getting fairly paid for the value we're delivering. I expect, Tim, this is over the next several years that we'll continue to drive it on an annual basis for sure, but it's going to take several years, I think, to get to those levels, Tim."
— Douglas Bettinger, CFO

Assessment: "Several years" with annual progress is a commitment to direction without a commitment to pace, and the difference between a three-year and a five-year path to mid-50s is material to any discounted valuation. The useful anchor came separately, when management said the 51% to 52% range is deliverable in the near term. We model gross margin at 52% to 53% through FY27 and treat the mid-50s as an FY29-plus outcome.

Advanced Packaging Raised Three Times in Three Quarters

A question on the advanced packaging trajectory laid out the sequence explicitly and asked whether the acceleration reflected genuine transition speed, HBM strength, or customers simply pulling next year's capability into this year. The answer declined to separate them, which is itself informative about how broadly the adoption is occurring.

Q: "And then on advanced packaging, you came into this year with a view of greater than 40% growth. You upped that to greater than 50% growth last earnings. Now you're looking at 70% growth today, right? Is the increase in the outlook due to HBM or 2.5D, the 3.5D SoIC advanced packaging transitions happening faster or is it customers just pulling in production capability from next year into this year, curious?"
— Harlan Sur, JPMorgan

A: "Yes. I mean, Harlan, it's just everything. I mean advanced packaging, I mentioned in my prepared remarks, it's becoming quite a technical tool for our customers and for the industry to drive greater performance. So it's everything you just mentioned. It's 2.5D, it's in the foundry/logic space, it's in HBM. And I think that what we're excited about is, obviously, as it moves also into panel packaging, that's a place where Lam we feel like we've gotten out to an early start there, and we think that transition is in very early stages but it's an important inflection point there as well. So I think just next few years, advanced packaging growth will be a little hard to predict because adoption is just occurring all over the place."
— Timothy Archer, CEO

Assessment: The pull-forward possibility embedded in the question was not addressed, which leaves it open. "Adoption is just occurring all over the place" is genuine breadth if true and a forecasting abdication if not, and the three consecutive raises argue for the former. The unquantified base remains the gap: 70% growth on an undisclosed number is a rate without a magnitude.

Third-Party Forecasts of NAND Oversupply in 2027

The sharpest challenge of the call put an external forecast directly to management: a divergence in memory next year with NAND oversupplied on weak consumer demand while DRAM stays short on data-center strength. This matters because NAND doubling was the quarter's headline driver. The answer was measured rather than dismissive and pivoted deliberately away from NAND dependence.

Q: "There is a third-party trend forecasting a divergence in memory fundamentals next year, they're calling for an oversupply in NAND because of the NAND demand getting pulled down by weaker consumer and shortages to continue on the DRAM side, which is more driven by data center. Are you seeing anything in your conversation with the memory makers that are pointing to some sort of oversupply or a reversal in the NAND fundamentals for next year?"
— Atif Malik, Citi

A: "I think that it's -- as we've said on NAND, I mean, clearly, customers are right now progressing through upgrades to 200-plus layer devices. There's some greenfield this year. it's a ways out before there's a lot of greenfield coming into NAND. So I think that, again, next year, we haven't -- we're still having conversations about how to continue to upgrade the existing installed base and get the fleet up to kind of current state of the art in terms of NAND. And so I think it's a little early on 2028. But our view right now is the fundamentals are still about the same for us. I think what I tried to highlight in my prepared remarks is the Lam story is a lot bigger than just NAND."
— Timothy Archer, CEO

Assessment: The substantive answer is that Lam's NAND revenue is driven by installed-base conversion rather than by bit demand, which decouples it from the oversupply question in the near term but not indefinitely: customers experiencing NAND oversupply eventually slow conversion spending too. The pivot to "the Lam story is a lot bigger than just NAND" was the right defensive framing and is supported by the DRAM and foundry disclosures, but it also concedes the premise rather than refuting it.

How the Clean-Room Constraint Bent from $140B to $150B

Because management had spent two quarters describing clean-room availability as the binding limit on 2026 spending, the $10B raise required an explanation. The answer was candid about the mechanism and about its limits, and included an unprompted correction of the prepared remarks on whether upside bias still attaches.

