Lam Research LRCX
Technology · Semiconductors · Synthos Deep Dive · 2026-08-04
The Overview
Making a computer chip means building it up in hundreds of layers on a silicon wafer. Some steps add material; some steps carve it away. Lam Research makes the machines that do the adding (deposition) and the carving (etch), plus the machines that clean the wafer between steps. It sells those machines to about a dozen customers worldwide — Samsung and TSMC are the two biggest — and then earns a second, steadier stream of money servicing the machines it has already sold. That service business was $6.9 billion last year, about 38 cents of every dollar of revenue, and it grows every time another machine ships.
Lam is unusually tied to memory chips — the kind that store data rather than do the calculating. That matters enormously right now, because memory is in a boom. Micron, the big American memory maker, has gone from $9.3 billion of quarterly revenue to $41.5 billion in a little over a year and is now worth more than a trillion dollars. Memory companies that make that much money eventually spend it on new factories, and new factories mean new Lam machines. That is the whole bull case, and today the market voted for it: Lam rose 7.85%, more than any other equipment maker.
Here is the catch, and it is the reason this is a "watch" and not a "buy." A year ago you could have bought this company for about $96 a share. It now costs $317.74. In that year the company's profits per share went up by 39%. The price went up by 230%. The difference is the market simply deciding to pay far more for each dollar of profit — from about 23 dollars of price per dollar of annual profit to about 55. So most of the last year's gain was a change of opinion, not a change of business.
And the memory boom has not actually shown up in Lam's numbers yet in the way you might expect. Over the same period that Micron's revenue multiplied by four and a half, Lam's grew 42%. Lam's own quarterly filing says most of this year's growth came from foundry customers — the ones making processors — not from memory. Two other quiet warnings sit in the filings: customers' advance deposits have shrunk from $2.77 billion to $2.22 billion over two quarters, and China, which is subject to US export restrictions, is now about 37% of sales.
Meanwhile every company insider in our records has been selling — the chief executive sold 30,000 shares at $390 in early July, two directors sold in the $335 to $359 range — all at prices above where the stock trades today. The company itself bought back its own shares at roughly $91 and roughly $219 in its two most recent accelerated programmes, which is a long way below $318.
We think fair value is about $353, roughly 11% above the current price and slightly below what the average analyst thinks. In a bad scenario we get $205. In a good one, $430. That is an even bet, and an even bet is not a reason to act.
- Downside Risk 6/10. The balance sheet cannot break; the multiple can. 55x trailing, beta 1.8, 37% China, two customers at a third of revenue.
- Growth Quality 8/10. Revenue +26.0%, gross margin 50.5% and rising, eight straight sequential increases, five straight beats — but it is capital-cycle growth, and FY2024 fell 14.5%.
- Exponential Potential 6/10. An exponential end-market routed through a cyclical, concentrated, export-controlled equipment channel.
Putting a number on it: our fair-value estimate is $353 against a current price of $317.74 — real upside if our numbers are right.
Our summary metrics
"Rated 6 — moderate-high, and the risk is valuation and cyclicality rather than solvency. The supports are real: net cash of $1.845B ($5.579B of cash against $3.735B of total debt, of which $3.75B is senior unsecured notes per the Q3 FY2026 10-Q and $4.1M is finance leases — no double-count); interest coverage of 66.0x; a current ratio of 2.63; gross margin of 50.5% for FY2026 and 51.7% in the June quarter; return on invested capital of 41.0% and return on equity of 67.0%; no outstanding commercial paper against a $2.00B programme and a $2.00B revolver; and five consecutive earnings beats. Against that: the equity trades at 55.2x trailing and 32.0x book value, with a beta of 1.805; it is already 26.7% below its 52-week closing high of $433.33 after tripling in twelve months, which is a demonstration of how far this name travels in both directions; China was approximately 37% of revenue for the nine months ended 2026-03-29 (up from 34% in FY2025) and the 10-K devotes a dedicated risk factor to export licence requirements there; two customers were 17% and 15% of FY2025 revenue; deferred revenue has fallen from $2.77B (September 2025 quarter) to $2.25B (December 2025) to $2.22B (March 2026) on declining customer down payments; operating cash flow FELL 5.1% in FY2026 while net income rose 35.6%, taking income quality to 0.81 as the ramp absorbs working capital; and the effective tax rate is 12.1%, which is a support for forward earnings and a fragile one. A 6 rather than a 7 because the balance sheet genuinely cannot break the company."
"Rated 8 — the growth is real, accelerating, high-quality and cyclical, and only the last word keeps it off a 9. Revenue: $14.626B (FY2021), $17.227B (FY2022), $17.427B (FY2023), $14.905B (FY2024, −14.5%), $18.436B (FY2025, +23.7%), $23.233B (FY2026, +26.0%) — note the FY2024 decline, which is what this industry does. The percentage changes for FY2025 and FY2024 are confirmed verbatim in the FY2025 10-K, so the vendor's revenue series ties to the filing. Quarterly revenue has risen sequentially in all eight quarters in the file, from $4.168B to $6.722B, and the June 2026 quarter was +30.0% year over year. Gross margin has expanded in every year since FY2023: 44.6% to 47.3% to 48.7% to 50.5%, with 51.7% in the June quarter, driven per the filings by factory efficiencies and customer mix. Operating margin went 28.6% to 32.0% to 35.3%. Research and development spend rose from $1.727B to $2.376B over three years and is 10.2% of revenue. Consensus models $34.573B of revenue in FY2027 (+48.8%, 20 analysts, range $29.329B-$37.757B) and $40.996B in FY2028. The reasons this is an 8 and not a 9: the growth is capital-cycle growth, not subscription growth; the nine-month FY2026 increase was per the 10-Q predominantly Foundry-driven rather than memory-driven; and deferred revenue is going the wrong way."
"Rated 6 — structurally levered to an exponential end-market without being exponential itself. The case for a high number: every advanced logic and memory device requires more deposition and etch steps as the industry moves from two-dimensional scaling to vertical architectures, high-aspect-ratio 3D NAND, DRAM capacitor structures and advanced packaging, so Lam's served available market expands faster than wafer starts; the 10-K states this directly as the long-term thesis. The knowledge base records that semi-cap went from roughly twenty firms to three giants by acquisition, which is an oligopoly with pricing power. Customer Support Business Group revenue — spares, upgrades, services on the installed base — was $6.944B in FY2025, 37.7% of revenue, and grows with cumulative shipments rather than with the order cycle, which converts a chunk of the business into an annuity. The case against a higher number: Lam sells capital equipment to roughly a dozen customers, two of whom are a third of revenue; it cannot grow faster than its customers' capital budgets; and those budgets fell 14.5% as recently as FY2024. Exponential end-demand routed through a cyclical, concentrated, export-controlled distribution channel is a 6."
What does “fair value” mean?
Fair value is Synthos’s estimate of what one share is worth today, based on our model of the company’s future cash generation and the risks to it. It is not a price target or a prediction of where the stock trades next quarter — it’s the price where we think risk and reward are balanced. Above it, you’re paying for outcomes better than our base case; below it, the market is offering a margin of safety.
The model’s inputs and weightings are proprietary — they’re the product. What we publish is the output, plus an independent, public-math cross-check further down the page (“Check our number”) so you can judge it for yourself.
What do the 5 tiers mean? (Core · Tactical · Watch · Hold · Avoid)
The Road Ahead
What we expect to matter in each window, and the evidence that would prove us wrong.
Short term 0-6 months
No differentiated view- Driver
- "The stock closed 2026-08-04 at $317.74, up 7.85% on the day — the largest move among the four WFE names (KLAC +6.95%, AMAT +5.48%, ASML +4.22%) and almost exactly Micron's +7.62%, which is itself the cleanest available confirmation that the market treats Lam as the memory-levered expression of the group. But the move is a bounce inside a correction, not a breakout. After a 7.85% day the price is STILL 6.1% BELOW its 50-day moving average of $338.28 and 26.7% below its 52-week closing high of $433.33. MACD is −14.22, deeply negative. RSI is 45.3 — still under 50 after an 8% day, which tells you how damaged the momentum structure was going in. The 200-day moving average sits 23.6% lower at $242.72. Every insider transaction in the file is a disposition, executed between $335 and $390, all above today's close; the chief executive exercised options at $30.03 and sold 30,000 shares at $390.01 on 2026-07-02. Deferred revenue has fallen from $2.77B to $2.22B across the last two reported quarters. And there is no print for 78 days."
- What we’re watching
- "Whether the 50-day moving average at $338.28 is reclaimed and held, which would be the first genuine technical repair and the trigger to move from Watch to a staged entry. Failing that, whether the pre-gap level around $294.61 and the $270-295 band hold on a retest. Also: any change to US export licence requirements affecting China, which was approximately 37% of nine-month FY2026 revenue; further insider dispositions below $320; and the 2026-10-21 print against the $8.085B revenue and $2.16 EPS consensus."
- Confidence
- Medium
Medium term 6-24 months
Tailwind- Driver
- "This is where the memory linkage should pay, if it is going to. Micron's quarterly revenue has gone from $9.3B to $41.5B over five quarters and it closed today at a $1.008 trillion market capitalisation. Memory profits convert into memory capital expenditure with a lag, and Lam is the most memory-exposed of the four WFE names — its Korea revenue alone was $4.128B in FY2025 (22.4% of the total) and rose 43.6% year over year, and the 10-K names Samsung Electronics as one of its two most significant customers. The March 2026 10-Q records the memory market segment strengthening across both DRAM and non-volatile memory. Consensus models $34.573B of FY2027 revenue against $23.233B delivered in FY2026, and $9.38 of FY2027 earnings per share against $5.76. If that lands, earnings roughly double in two years from a 50.5% gross-margin base with 41% return on invested capital. The medium-term stance is a tailwind on the business; it is the price that is the problem, not the trajectory."
- What we’re watching
- "Whether deferred revenue rebuilds — it has fallen $550M across two quarters and the 10-Q attributes the decline to customer down payments, which are the earliest hard signal of an order wave. Whether the memory share of the market-mix disclosure keeps rising (the percentage table itself was not extractable from these filings and must be read from the source documents). Whether gross margin holds above 50% as China mix and tariff-related spend move around. Whether operating cash flow reconverges with net income — income quality fell to 0.81 in FY2026 as working capital absorbed the ramp. And whether capital expenditure, which has risen from $396.7M to $759.2M to $966.4M and now runs at 2.19x depreciation, is capacity for a real multi-year wave or a cost the cycle will strand."
- Confidence
- Medium
Long term 2+ years
Tailwind- Driver
- "The structural case is the strongest thing about this name and it is not in dispute. Two-dimensional scaling is running out, so the industry is going vertical — 3D NAND with extreme high-aspect-ratio holes, DRAM capacitor cells, gate-all-around transistors, multiple patterning, advanced packaging. Every one of those inflections adds deposition and etch steps per wafer, which means Lam's served available market grows faster than wafer starts. The 10-K states this as the long-term thesis in plain terms. The industry has consolidated to a handful of scaled suppliers — the knowledge base records the move from roughly twenty firms to three giants by acquisition — and the Customer Support Business Group, at $6.944B or 37.7% of FY2025 revenue, converts the cumulative installed base into recurring spares, upgrades and service revenue that does not swing with the order cycle. Ninety-three percent of revenue is non-US, which is a risk and also a statement of where semiconductor manufacturing actually happens."
- What we’re watching
- "Whether export controls harden into a structural loss of the China business rather than a periodic constraint — China was 26% of revenue in FY2023, 42% in FY2024, 34% in FY2025 and approximately 37% in the first nine months of FY2026, which is not the profile of a settled situation. Whether customer concentration worsens beyond the FY2025 level of two customers at 17% and 15%. Whether Chinese domestic equipment makers close the technology gap in deposition and etch, which is where import substitution is most feasible. Whether the Customer Support share of revenue keeps rising, which is the quiet de-risking of the model. And whether the 12.1% effective tax rate survives global minimum-tax regimes."
- Confidence
- Medium
Exponential Potential
"Rated 6 — structurally levered to an exponential end-market without being exponential itself. The case for a high number: every advanced logic and memory device requires more deposition and etch steps as the industry moves from two-dimensional scaling to vertical architectures, high-aspect-ratio 3D NAND, DRAM capacitor structures and advanced packaging, so Lam's served available market expands faster than wafer starts; the 10-K states this directly as the long-term thesis. The knowledge base records that semi-cap went from roughly twenty firms to three giants by acquisition, which is an oligopoly with pricing power. Customer Support Business Group revenue — spares, upgrades, services on the installed base — was $6.944B in FY2025, 37.7% of revenue, and grows with cumulative shipments rather than with the order cycle, which converts a chunk of the business into an annuity. The case against a higher number: Lam sells capital equipment to roughly a dozen customers, two of whom are a third of revenue; it cannot grow faster than its customers' capital budgets; and those budgets fell 14.5% as recently as FY2024. Exponential end-demand routed through a cyclical, concentrated, export-controlled distribution channel is a 6."
What could take this further than the base case — and where the market may be underpricing it. Full forward-growth and acceleration math in Deeper analysis, §4 below.
Deeper analysis
Technicals, fundamentals, valuation, and the full expert-claim evidence panel — the detail behind the numbers above.
Reference table
| Street consensus | $371.29 (+16.9%) · median $355 · high $500 · low $290 · 39 buy, 10 hold, 1 sell |
| Valuation | 55.2x trailing GAAP diluted EPS · 33.9x FY2027E · 27.0x FY2028E · 32.0x book · EV/sales 17.0x · EV/EBITDA 45.4x |
| Balance sheet | NET CASH of $1.845B ($5.579B cash against $3.735B total debt). The vendor's enterprise value of $395.51B is correct — a rare clean case (Section 4) |
| Conviction | Low-Moderate — 10 raw KB hits, 8 used, 2 discarded; 3 of the 8 are management voice at low weight, leaving 5 independent claims across 3 sources. A thin lane for a $397B company |
| Technicals | 6.1% BELOW the 50-DMA ($338.28) even after today's 7.85% · 26.7% below the 52-week closing high of $433.33 · 30.9% above the 200-DMA · RSI 45.3 · MACD −14.22 · 12-month return +229.7% vs SPY +24.3% |
What the experts actually said 7 traceable claims on LRCX · showing the highest-conviction voices
“Best S&P performer of early 2026 and a best-stocks-of-2025 write-up; the winning trend continues into the new year.”
“AI-driven demand is reshaping the semiconductor industry; Lam's investments and execution velocity are generating momentum and driving outperformance during this critical growth phase.”
“Lam Research is stretched—up ~30% in three weeks with high stochastics; due for a pullback.”
Every claim reconciles to a real claim_id in the Synthos knowledge base — this is the evidence the verdict is built on, not vibes. Management (the company itself) is shown but half-weighted; one cautionary voice is included on purpose.
Price & moving averages 12 months · 50 & 200-day averages · 52-week range
Solid line = price · dashed line = 50-day average · dotted line = 200-day average · the two thin horizontal lines mark the 52-week high and low. Price above both averages is an uptrend.
Data summary: last close $309.14, 9% below the 50-day average ($338), 27% above the 200-day average ($244) — a mixed trend. 29% below the 52-week high of $433, 222% above the 52-week low of $96.
Bollinger Bands 20-day average ± 2 standard deviations
The shaded band widens when the stock gets more volatile. Riding the upper edge = strong momentum (sometimes stretched); the lower edge = weak / potentially oversold.
Data summary: price $309.14 is currently inside the band (band $263–$362).
RSI (14) momentum gauge · 0–100
Above 70 (overbought zone, shaded) = overbought, below 30 (oversold zone, shaded) = oversold. Currently 48.
MACD 12 / 26 / 9 · trend & momentum
The MACD line crossing above the signal line (bars flip to the up color) = momentum turning up; crossing below (bars flip to the down color) = turning down. Bar height = the size of that gap.
Data summary: MACD is currently above its signal line by 1.04, positive momentum.
Relative performance vs S&P 500 & its sector (XLK (sector)), set to 100 a year ago
Solid = LRCX · dashed = S&P 500 · dotted = XLK (sector). A rising line means it is beating that benchmark — the sector line shows whether it is a leader or laggard within its own group.
Forward revenue & earnings actual → estimate · "FY" = fiscal year, "E" = estimate
Darker bars = actual results, brighter = analyst estimates. Taller bars to the right = expected growth.
Key stats an RIA wants
1. What the business is — one segment, two revenue streams, three markets
Lam Research is, in the words of the FY2025 10-K, "a global supplier of innovative wafer fabrication equipment and services to the semiconductor industry." Incorporated 1980, headquartered in Fremont, California, listed since 1984. Chief executive Timothy Archer.
The 10-K states the company operates in ONE reportable business segment: manufacturing and servicing of wafer processing semiconductor manufacturing equipment. That matters for how the vendor data is read, and is addressed in Section 8.
Product lines (10-K Item 1): deposition — atomic layer deposition, chemical vapour deposition, electrochemical deposition (the ALTUS, VECTOR, SPEED, SABRE families) — and etch, including conductor etch, dielectric etch (the Vantex platform) and cryogenic low-temperature etch for high-aspect-ratio structures, plus wafer clean. The 10-K is specific that leading-edge memory devices have "especially challenging structures, such as extremely deep holes and trenches" and that its dielectric etch capability targets "3D NAND high aspect ratio hole, trench, contact, and capacitor cell applications." This is the technical basis of the memory leverage.