Q: "Tim, on your WFE comment about $150 billion I think you said -- you alluded to maybe further upward bias for the year. I'm just trying to get a sense of what your lead times are? And in case if there is more upside, I guess, in the second half of the year, how well positioned you are to be able to supply to any potential upside?"
— Srinivas Pajjuri, RBC Capital Markets

A: "Yes. Sorry, that might have been poorly worded in the remarks. It was -- our previous quarter's guidance was $140 billion with upside bias and that upside bias played out to get us to $150 billion now. So this outlook was in the $150 billion range. We didn't say upside bias for this this current outlook... I think as we move through the second half of the year, that becomes more and more difficult to see anything further in this year as true surprise upside. And that's why the discussions are now out in the '27 and beyond."
— Timothy Archer, CEO

Assessment: The correction is the valuable part. April's upside bias has been consumed, not renewed, so anyone modeling further 2026 WFE upside is double-counting. The constraint proved elastic under demand pressure through bottleneck-tool resolution and squeezed space, but management explicitly flagged that lever as largely exhausted for this year. That makes 2027 dependent on genuinely new clean-room capacity rather than on better utilization of existing space.

Whether Sequential Growth Can Continue Through the End of 2027

Having failed to extract a 2027 WFE number, a line of questioning tried the same target from a different angle: not how big, but for how long. The reframing was clever because a commitment to sequential growth through the end of 2027 would imply a magnitude without stating one. Management declined again but gave its most forward-leaning phrasing of the call.

Q: "Got it. For my follow-up, let me try one more way. Again, I'll try to ask this without having to put numbers on it, but if '27 is going to be that good. Do you think the setup is good enough where, at a minimum, we could see sequential growth from here until the end of '27? Is the setup and I guess, the availability of space enough that at least you could be willing to sign up for something like that?"
— Stacy Rasgon, Bernstein Research

A: "Yes, maybe, Stacy, when you look at it, this doesn't all come on in any one quarter. So it comes on kind of bit by bit. again, not going to guide you quarter by quarter through next year. But I feel incrementally good about each successive quarter as I sit here right now. And as we get a little bit closer, maybe I'll give you a little more color."
— Douglas Bettinger, CFO

Assessment: "I feel incrementally good about each successive quarter" is the closest management came to a 2027 commitment and it is a meaningful one, because it is a statement about direction across eighteen months rather than about a single quarter. Combined with the separately disclosed 8 to 10 new fabs opening by end-2027, it supports a model where growth continues through the period, which is what our FY27 and FY28 estimates assume. It is not a guarantee and was not offered as one.

The Second NAND Upgrade Wave Beyond the $40B Program

A question probed whether the $40B conversion opportunity, now nearly realized, is a one-time event or the first iteration of a recurring cycle. The answer confirmed recurrence, explained why the next wave should be larger, and volunteered that it has not been sized.

Q: "So it seems like the company is pretty close to realizing the $40 billion upgrade opportunity faster than expected. But as you've described in the past, that's not so much of a static opportunity. Like whatever has been upgraded to 200 layers eventually has to migrate to 300 layers and above, which could trigger another wave of spending for NAND where are we in that kind of second phase of upgrades? And is there a way to kind of contextualize how big that opportunity could be relative to the initial $40 billion upgrade opportunity you saw in the last couple of years?"
— Michael Mani (for Vivek Arya), Bank of America Securities

A: "most of that $40 billion we had previously said would likely occur -- and upgrades would likely occur before the end of 2027 but then as you pointed out, it kind of all starts again. But the key is since greenfield additions have been made in that period of time. The next time it rolls through, you go from 200 to 300 plus or 400 plus, it's an even bigger installed base. And so while we have not -- we haven't quantified that, but it's a good action item for us to get to you into the future."
— Timothy Archer, CEO

Assessment: This is the most important forward disclosure for the NAND franchise beyond 2027 and it is entirely unquantified, with management explicitly accepting the action item. The structural logic is sound: a larger installed base multiplied by a per-wafer SAM that doubles from 128-layer to 500-plus-layer implies the second wave exceeds the first. We carry no explicit revenue for it and flag it as the largest unmodeled upside in the name.