Two revenue streams, per Note 4 of the 10-K and the vendor's seg_prod, which tie exactly:
| Fiscal year | Systems | Customer Support and Other | Total | CSBG share |
|---|---|---|---|---|
| FY2021 | $9.765B | $4.861B | $14.626B | 33.2% |
| FY2022 | $11.322B | $5.905B | $17.227B | 34.3% |
| FY2023 | $10.696B | $6.733B | $17.427B | 38.6% |
| FY2024 | $8.922B | $5.984B | $14.905B | 40.1% |
| FY2025 | $11.491B | $6.944B | $18.436B | 37.7% |
Systems is new leading-edge equipment. Customer Support Business Group is service, spares, upgrades and the non-leading-edge Reliant product line. The CSBG share rises when systems fall — 40.1% in the FY2024 downturn, 33.2% at the FY2021 trough of the previous ratio — which is precisely the stabiliser you want in a capital-equipment business. FY2026 product-line data is not in the file (the most recent year is FY2025); the FY2026 10-K has not yet been filed.
Three markets — memory, foundry, and logic/integrated device manufacturing. The 10-K and both 10-Qs reference a table giving the percentage of revenue from each market, but that table is not present in the extracted filing text. We therefore cannot quote the memory/foundry/logic split and say so rather than estimating it. What we can quote is the narrative, which is more useful anyway:
> FY2025 10-K: "Revenue increased in fiscal year 2025 compared to fiscal year 2024 due to increased equipment spending by our customers across Memory and Foundry market segments."
> FY2024 (from the same 10-K): "Systems and customer-support related revenues declined in fiscal year 2024 primarily from weakness in the non-volatile memory market, partially offset by strength in DRAM."
> Q3 FY2026 10-Q (March 2026 quarter): "The decrease in the foundry market segment for the March 2026 quarter compared to the December 2025 quarter was primarily driven by lower mature node investments, while the memory market segment saw strengthened investments across both DRAM and non-volatile memory."
> Q3 FY2026 10-Q (nine months): "The increase in revenue in the nine months ended March 29, 2026 compared to the same period in the prior year was predominantly driven by increases in Foundry equipment spending by our customers, as well as higher customer support-related revenue mainly due to spares revenue."
That last sentence is the most important disclosure in this dive and it cuts against the consensus story. Through nine months of FY2026 — a period in which memory makers were printing historic profits — the growth was predominantly Foundry. Memory strengthened only in the March quarter.
Geography (FY2025, vendor seg_geo, which ties exactly to reported revenue and to the 10-K's stated percentages):
| Region | FY2025 | Share | FY2024 | Change |
|---|---|---|---|---|
| China | $6.205B | 33.7% | $6.294B | −1.4% |
| Korea | $4.128B | 22.4% | $2.874B | +43.6% |
| Taiwan | $3.445B | 18.7% | $1.672B | +106.1% |
| Japan | $1.881B | 10.2% | $1.460B | +28.8% |
| United States | $1.377B | 7.5% | $1.104B | +24.7% |
| South East Asia | $0.837B | 4.5% | $0.794B | +5.4% |
| Europe | $0.563B | 3.1% | $0.707B | −20.4% |
| Total | $18.436B | $14.905B |
The 10-K states China was "approximately 34%, 42%, and 26% of our total revenue for fiscal years 2025, 2024, and 2023" and that non-US sales were "approximately 93%, 93%, and 91%" — both tie to the vendor table to within rounding. But the 10-Qs update it and the vendor cannot: China was approximately 39% of revenue for the six months ended 2025-12-28 and approximately 37% for the nine months ended 2026-03-29. China exposure is rising again, and the 10-K carries a dedicated risk factor titled "Our Sales to Customers in China, a Significant Region for Us, Have Been Impacted, and are Likely to Be Materially and Adversely Affected by Export License Requirements."
Customer concentration (10-K Note 19): in FY2025 two customers were approximately 17% and 15% of total revenues; in FY2024 one customer was 17%; in FY2023 two were 22% and 16%. The 10-K names "Samsung Electronics Company, Ltd. and Taiwan Semiconductor Manufacturing Company" as the most significant customers across FY2023-FY2025. Three customers were 19%, 15% and 12% of accounts receivable at FY2025 year-end. Headcount was approximately 20,600 at 2026-03-29 per the 10-Q; the vendor profile says 19,000 — the filing wins.
2. The single most important thing — you are buying the memory cycle at a multiple that has already tripled
Two facts, placed next to each other, are the whole dive.
Fact one: the memory supercycle is real and Lam is its most levered equipment expression. Micron closed today at $892.67, +7.62%, a $1.008 trillion market capitalisation, with quarterly revenue up from $9.3B to $41.5B across five quarters. Memory profits become memory capital expenditure become Lam orders. Lam's Korea revenue — Samsung and the memory complex — grew 43.6% in FY2025 to $4.128B. The March 2026 10-Q records memory strengthening across both DRAM and non-volatile memory. Today's price action ranked the four WFE names in exactly the order of their memory leverage: LRCX +7.85%, KLAC +6.95%, AMAT +5.48%, ASML +4.22%, against Micron's +7.62%. The market is pricing Lam as memory beta, and it is right to.
Fact two: the conversion has been far smaller than the narrative, and the re-rating far larger.
| Five quarters ago | Latest | Change | |
|---|---|---|---|
| Micron quarterly revenue (batch reference) | $9.3B | $41.5B | +346% |
| Lam quarterly revenue | $4.720B (Mar-2025 qtr) | $6.722B (Jun-2026 qtr) | +42.4% |
And the re-rating:
| ~2026-08-04 minus 12 months | 2026-08-04 | Change | |
|---|---|---|---|
| Price (derived from the 12-month return of +229.7%) | ~$96.37 | $317.74 | +229.7% |
| Trailing diluted EPS | $4.15 (FY2025) | $5.76 (FY2026) | +38.8% |
| Trailing P/E | ~23.2x | 55.2x | +138% |
Multiply it out: 1.388 × 2.380 = 3.30. Roughly two-thirds of a 229.7% return was the market changing its mind about what a dollar of Lam's earnings is worth. That is not a criticism of the company — the earnings genuinely grew 38.8% and the gross margin genuinely expanded 180 basis points. It is a statement about what is left to capture.
Why this matters more for a WFE name than for almost anything else. Wafer-fab equipment is a capital cycle. Historically these businesses trade at low multiples on peak earnings and high multiples on trough earnings, because the market correctly discounts the mean reversion. Lam's own history in this file demonstrates it: revenue fell 14.5% in FY2024, diluted EPS fell from $3.32 to $2.90, and the FY2024 estimate row in the vendor's own file was $15.591B of revenue against $14.905B delivered. At $317.74 the market is paying 55.2x trailing and 27.0x FY2028 consensus earnings — a growth multiple, on rising earnings, in a cyclical. That is a bet that this cycle is structurally different. It might be: the 10-K's argument that 3D scaling, multiple patterning and advanced packaging permanently increase the number of deposition and etch steps per wafer is a genuine argument for a higher through-cycle multiple. But it is a bet, and it is already in the price.
The third fact, from the filings, that neither narrative accommodates. Deferred revenue — the balance of customer money received ahead of recognised revenue, and the earliest hard signal of an order wave — has gone the wrong way:
| Balance-sheet date | Deferred revenue | Source |
|---|---|---|
| 2024-06-30 (FY2024 end) | $1.552B | 10-K |
| 2025-06-29 (FY2025 end) | $2.7B | 10-K |
| 2025-09-28 (Q1 FY2026) | $2.77B | Q2 FY2026 10-Q |
| 2025-12-28 (Q2 FY2026) | $2.25B | Q2 and Q3 FY2026 10-Qs |
| 2026-03-29 (Q3 FY2026) | $2.22B | Q3 FY2026 10-Q |
Down $550M, or 20%, across two quarters. The Q2 10-Q attributes the December decline to "a decrease in customer down payments." The Q3 10-Q says the March quarter's small decline "included approximately $300 million of decreases in customer down payments that were largely offset by increases across other components of deferred revenue balance associated with growing business levels." This is management's own framing and it is discounted as such, but the direction of the balance is a fact, not a framing. It is not consistent with a customer base rushing to secure equipment slots.
3. Growth and margin trajectory — genuinely excellent, genuinely cyclical
| Fiscal year (ends late June) | Revenue | Growth | Gross margin | Operating margin | Net income | Diluted EPS | R&D | Diluted shares |
|---|---|---|---|---|---|---|---|---|
| FY2021 | $14.626B | — | 46.5% | 30.6% | $3.908B | $2.69 | $1.493B | 1.453B |
| FY2022 | $17.227B | +17.8% | 45.7% | 31.2% | $4.605B | $3.27 | $1.604B | 1.406B |
| FY2023 | $17.427B | +1.2% | 44.6% | 29.7% | $4.511B | $3.32 | $1.727B | 1.358B |
| FY2024 | $14.905B | −14.5% | 47.3% | 28.6% | $3.828B | $2.90 | $1.902B | 1.320B |
| FY2025 | $18.436B | +23.7% | 48.7% | 32.0% | $5.358B | $4.15 | $2.096B | 1.290B |
| FY2026 | $23.233B | +26.0% | 50.5% | 35.3% | $7.265B | $5.76 | $2.376B | 1.261B |
The FY2025 (+23.7%) and FY2024 (−14.5%) growth rates are stated verbatim in the FY2025 10-K, so the vendor's revenue series ties to the filing. Note the shape: this is not a compounder, it is a cycle with an upward drift. FY2024 fell 14.5% and EPS fell 12.7%.
Quarterly, the run is uninterrupted:
| Quarter | Revenue | Sequential | Gross margin | Operating margin | Diluted EPS |
|---|---|---|---|---|---|
| Q1 FY2025 (Sep-24) | $4.168B | — | 48.0% | 30.3% | $0.86 |
| Q2 FY2025 (Dec-24) | $4.376B | +5.0% | 47.4% | 30.5% | $0.92 |
| Q3 FY2025 (Mar-25) | $4.720B | +7.9% | 49.0% | 33.1% | $1.03 |
| Q4 FY2025 (Jun-25) | $5.171B | +9.6% | 50.1% | 33.7% | $1.35 |
| Q1 FY2026 (Sep-25) | $5.324B | +3.0% | 50.4% | 34.4% | $1.24 |
| Q2 FY2026 (Dec-25) | $5.345B | +0.4% | 49.6% | 33.9% | $1.26 |
| Q3 FY2026 (Mar-26) | $5.841B | +9.3% | 49.8% | 35.0% | $1.45 |
| Q4 FY2026 (Jun-26) | $6.722B | +15.1% | 51.7% | 37.4% | $1.81 |
Eight consecutive sequential increases; the June quarter was +30.0% year over year and the largest sequential step in the series. Gross margin at 51.7% is the highest in the file. The filings attribute the margin expansion to "improved factory efficiencies and favorable product mix" (FY2025 10-K) and, for the March quarter, "improved factory efficiencies" (Q3 10-Q), with the nine-month improvement due to "favorable changes in customer mix, slightly offset by reduced factory efficiencies from higher tariff-related spend." Tariffs are named as a live cost headwind.
Five consecutive earnings beats per earn_cal: $1.33 vs $1.21 · $1.26 vs $1.22 · $1.27 vs $1.17 · $1.47 vs $1.36 · $1.82 vs $1.69. Beats of 3% to 9%. (The earn_cal actuals differ from the GAAP diluted EPS in the income statement by one to two cents in three of the five quarters — a basis difference, immaterial, noted in Section 8.)
4. Balance sheet and the enterprise-value rebuild — a rare clean file
As at 2026-06-28 (FY2026 year-end, from the earnings release; the FY2026 10-K is not yet filed):
- Cash and equivalents $5.579B · short-term investments $0 · inventories $4.276B
- Total assets $23.530B · total liabilities $11.059B · total equity $12.471B
- Long-term debt $3.730B · finance leases $4.1M · short-term debt $0 · total debt $3.735B
- Net cash: $5.579B − $3.735B = +$1.845B
The debt reconciles to the filings and is not double-counted. The Q3 FY2026 10-Q states: "We have $3.75 billion in aggregate principal amount of senior unsecured notes outstanding." Carrying value of $3.730B against $3.75B principal is the normal discount and issuance-cost difference. The 10-Q also confirms that $750.0 million of senior notes matured and were settled during the March 2026 quarter, which exactly explains the vendor's short-term debt going from $749.7M at FY2025 to $0 at FY2026. The company had no outstanding borrowings under its $2.00B commercial paper programme (enlarged from $1.50B on 2026-03-10), backstopped by a $2.00B revolver.
The vendor's enterprise value is correct, and we say so because we usually cannot. enterpriseValueTTM is $395.507B; rebuilt independently, $397.356B market cap + $3.735B debt − $5.579B cash = $395.512B, a difference of 0.001%. Lam holds essentially no separate short-term investment portfolio — the Q3 10-Q describes liquidity purely as "cash, cash equivalents, and restricted cash" — so the classic omission does not arise here. Every EV-derived multiple in this file is therefore usable, which is unusual and worth stating plainly: EV/sales 17.0x, EV/EBITDA 45.4x, EV/free cash flow 80.9x, net debt to EBITDA −0.21x.
Returns and quality: return on invested capital 41.0%, return on equity 67.0%, return on assets 30.9%, return on capital employed 46.6%. Interest coverage 66.0x, debt to equity 0.299, financial leverage 1.89x, current ratio 2.63, quick ratio 1.91. Book value per share $9.94, tangible book $8.43 — so the shares trade at 32.0x book, which is what happens when a company earns 67% on equity and returns most of it.
Working capital is where the ramp shows up. Days of inventory outstanding 135.6, days of sales outstanding 83.9. Inventories of $4.276B are 18.2% of annual revenue. This is a build-to-order equipment maker at the start of an up-leg, and inventory ahead of shipments is the correct behaviour — but it is also cash, and Section 5 shows where it went.
5. Cash flow and capital returns — and the yield that quietly disappeared
| Fiscal year | Operating cash flow | Capex | Free cash flow | Buyback | Dividends | Total returned | SBC | D&A |
|---|---|---|---|---|---|---|---|---|
| FY2023 | $5.179B | $501.6M | $4.677B | $2.017B | $907.9M | $2.925B | $286.6M | $342.4M |
| FY2024 | $4.652B | $396.7M | $4.256B | $2.843B | $1.019B | $3.862B | $293.1M | $359.7M |
| FY2025 | $6.173B | $759.2M | $5.414B | $3.422B | $1.150B | $4.572B | $343.4M | $386.3M |
| FY2026 | $5.858B | $966.4M | $4.891B | $3.851B | $1.271B | $5.122B | $386.4M | $441.5M |
Three things in this table deserve attention.
First, operating cash flow FELL 5.1% in FY2026 while net income rose 35.6%. Income quality — operating cash flow divided by net income — dropped from 1.15 in FY2025 to 0.81 in FY2026. That is the working-capital absorption of a ramp: inventory and receivables funding shipments that have not yet been collected. It is the right kind of cash-flow deterioration if the shipments are real, and the wrong kind if they are not. It is a number to watch, not yet a number to worry about.
Second, capital expenditure has more than doubled in two years — $396.7M to $759.2M to $966.4M — and now runs at 2.19x depreciation and 4.16% of revenue. Lam is building capacity for a wave it believes is coming. If the wave lands, this is the correct investment. If it does not, it is stranded fixed cost against a business whose revenue fell 14.5% two years ago.
Third, and most importantly: the capital-return story has become immaterial at this price. Lam returned $5.122B in FY2026 — a genuinely large sum, more than its $4.891B of free cash flow. On a $397.36B market capitalisation that is a 1.29% shareholder yield (0.97% buyback, 0.33% dividend), against a 1.23% free-cash-flow yield. Screens and sector narratives still file Lam under "capital returner." At 55x earnings it is not one. The same $3.851B of repurchase against the roughly $120B market capitalisation of a year ago would have been a 3.2% yield. Nothing about the company's policy changed; the denominator tripled.
And the company itself has not been buying at these prices. The filings disclose the actual execution levels of its two most recent accelerated share repurchase agreements:
- April 2025 ASRs — $500.0M across two financial institutions; initial delivery ~5.2M shares; final settlement September 2025 delivering 317 thousand additional shares at "a weighted-average share price of $91.00 for the transaction period" (Q3 FY2026 10-Q).
- March 2026 ASR — $200.0M with one institution on 2026-03-11; initial delivery of ~685 thousand shares representing 75% of the prepayment divided by the closing price, which implies a closing price of approximately $219 on 2026-03-11.
$91 and ~$219. Today's price is $317.74. The May 2024 board authorisation of an additional $10.0B remains in force with no termination date, so capacity is not the constraint — but the two most recent structured programmes were struck at 29% and 69% of today's price, and the share count is now falling at only about 2%–2.5% a year (1.358B diluted in FY2023 to 1.261B in FY2026, −7.1% over three years).
Dividend. The FY2025 10-K states "During fiscal year 2025, our quarterly dividend declared was $0.23 per share"; the board has declared quarterly dividends since April 2014. The vendor's trailing dividend per share of $1.04 implies roughly $0.26 a quarter in FY2026, consistent with dividends paid of $1.271B on ~1.26B shares (~$1.01 paid). The FY2026 declared rate is not disclosed in any filing in this folder; we use the vendor's $1.04 and label it. Payout ratio 17.5%. Yield 0.33%.