Pricing as a Named Gross Margin Driver, Declined Twice

Pricing was listed first among the gross margin drivers in the prepared remarks, which is a change in emphasis from prior quarters. Two separate lines of questioning tried to get behind it, one asking what specifically drove the pricing gains and whether like-for-like increases were achievable within existing programs, the other asking directly about industry-wide like-for-like pricing. Both received the same framework-level answer.

Q: "I actually wanted to ask on pricing. It's been kind of an investor theme, and I'm kind of just curious what you're seeing like-for-like pricing in the industry and you."
— Multiple analysts incl. Blayne Curtis, Jefferies; James Schneider, Goldman Sachs

A: "Yes, Blayne, I'm not going to talk about like-for-like pricing. When I described the solid gross margin that we saw last quarter, I talked a little bit about pricing, about operational and scale efficiencies and about product mix, all of that contributed. We're always doing everything we can to get fairly paid for the value we're delivering. That's true today. It's been true for, I don't know, a decade or longer, but we're working on all of those things, Blayne."
— Douglas Bettinger, CFO

Assessment: The refusal is standard practice and not itself a red flag, but it leaves the most cyclically sensitive gross margin contributor unquantified at exactly the moment management raised its long-term margin target by 500bps. Pricing power in a supply-constrained upcycle is real and also the first thing to reverse when the constraint lifts. Operational efficiency and footprint are durable; pricing is conditional. Without the split, the durability of the mid-50s target through a downturn cannot be assessed from the outside, and that is the strongest remaining argument for the bear case on the framework revision.

What They're NOT Saying

1. Any 2027 WFE number: Four separate attempts, four refusals, against language as strong as "extraordinary setup" and "unprecedented long-term demand visibility." The gap between the qualitative confidence and the quantitative silence is the single largest disclosure asymmetry in the print.
2. The timeline to mid-50s gross margin and mid-40s operating margin: "Several years" only. A three-year path and a six-year path are very different assets, and management declined to narrow it even when asked directly.
3. Like-for-like pricing: Named first among gross margin drivers in prepared remarks, then declined twice under questioning. This is the contributor most likely to reverse in a downturn and the one we cannot size.
4. The size of the second NAND upgrade wave: The 200-to-300-to-500-layer cycle was confirmed as recurring and larger than the first, then explicitly left unquantified with an acknowledged action item.
5. The receivables build: DSO went from 59 to 72 days year-over-year and free cash flow fell 9.7% while net income rose 35.6%. One question was asked about inventory. None was asked about receivables, and management did not raise it.
6. Why customer down payments reversed: Down payments were the largest driver of the $213M deferred revenue increase this quarter, three months after management framed their decline to a four-year low as a positive signal about customer quality. Both framings were offered without reconciliation.
7. Advanced packaging revenue base: Growth above 70% disclosed for a third consecutive quarter; the dollar base has never been given, so the contribution cannot be computed.
8. The new US logic customer: Confirmed as "happening" with no name, timeline, magnitude, or qualification stage.
9. Akara, Argos and VECTOR revenue contribution: Installed base doubling annually, tool-of-record wins at the most challenging gate applications, and specific performance claims (20% cost savings, 5% capacitance reduction), with no revenue attached to any of them.
10. HBM4, HBM4E and high-bandwidth flash capital intensity: Asked for a per-wafer sizing and declined. Management confirmed taller stacks mean longer process times and more equipment, but would not quantify.
11. September CSBG guidance: Declined specifics, offering only that the profile will look similar. Given CSBG beat its own informal April guidance by 17%, the refusal to re-guide is notable.
12. China domestic lithography: Not raised by management and not asked by any analyst, in a month when China began mass-producing domestic immersion DUV systems and the headline contributed to a double-digit decline in the stock.