6. Valuation — priced in or room?
At $317.74 (market cap $397.36B, EV $395.51B, 1.2506B shares):
| Trailing FY2026 | FY2027E | FY2028E | FY2029E | |
|---|---|---|---|---|
| Revenue | $23.233B | $34.573B (20 analysts) | $40.996B (19) | $45.373B (13) |
| Revenue growth | +26.0% | +48.8% | +18.6% | +10.7% |
| Diluted EPS | $5.76 | $9.38 (20 analysts, range $7.64–$10.36) | $11.77 (16, range $9.98–$12.70) | $13.51 (9) |
| P/E | 55.2x | 33.9x | 27.0x | 23.5x |
| EV/sales | 17.0x | 11.4x | 9.6x | 8.7x |
| Price to book | 32.0x | — | — | — |
| ROIC / ROE | 41.0% / 67.0% | — | — | — |
Basis note. The FY2026 estimate row ($23.196B revenue, $5.695 EPS) came within 0.2% and 1.1% of the GAAP actuals ($23.233B, $5.76). For Lam, consensus EPS is effectively on a GAAP basis — the company carries little acquisition amortisation — so the forward multiples above are comparable to the trailing one without adjustment. This is not true of most names and is worth stating.
Peer context. Within the batch, street upside is: ASML +34.7% ($1,711.89 to $2,305.75), LRCX +16.9%, AMAT +15.9% ($546.62 to $633.54), KLAC +15.2% ($195.45 to $225.18). The four WFE names are one trade with four expressions: ASML owns lithography outright and carries the largest street discount; KLAC is process control and the highest-margin, most foundry-levered; AMAT is the broadest and most diversified; LRCX is the memory-levered one, which is why it moved most today. No peer earnings or multiples are supplied in this file, so no peer-multiple comparison is drawn — only the price-target gaps above, which are the street's, not ours.
Ratings. 39 buy, 10 hold, 1 sell — consensus Buy. Targets: high $500, low $290, median $355, consensus $371.29. The vendor's own quality model rates LRCX B, overall score 3, with return-on-equity and return-on-assets sub-scores of 5 (the maximum in that scale) and price-to-earnings and price-to-book sub-scores of 1 (the minimum) — the model's own summary of this dive: excellent business, expensive stock.
6a. What today's price assumes (the inversion)
At $317.74 — 55.2x trailing, 33.9x FY2027E, 27.0x FY2028E — the price embeds roughly the following falsifiable claims:
- Revenue compounds from $23.233B (FY2026 actual) to $34.573B in FY2027 and $40.996B in FY2028 — +48.8% then +18.6%. (Consensus-derived, 20 and 19 analysts.) The FY2027 range is $29.329B to $37.757B, an $8.4B spread, so even the consensus is not confident.
- The September 2026 quarter lands at approximately $8.085B of revenue and $2.16 of EPS — a 20.3% sequential step on the June quarter's $6.722B and +51.9% year over year. (Consensus, last updated 2026-08-04, i.e. anchored on management's guidance issued 2026-07-29 — HALF-WEIGHT, management framing.) This is the largest single implied step in the entire estimate series and the first thing the 2026-10-21 print will confirm or break.
- Gross margin holds at or above 50%. (FY2026 actual 50.5%; June quarter 51.7%; our number, from the reported series.) The 10-Q already names "higher tariff-related spend" as a partial offset to factory efficiencies.
- The market still pays roughly 27x FY2028 earnings in mid-2027 — i.e. no cyclical de-rating as earnings approach their peak. (Our derivation from the consensus EPS path at a constant price.) This is the most fragile assumption in the price. Wafer-fab equipment historically de-rates into peak earnings; the same company's revenue fell 14.5% in FY2024 and its trailing multiple twelve months ago was 23.2x.
- China, at approximately 37% of nine-month FY2026 revenue, is not further restricted. (10-Q disclosure; the risk factor is the 10-K's own.) No part of the consensus revenue path survives a material tightening.
6b. The return bridge (why the multiple moves)
Expected return over the dive's horizon decomposes as: EPS growth (very large) + multiple drift (large COMPRESSION) + shareholder yield (1.29%).
Consensus EPS compounds from $5.76 (FY2026 actual) to $11.77 (FY2028E) — a 42.9% annualised rate. Our base fair value of $353 is 30x FY2028E, against 55.2x trailing today. So the base case explicitly assumes substantial multiple compression, and the entire modest +11.1% expected return is what survives after handing most of a doubling of earnings back through a de-rating.
State it as plainly as possible: at 55x trailing, earnings can double and the stock can go up 11%. That is the arithmetic of buying a cyclical after it has been re-rated. The reason we assume compression rather than a hold is threefold — growth maturation (consensus itself decelerates to +10.7% revenue by FY2029), risk normalisation (a 12.1% effective tax rate and a 37% China mix are both fragile inputs to the forward EPS), and cycle position (equipment multiples contract as customer capital budgets peak).
The bull case at $430 does not require multiple expansion either — it is 36.5x FY2028E, roughly a hold of the current forward multiple, on delivered earnings. The bear case at $205 is 17.4x FY2028E, which is what a normal cyclical de-rating looks like. None of the three scenarios needs the market to pay more than it does today. That is a coherent way to be constructive on a business and neutral on its stock.
6c. Variant perception (where we differ, what would surprise)
On valuation we have essentially NO variant perception, and we say so because the house rule makes that a verdict input. Our base of $353 is 30x FY2028E; the street's $371.29 is 31.5x the same number; the street median of $355 is 30.2x. We and the crowd are looking at the same estimates and applying almost the same multiple. A no-edge name defaults toward Watch, and this one does. Where we differ is on four narrower points, all filing-sourced:
- We think the deferred-revenue decline is under-weighted. The balance has fallen from $2.77B (September 2025) to $2.22B (March 2026), −20%, with the 10-Qs attributing it to "a decrease in customer down payments." Consensus models FY2027 revenue +48.8%. Those two facts sit awkwardly together. Watchable number: the deferred-revenue balance in the FY2026 10-K when it is filed, and in the Q1 FY2027 10-Q after 2026-10-21. A rebuild above $2.7B would resolve it in the bulls' favour and would be a genuine positive surprise.
- We think the memory-conversion lag is under-modelled in BOTH directions. Micron's quarterly revenue rose 346% over five quarters; Lam's rose 42.4%; and the 10-Q says the nine-month increase was predominantly Foundry. The bear reading: the memory capex wave is still mostly narrative in Lam's actual numbers, and the equity has already been paid for it. The bull reading: the conversion has barely begun, Lam is the most memory-levered of the four names, and the high end of the FY2028 estimate range ($12.70) may prove low. Watchable disclosure: the market-mix percentage table in the FY2026 10-K — not extractable from these filings, and the single most informative number Lam publishes.
- We think the market still credits Lam with a capital-return story that no longer exists at this price. $5.122B returned in FY2026 is a 1.29% shareholder yield; the company's two most recent accelerated repurchases executed at approximately $91 and $219 a share; the diluted count is shrinking at only ~2.4% a year. Watchable number: repurchase dollars in the FY2027 quarters. If management materially slows the buyback at $318, that is a valuation signal from the best-informed party.
- We think China at 37% and rising is a larger single-point risk than a 27x FY2028 multiple accommodates. The trajectory — 26% (FY2023), 42% (FY2024), 34% (FY2025), 39% (six months to December 2025), 37% (nine months to March 2026) — is not a settled exposure. Watchable event: any change to US export licence requirements. This is the negative surprise most capable of breaking the thesis in a single headline.
Synthos fair values
All three anchors are multiples of consensus FY2028E diluted EPS of $11.77 (16 analysts, range $9.98–$12.70), cross-checked against FY2027E EPS of $9.38 and against the technical structure.
- Bear ~$205 — 17.4x FY2028E, cross-checked at 21.9x FY2027E. The scenario: the memory capex wave arrives smaller or later than modelled, China tightens, FY2027 lands near the low end of its $7.64 estimate range, and the multiple compresses to a normal cyclical level. Cross-check: $205 is 15.6% below the 200-day moving average of $242.72 — a full cycle unwind, which this name did as recently as FY2024. −35.5%.
- Base ~$353 — 30.0x FY2028E, cross-checked at 37.6x FY2027E and 26.1x FY2029E. The scenario: consensus is broadly delivered, gross margin holds above 50%, China is constrained but not cut off, and the market pays a premium-but-not-peak multiple for an oligopolist with 41% return on invested capital. Cross-check: $353 is fractionally below the street median of $355 and 4.9% below the consensus mean of $371.29. +11.1%, plus a 1.29% shareholder yield.
- Bull ~$430 — 36.5x FY2028E, cross-checked at 31.8x FY2029E. The scenario: the memory conversion runs above the high end of the estimate range, the CSBG installed-base annuity keeps rising as a share of revenue, and the market treats WFE as structural growth rather than cycle. Cross-check: $430 is a return to the 52-week closing high of $433.33 — i.e. the bull case is the market re-paying what it paid two months ago. +35.3%.
Base is 11.1% above spot; the asymmetry is 1.00:1 (35.5% down, 35.3% up). An even bet on a 1.8-beta cyclical, six days after a print and on the day of an 8% gap, with the base below the street median. That is a Watch, not a Buy.