Market Reaction

• Pre-print setup (July 29 close, $252.35): down 38.6% over the trailing 30 days from $410.91 on June 29, yet still up 47.4% year-to-date and up 155.1% over twelve months. The 52-week closing range was $94.84 to $433.33, so the stock entered the print 41.8% below its own high while still having more than doubled over a year. The S&P 500 was up 6.9% year-to-date over the same period.
• What drove the drawdown: a sector-wide de-rating on the thesis that AI infrastructure spending is nearer a peak than a floor, triggered in the first week of July when a report that a major hyperscaler was building a business to resell excess AI compute capacity was read as evidence of overbuild. Every wafer-fab-equipment name fell together. The unwind was amplified by positioning, since the stock had roughly doubled in the first half of 2026, and compounded mid-month by reports that China had begun mass-producing domestic immersion DUV lithography systems, then by pre-earnings de-risking in the final week.
• Reaction session (July 30): opened at $318.29, a 26.1% gap, traded a range of $294.25 to $319.50, and closed at $297.72, up 18.0% or $45.37. Volume of 22.1M shares against a 13.0M thirty-day average, 1.7x normal. The S&P 500 rose 1.7% the same session.
• Net position: after an 18.0% single-session gain, the stock remains 31.3% below its 52-week closing high and 27.5% below where it traded a month earlier.

The reaction is best understood as a repricing of the fundamentals, not a repricing of sentiment. The July drawdown was an argument about whether AI-driven equipment demand was peaking. The print answered that argument with the largest guidance beat in the company's recent history, a second consecutive increase in the industry spending outlook, and a multi-year margin framework raised rather than defended. None of those are sentiment items.

The more telling detail is the fade. A 26.1% opening gap closed at 18.0%, giving back roughly a third of the initial move intraday on 1.7x volume. That is not the signature of a market that has fully re-embraced the story. It is the signature of one absorbing a genuine earnings revision while remaining unwilling to pay the prior multiple, which is a rational response given that the drawdown's underlying concern was about 2028, and this print addressed the next four quarters rather than 2028.

The most useful arithmetic sits underneath the price. Before the print, at $252.35 against our prior FY27 non-GAAP EPS estimate of roughly $7.05, the stock traded near 35.8x. After the print, at $297.72 against a revised estimate of roughly $9.35, it trades near 31.8x. The shares rose 18.0% and de-rated roughly 11% at the same time, because earnings power was revised up by roughly a third. A stock that gets cheaper on an 18% rally is an unusual setup and it is the core of the case for staying constructive here.

Street Perspective

Debate: Is 2027 the peak, or the middle?

Bull view: The industry is undersupplied entering 2027, 8 to 10 new fabs are opening by the end of next year, and customers are securing equipment at or beyond lead time to fill clean-room space that does not yet exist. Spending is being driven by capacity that has already been committed and announced, which is a far better-quality signal than order momentum. Management's willingness to raise a multi-year margin framework implies internal confidence extending well past 2027.

Bear view: A 20.5% sequential revenue guide is what the top of a cycle looks like, not the middle. The July drawdown was the market pricing the possibility that hyperscaler capex is closer to a peak than a floor, and nothing in this print speaks to 2028. Management refused a 2027 number four times, which for a team that just raised a multi-year margin target is a conspicuous asymmetry. Clean-room elasticity has been spent, so 2027 requires genuinely new capacity to arrive on schedule.

Our take: The bull case has the better evidence for the next four to six quarters and the bear case has the better question about what follows. Announced fab openings are hard commitments with multi-year lead times and do not unwind on a quarter of softer hyperscaler spending, which is why we are comfortable through FY27. The honest position is that neither side can currently evidence 2028, and management's refusal to quantify 2027 is more likely genuine uncertainty about magnitude than concealment about direction. That argues for maintaining the rating and not for increasing position size.

Debate: Is the mid-50s / mid-40s framework credible, or is it peak-cycle extrapolation?

Bull view: Lam has already passed a target model set only eighteen months ago, is printing a 20-year-high gross margin, and delivered a 60.6% incremental operating margin in the quarter with 44.9% guided. The drivers named (manufacturing footprint proximity, scale, new product value capture) are structural and were built over five years. Each incremental point of operating margin at the September revenue run rate is worth roughly $324M annualized.

Bear view: Pricing was listed first among the gross margin drivers and management declined to quantify it twice. Pricing power in a supply-constrained upcycle is the most reversible margin component, and the framework was raised at exactly the moment that power peaked. No timeline was attached, and a target reachable in "several years" is unfalsifiable in the interim.