7. Knowledge base — a thin lane, and the thinness is the finding
Raw hits on the direct terms (LRCX, Lam Research, WFE, wafer fab equipment): 10. Used: 8. Discarded: 2. Both discards are intra-document duplicates — the same claim restated twice within one source document (compound_and_friends TLMfVxCP5-U on 2026-01-02, and MV76MRQndTg on 2026-02-27). No name-collision problem exists for this ticker: "LRCX" and "Lam Research" are unambiguous strings, unlike the acronym collisions that plague other names in this batch.
Of the 8 used, 3 carry the thinker: LRCX_mgmt tag — the company's own earnings-release voice, weighted LOW under the half-weight management rule and excluded from the conviction score. That leaves 5 independent-voice claims across 3 sources — compound_and_friends, brent_johnson and invest_like_the_best — for a $397 billion company. For comparison, the strongest lane in this batch of dives runs to six claims across two sources with three years of continuity and a specific dated valuation call. This lane has none of that. A thin lane on a mega-cap is a finding: the panel has not underwritten this name.
Note on attribution. Eight of the ten hits carry a null speaker field, so they are attributable only to the source document, not to a named individual. Where a speaker is named we attribute to the speaker and note the channel separately. This is stated rather than smoothed over.
The independent lane, verbatim:
> 2026-01-02 · bullish · conviction 55 · entities: Lam Research, LRCX · source: compound_and_friends
> "Best S&P performer of early 2026 and a best-stocks-of-2025 write-up; the winning trend continues into the new year."
> 2026-01-11 · BEARISH · conviction 65 · entities: Lam Research, LRCX · source: brent_johnson (skill weight 0.7)
> "Lam Research is stretched — up ~30% in three weeks with high stochastics; due for a pullback."
> 2026-02-27 · bullish · conviction 70 · entity: Lam Research · source: compound_and_friends
> "Semiconductor equipment is the best spot in semis — the one area without parabolic blowoffs, unlike hyperscalers (2024) or memory (2025)."
> 2026-03-03 · bullish · conviction 70 · entities: Applied Materials, Lam Research, KLA · source: compound_and_friends
> "Semi-cap went from ~20 firms to three giants via acquisition; illustrates how every sector consolidates."
> 2026-04-29 · NEUTRAL · conviction 50 · entity: LRCX · speaker: Colin · channel: invest_like_the_best
> "Lam Research got a little frothy but is a good long-term hold; won't sell despite its epic run."
The management lane (LOW weight, reported for completeness, all 2026-04-22, all from the March-quarter earnings release):
> "AI-driven demand is reshaping the semiconductor industry; Lam's investments and execution velocity are generating momentum and driving outperformance during this critical growth phase." (bullish, conviction 78)
> "Lam expects to keep enabling customers' AI roadmaps with leading-edge deposition and etch equipment as nearly every advanced chip is built with Lam technology." (bullish, conviction 72)
> "Management is guiding the June 2026 quarter... signaling continued forward momentum off record March-quarter results." (bullish, conviction 70)
These three are promoter-class by construction — a company describing its own prospects in its own press release — and they carry no weight in the conviction rating. They are recorded because omitting them would misstate the raw count.
Supplementary sector sweep (semiconductor equipment, semi-cap, etch, deposition): 7 raw hits, 2 used as sector-level colour, 5 discarded. The discards: two are duplicates of claims already in the direct set; one is a false positive where "semi-cap" appears only as an analogy inside a claim about cybersecurity software consolidation; one is about Intel and a fab joint venture, not about equipment demand; one is about Japan-China trade with no Lam relevance. The two used are sector-level and are explicitly NOT name-level conviction:
> 2026-05-28 · neutral · conviction 70 · speaker: Dan Loeb · channel: invest_like_the_best
> "Even when semiconductor, semi-cap and memory fundamentals are genuinely super-strong, stocks fall when expectations run too high — as seen when Micron reported up 80% and the stock still went down."
This is the single most relevant claim in the entire knowledge base for today's decision, and it is a caution. It is attributed to a named speaker, it is recent, it is precisely about the semi-cap and memory complex, and it describes exactly the risk this dive identifies: strong fundamentals are not sufficient when the multiple has already moved. It is sector-level, not a call on LRCX, and is labelled as such.
What the lane establishes, honestly. The independent voices are constructive on the sector and cautious on the price. Two of the five are explicit warnings on stretch or froth (brent_johnson, January 2026, at a lower price; and the neutral "got a little frothy" claim from April 2026). One is a structural argument about oligopoly. One is momentum-following. One — the February 2026 claim that semi-cap was "the one area without the parabolic blowoff moves seen elsewhere" — has arguably been invalidated by the subsequent price action: the stock is up 33.8% in six months and 229.7% in twelve. Net conviction: positive-low. The direction of the sector view is favourable; the name-level underwriting is absent; and the most on-point recent claim is a warning.
8. Data integrity — what we rejected from the vendor file and why
Nine items. This file is unusually clean by the standards of this exercise, and the clean parts are stated as explicitly as the broken ones.
1. est net-income and EBIT/EBITDA rows are internally impossible from FY2027 onward — REJECTED; all forward valuation runs on epsAvg. In every estimate row from FY2027 to FY2031, niAvg EXCEEDS ebitAvg: FY2027 $11.827B vs $11.062B; FY2028 $14.225B vs $13.117B; FY2029 $16.717B vs $14.518B; FY2030 $17.782B vs $15.398B; FY2031 $21.817B vs $17.837B. Net income cannot exceed operating profit at a positive tax rate for a company with net interest income of this size. The FY2026 row is separately understated against the actual: ebitAvg $7.422B against reported operating income of $8.200B, and ebitdaAvg $7.900B against a reported operating income plus D&A of $8.642B. No EBITDA or EBIT estimate appears anywhere in this dive; epsAvg is used throughout, and it is validated below.
2. epsAvg is validated against actuals and is on a near-GAAP basis — ACCEPTED, with the basis stated. The FY2026 estimate row ($23.196B revenue, $5.69531 EPS) came within 0.2% and 1.1% of the reported GAAP result ($23.233B, $5.76). Lam carries minimal acquisition amortisation, so GAAP and adjusted earnings are close. Forward multiples in Section 6 are therefore directly comparable to the trailing GAAP multiple, which is not true of most names.
3. Vendor seg_prod is labelled as segment data but Lam reports ONE segment — a labelling error, not a numbers error. The 10-K states: "The Company operates in one reportable business segment: manufacturing and servicing of wafer processing semiconductor manufacturing equipment." The vendor's "System" and "Customer Support and Other" categories are the Note 4 revenue disaggregation, not reportable segments. The numbers are correct and tie exactly — FY2025 $11.491B + $6.944B = $18.436B, precisely reported revenue, and the same holds for every year back to FY2020. We use them, relabelled as revenue disaggregation. Separately, FY2026 product-line data is MISSING (latest year is FY2025) and is stated as a gap.
4. seg_geo ties to the 10-K and is USED — but it is a year stale and the 10-Qs supersede it. FY2025 geographic revenue sums to $18.436B, exactly reported revenue; the implied China share of 33.66% matches the 10-K's stated "approximately 34%", FY2024's 42.2% matches "42%", FY2023's 25.6% matches "26%", and the implied non-US share of 92.5% matches "approximately 93%". This is a correct geographic file, which is worth recording. However seg_geo stops at FY2025, and the 10-Qs disclose China at approximately 39% for the six months ended 2025-12-28 and approximately 37% for the nine months ended 2026-03-29. The vendor implies China is 34%; the 10-Q shows 37%; we use 37%.
5. The tech block and the quote block disagree on the 52-week range — reconciled, and the CLOSING-basis figures are used. quote.yearHigh is $438.50 and quote.yearLow is $94.11; tech.hi52 is $433.33 and tech.lo52 is $95.94. The gap is the difference between intraday extremes (quote) and highest/lowest daily closes (tech, computed from six years of closes per the data contract). We use the tech block's closing-basis $433.33 / $95.94 throughout, because the moving averages, RSI, MACD and drawdown in the same block are computed from the same closing series, and mixing bases would corrupt the drawdown figure. On the closing basis the drawdown from peak is −26.7%; on the intraday basis it would be −27.5%. Both are stated so neither is hidden. The moving averages disagree trivially and immaterially (quote.priceAvg50 $337.923 vs tech.sma50 $338.283; priceAvg200 $241.074 vs sma200 $242.720); the tech block is used.
6. daysOfPayablesOutstandingTTM is 0 and payablesTurnoverTTM is 0 — arithmetically impossible; the cash conversion cycle is REJECTED. The vendor reports averagePayablesTTM of $535.8M, so payables exist; a turnover of zero is a computation failure. Consequently cashConversionCycleTTM of 219.5 days is simply operatingCycleTTM with the payables leg omitted. Rebuilt with FY2026 cost of goods sold of $11.508B, payables days would be roughly 17, giving a cycle nearer 202 days — but $535.8M of average payables against $11.5B of COGS is itself implausibly low for a hardware manufacturer of this scale, so we reject the payables data outright and quote only days of inventory (135.6) and days of sales outstanding (83.9), which are internally consistent with the balance sheet.
7. priceToFairValueTTM is not a fair-value estimate — REJECTED. It reports 31.977494, identical to priceToBookRatioTTM to six decimal places. It is book value relabelled, and carries no valuation information. It is not used and no reader should read it as a fair-value signal.
8. Goodwill is reported as ZERO for FY2026 — a mapping error, though the aggregate is preserved. The vendor shows FY2025 goodwill $1.626B and intangibles $182.2M (total $1.808B), then FY2026 goodwill $0 and intangibles $1.896B (total $1.896B). The FY2025 10-K has a dedicated goodwill and intangible assets note and discusses tax-deductible portions of "the goodwill balance", so Lam plainly carries goodwill. The vendor has folded goodwill into intangibles for FY2026. The aggregate is preserved, so tangible book value per share of $8.43 is unaffected and is used; but the goodwill line for FY2026 is rejected and no goodwill-specific conclusion is drawn.
9. The 2026-07-29 8-K contains the cover page ONLY — Exhibit 99.1, the earnings release with the financial statements, is not in the extracted text. The freshest hard data in the folder is therefore not readable from the 8-K itself. However, the vendor payload IS current: inc_q carries the quarter ended 2026-06-28 (revenue $6.722B, diluted EPS $1.81), inc_a carries FY2026 in full ($23.233B, $5.76) and the four quarters sum to the year to within $1M, and earn_cal records the 2026-07-29 print with epsActual 1.82 and revenueActual $6,722,238,000, lastUpdated 2026-08-04. We use the vendor's June-quarter data and state that it could not be independently corroborated against the 8-K text. The corollary is important: the FY2026 balance sheet and cash-flow statement in this file (dated 2026-06-28) come from an unaudited earnings release, and the FY2026 10-K has not yet been filed — the 10-K in this folder is FY2025, filed 2025-08-11. FY2026 balance-sheet and cash-flow figures are therefore uncorroborated by any filing here, and are labelled as such wherever used.
Two further minor discrepancies, flagged not rejected. (a) earn_cal EPS actuals differ from income-statement GAAP diluted EPS by one to two cents in three of five quarters ($1.33 vs $1.35, $1.27 vs $1.26, $1.82 vs $1.81) — a basis difference of no consequence, but the income-statement figures are used for all trailing multiples. (b) profile.fullTimeEmployees reports 19,000; the Q3 FY2026 10-Q states headcount was approximately 20,600 at 2026-03-29. The filing wins. (c) profile.lastDividend of $1.04 is an annualised figure, not the quarterly declared rate — the FY2025 10-K discloses $0.23 per share per quarter for FY2025 — and the FY2026 declared quarterly rate is not disclosed in any filing in this folder.
Not defects, correctly reported: enterpriseValueTTM of $395.507B rebuilds independently to $395.512B (Section 4); capexToDepreciationTTM of 2.189 ties to $966.4M of capex against $441.5M of D&A; freeCashFlowYieldTTM of 1.231% ties to $4.891B of free cash flow on a $397.36B market capitalisation; netDebtToEBITDATTM of −0.212 correctly signs the net cash position; and interestBurdenTTM of 1.0 with ebtPerEbitTTM of 1.008 correctly reflects interest income exceeding interest expense.
Non-equity tripwire: CHECKED AND PASSED. LRCX is common stock on NASDAQ, IPO 1984-05-04, isEtf false, isFund false, isAdr false. Price $317.74 with a 52-week range of $94.11 to $438.50 — a 366% band, which no preferred or baby bond exhibits. Beta 1.805. Volume 14.73M shares. The dividend has been raised (the 10-K's $0.23 quarterly for FY2025 against a trailing $1.04 annual) rather than fixed. This is not a preferred, a baby bond or a tracking instrument inheriting parent fundamentals.
9. Technicals — a bounce inside a correction
- Price $317.74. 26.7% below the 52-week closing high of $433.33; 231.2% above the 52-week closing low of $95.94. The maximum drawdown from peak over the trailing year is −26.7%, i.e. the stock is at or near its worst point since the peak.
- BELOW the 50-day moving average of $338.28 by 6.1%. ABOVE the 200-day moving average of $242.72 by 30.9%. That combination — under the short average, well over the long one — is the signature of a correction inside an uptrend, not a breakdown and not a breakout.
- RSI 45.3. MACD −14.22, deeply negative. Note that RSI is still below 50 after a 7.85% day, which is a measure of how damaged the momentum structure was going in.
- Relative performance: 3-month +22.9% vs SPY +7.6% and QQQ +7.7%; 6-month +33.8% vs SPY +11.1% and QQQ +15.7%; 12-month +229.7% vs SPY +24.3% and QQQ +30.8%. This is one of the great twelve-month runs in the large-cap market, and it is the reason for caution, not the reason for enthusiasm.
Today's move and what it does to the entry
The facts. LRCX opened at $308.83 against a previous close of $294.61 — a 4.8% opening gap — traded a range of $306.48 to $320.20, and closed at $317.74, near the top of the day's range, +7.85% (+$23.13) on 14.73M shares. It was the largest move of the four WFE names (KLAC +6.95%, AMAT +5.48%, ASML +4.22%) and almost exactly matched Micron's +7.62%. That ranking is itself evidence for the thesis: the market sorted the equipment names by memory leverage, and put Lam at the top.
What is genuinely different about this gap, and it matters. Most of the names that moved violently today were gapping toward or through their highs. LRCX was not. Even after 7.85%, the stock is 6.1% below its 50-day average and 26.7% below its 52-week high. The insiders who sold in June and July sold at $335 to $390 — above today's close. The company's own March 2026 accelerated repurchase executed at roughly $219. So the honest characterisation is: this is a recovery bounce off a 27% correction, not a chase into a new high. That is a materially better setup than a name printing all-time highs on an 11% day, and it deserves to be said.
But it is not yet an entry, for four reasons. First, the 50-day average at $338.28 has not been reclaimed — that is the first genuine technical repair and it is 6.4% away. Second, MACD at −14.22 is nowhere near a crossover; a single day does not fix that. Third, we are being asked to pay 8% more than yesterday for the same disclosures — nothing about Lam changed today; the sector's re-rating did. Fourth, and decisively, the valuation work independently lands at $353 against $317.74 — an 11% base return with a 35% bear. Paying up 8% into an even-money proposition, in the middle of a sector-wide up-day, is the definition of a poor entry.
Practical read. There are two entries we would take and neither is today. (a) A reclaim of the 50-day average at $338.28 that holds for several sessions with MACD turning positive — that costs 6.4% more than today's close and buys confirmation that the correction is over; it converts this to Buy — Tactical. (b) A retest that holds the $270–$295 band — the pre-gap level was $294.61 — which would restore the asymmetry to roughly 1.5:1 and would convert this to a staged Buy — Tactical without needing any confirmation at all. Below the 200-day average at $242.72, the cyclical bear case is in play and the position should be reassessed from scratch rather than averaged into. There is no print for 78 days, so there is no cost to waiting.
10. Insiders — eight transactions, every one a disposition, all above today's price
| Date (txn) | Person | Role | Action | Shares | Price | Holding after |
|---|---|---|---|---|---|---|
| 2026-07-13 | Abhijit Y. Talwalkar | Director | Sale | 18,282 | $335.00 | 87,142 |
| 2026-07-02 | Timothy Archer | President & CEO | Option exercise (M-Exempt) | 30,000 | $30.033 | 1,105,966 |
| 2026-07-02 | Timothy Archer | President & CEO | Sale | 30,000 | $390.01 | 1,075,966 |
| 2026-06-12 | Eric Brandt | Director | Sale | 3,828 | $359.08 | 192,584 |
| 2026-06-12 | Eric Brandt | Director | Sale | 1,100 | $357.98 | 196,412 |
| 2026-06-12 | Eric Brandt | Director | Sale | 1,393 | $357.24 | 197,512 |
| 2026-06-12 | Eric Brandt | Director | Sale | 300 | $355.71 | 198,905 |
| — |
Every transaction in the file is a disposition or the option exercise funding one. There is not a single open-market purchase.
The reading, stated carefully and without over-claiming. The chief executive's transaction is an exercise-and-sell: options struck at $30.033 exercised and the resulting 30,000 shares sold at $390.01, a 13x spread. That is the textbook profile of a scheduled trading-plan transaction, and it left him holding 1,075,966 shares — a 2.7% reduction. Eric Brandt sold 6,621 shares in four tranches, 3.3% of his holding. Only Abhijit Talwalkar's sale is proportionally meaningful at 17.3% of his position.