Our take: The bear case correctly identifies the unquantified component but overstates its weight. Gross margin rose 210bps sequentially while China, historically the higher-margin mix, fell from 34% to 26% of revenue, which means the expansion happened against an adverse mix rather than because of a favorable one. That is hard to explain through pricing alone. We assume the mid-50s target is directionally real and slower than "several years" implies, and we model 52% to 53% through FY27 rather than underwriting the destination.

Debate: Does the free cash flow deterioration matter?

Bull view: Receivables growing faster than revenue is the arithmetic of a business shipping heavily into the back end of an accelerating quarter. Inventory turns improved to a five-year high, so nothing is stuck. The cash will convert as growth rates normalize, and the capital-return policy is a long-run commitment rather than a quarterly one.

Bear view: Fiscal 2026 free cash flow fell 9.7% while net income rose 35.6%, and conversion dropped from 101% to 67%. A policy of returning at least 85% of free cash flow applied to a shrinking free cash flow base means less capital return, and the June quarter already showed it: the buyback was cut 79%. If growth continues at this pace, working capital consumes more cash, not less.

Our take: This is the most under-discussed item in the print and neither camp is engaging with it. The bear arithmetic is correct and the bull explanation is also correct, which is the uncomfortable resolution: the cash gap is real, it is caused by growth rather than by deterioration, and it will persist for as long as the growth does. The practical consequence is that buyback support should be modeled down, not up, through FY27 despite record earnings. That is a modest negative for EPS growth and it does not change the thesis.

Model Update Needed

Valuation: At the July 30 close of $297.72 and roughly 1.255B diluted shares, market capitalization is about $374B, with $5.6B of cash against $3.7B of long-term debt for an enterprise value near $372B. On our revised FY27 non-GAAP EPS midpoint of $9.35, the shares trade at 31.8x, against 35.8x on the pre-print price and our prior estimate. On the FY28 midpoint of $11.35, they trade at 26.2x.

Our 12-month fair value range: $320 to $370, representing 7.5% to 24.3% above the July 30 close, and 28.2x to 32.6x our FY28 non-GAAP EPS midpoint. The range is set deliberately below the Street's consensus target of roughly $382 because we are unwilling to underwrite the mid-40s operating margin destination on an undefined timeline, and because the print addressed the next four to six quarters rather than the 2028 question that caused the July drawdown. The four supports are: (a) forward earnings power revised up by roughly a third with the stock still 31.3% below its 52-week high, (b) an industry spending outlook raised twice inside one calendar year with a third consecutive year of Lam outgrowing it, (c) a long-term margin framework raised rather than defended, and (d) SAM expansion now quantified at 36% to 36.5% and running ahead of the Investor Day pace.

Thesis Scorecard Post-Earnings

Overall: Thesis strengthened materially. Every commitment management made on the April call was delivered, and the two that mattered most (the framework update and the NAND conversion pull-forward) were delivered well above what we modeled. The bull pillars are now better evidenced than at any point in this coverage arc, with SAM expansion quantified, the margin ceiling formally raised, and the NAND thesis converted from forecast to print. Two new bear points enter the standing thesis, both quantitative and neither fatal: cash conversion has deteriorated on working capital, and the tax rate has stepped up permanently.

Action: Maintaining Outperform, position sizing 3% to 5% unchanged. The case for holding rather than adding is that this print resolved the next four to six quarters and left 2028 exactly where the July drawdown found it, and management's four refusals to quantify 2027 mean we are extrapolating where they will not. The case for not trimming into an 18% rally is arithmetic: forward earnings rose faster than the price, so the shares are cheaper now than before the print and still sit 31.3% below their high. Re-rate to high conviction on (a) a quantified CY27 WFE framework, (b) evidence that free cash flow conversion is recovering toward 80%-plus, or (c) a named commitment on the new US logic customer. Re-rate down to Hold on (a) a sequential guide that breaks the growth pattern management described, (b) gross margin retreating below 51% while pricing remains unquantified, or (c) evidence that the NAND conversion wave is pulling forward 2027 demand rather than adding to it.