So this is not a stampede. What it is, is uniformly one-directional at prices that are all above today's close. The realised prices — $335.00, $355.71, $357.24, $357.98, $359.08, $390.01 — bracket a range 5.4% to 22.7% above $317.74. Combined with the company's own repurchase executions at approximately $91 and approximately $219, the best-informed parties to this security have been sellers into the $335–$390 zone and buyers only far below the current price. That is a modest negative and it is consistent with the valuation conclusion rather than independent of it.
11. Verdict, kill-criteria and flip conditions
Watch.
What is genuinely excellent. Revenue of $23.233B, up 26.0%, with eight consecutive sequential quarterly increases and a June quarter up 30.0% year over year. Gross margin at 50.5% for the year and 51.7% in the quarter, expanding for three straight years on factory efficiency and mix. Operating margin 35.3%. Return on invested capital 41.0%; return on equity 67.0%. Net cash of $1.845B, interest coverage of 66x, no commercial paper drawn against a $2.00B programme. Five consecutive earnings beats. A 37.7%-of-revenue installed-base service business that rises when systems fall. An oligopoly position in deposition and etch, in an industry whose technology roadmap adds more deposition and etch steps every generation. And a genuine, filing-visible memory linkage into the largest memory capital cycle on record.
What we are not willing to pay for it today. 55.2x trailing and 27.0x FY2028 consensus earnings, after a re-rating from 23.2x to 55.2x in twelve months that supplied roughly two-thirds of a 229.7% return. Our base fair value of $353 is 11.1% above spot and below the street median of $355 — we have no valuation edge, and the house rule says a no-edge name defaults toward Watch. The bear at $205 and the bull at $430 are symmetric at ±35% on a 1.805-beta cyclical. Deferred revenue has fallen 20% across two quarters on declining customer down payments. China is approximately 37% of nine-month FY2026 revenue and rising, with a dedicated export-control risk factor. Two customers are 32% of revenue. The nine-month FY2026 growth was, per the 10-Q, predominantly Foundry rather than memory — the memory conversion the price assumes has barely started to appear in the reported numbers. Every insider transaction is a sale, all above spot. The shareholder yield is 1.29%. The stock is below its 50-day average after an 8% day, with MACD at −14.22.
On today's move specifically: we are not asking the reader to buy an 8% gap, and we do not think they should. The mitigation — that this is a bounce inside a 27% correction rather than a chase into new highs — is real and is why this is a Watch and not a Hold or an Avoid. But the correct response to a violent sector up-day on a name whose base case is +11% with a −35% bear is to write down the entry levels and wait for one of them.
Pre-registered KILL criteria — what would take this to Avoid:
- A material tightening of US export licence requirements affecting China. At approximately 37% of nine-month revenue this is the single event capable of invalidating the entire consensus revenue path in one headline.
- The 2026-10-21 print missing the $8.085B revenue consensus, particularly if accompanied by December-quarter guidance below the June quarter's $6.722B. A sequential decline would signal the up-leg has already peaked.
- Deferred revenue falling below $2.0B, continuing the $2.77B → $2.22B decline. Customer down payments are the earliest hard order signal Lam discloses.
- Gross margin falling back below 49%, which would indicate the mix and factory-efficiency gains were cycle artefacts rather than structural.
- A decisive break below the 200-day moving average of $242.72 on company or policy news rather than sector beta.
- The buyback being materially slowed while the stock trades above $300 — management declining to buy at a price it declined to buy at $219 and $91 would be an informative signal.
Pre-registered FLIP conditions — what would take this to Buy — Tactical:
- A reclaim of the 50-day moving average at $338.28, held for several sessions, with MACD crossing positive. Paying 6.4% up from here for confirmation that the correction is over is a fair trade on a name with this much beta.
- A retest that holds the $270–$295 band (the pre-gap level was $294.61). At $285 the base of $353 is +23.9% against a bear of −28.1% — a 1.5:1 asymmetry, which is a different proposition entirely. This is the preferred entry.
- The 2026-10-21 print delivering approximately $8.085B of revenue AND a rebuilding deferred-revenue balance. Revenue alone is not sufficient; the order-book signal is what is currently missing.
- The FY2026 10-K disclosing a materially higher memory share of the market mix than FY2025, which would confirm the conversion the price assumes.
- Independent, non-management knowledge-base claims initiating name-level coverage. The current lane is five independent claims across three sources, two of which are cautions. Conviction is the missing input, not analysis.
Where LRCX fits in the Synthos Framework Portfolio. The semiconductor capital-equipment sleeve, currently at 0% — a watch-list position with defined entry levels, not a holding. On a flip to Buy — Tactical the suggested size is 1.5–2%, built in three tranches rather than one, because the 1.805 beta and the 366% 52-week band mean position sizing does more work here than entry timing. On overlap: the four WFE names are one trade with four expressions and should not be held as four positions. Within the group, LRCX is the highest-beta, most memory-levered and — on the street's own numbers — mid-pack on upside (+16.9%, against ASML's +34.7%, AMAT's +15.9% and KLAC's +15.2%). An investor who wants the memory capital cycle already owns it more directly through the memory makers themselves; an investor who wants equipment exposure with less memory beta has three other choices in this batch. LRCX's distinct claim is that it is the most levered equipment expression of the memory cycle — which is exactly why the entry price matters more here than for its peers. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $317.74.
Single biggest risk: a cyclical de-rating as earnings peak. At 27.0x FY2028 consensus earnings the price assumes the market continues to pay a growth multiple for a capital-equipment maker at the top of a spending cycle. History in this very file argues otherwise: revenue fell 14.5% in FY2024, and the trailing multiple twelve months ago was 23.2x. If FY2028 consensus EPS of $11.77 is delivered in full and the market applies a 20x mid-cycle multiple, the stock is $235 — 26% below today's price, with the earnings having doubled. That is the specific way an investor loses money owning an excellent business at this price, and it requires no bad news at all.
Most fragile assumption in the price: that the September 2026 quarter lands at approximately $8.085B of revenue — a 20.3% sequential step. Everything downstream depends on it. It is the largest implied step in the estimate series, it is anchored on management's own guidance issued six days ago and therefore half-weighted, and it sits awkwardly beside a deferred-revenue balance that has fallen 20% in two quarters and a 10-Q that describes nine-month growth as predominantly Foundry-driven. If that quarter lands, the flip conditions above are met and this becomes a straightforward Buy — Tactical. If it does not, $205 is not a pessimistic number.
Provenance & disclosures
- Traceability: 10 raw knowledge-base hits on the direct terms (
LRCX,Lam Research,WFE,wafer fab equipment), 8 used, 2 discarded as intra-document duplicates (compound_and_friends documentsTLMfVxCP5-UandMV76MRQndTg, each restating one claim twice). Of the 8 used, 3 carrythinker: LRCX_mgmt— the company's own earnings-release voice, weighted LOW and excluded from the conviction rating — leaving 5 independent-voice claims across 3 sources (compound_and_friends ×3, brent_johnson ×1 at skill weight 0.7, invest_like_the_best ×1). A supplementary sector sweep (semiconductor equipment,semi-cap,etch,deposition) returned 7 raw hits, 2 used as clearly-labelled SECTOR-level colour, 5 discarded — two duplicates of the direct set, one false positive where "semi-cap" appears only as an analogy inside a claim about cybersecurity consolidation, one about Intel and a fab joint venture, one about Japan-China trade. Sector-level claims are NOT name-level conviction and are labelled as such wherever quoted. Eight of the ten direct hits carry a null speaker field and are attributable only to the source document; where a speaker is named (Colin; Dan Loeb) we attribute to the speaker and note the channel separately. Net stance: positive-low. All quotes verbatim from stored claim text. - Data as-of: income statement 2026-06-28 (FY2026 Q4 and full year, released 2026-07-29, six days before this dive) · balance sheet and cash flow 2026-06-28, unaudited, from the earnings release; the FY2026 10-K has not been filed · segment and geographic data 2025-06-29 (FY2025) · estimates 2026-08-04 · prices 2026-08-04 (quote timestamp 1785873601 = 2026-08-04T20:00:01Z, the 16:00 ET close; 50-DMA $338.28, 200-DMA $242.72, RSI 45.3, MACD −14.22) · knowledge-base claims 2026-08-04. Filings used: 10-K filed 2025-08-11 (FY2025, ended 2025-06-29); 10-Q filed 2026-01-29 (Q2 FY2026); 10-Q filed 2026-04-23 (Q3 FY2026); 8-K 2026-02-03 (director appointment and COO transition); 8-K 2026-04-22 and 8-K 2026-07-29 (earnings releases — cover pages only; Exhibit 99.1 is not in the extracted text).
- Filing-versus-vendor corrections made (detailed in Section 8): China revenue share — the vendor's
seg_geoimplies 34% (FY2025); the 10-Q shows approximately 37% for the nine months ended 2026-03-29 and 39% for the six months ended 2025-12-28; we use 37%. Headcount — the vendor reports 19,000; the Q3 FY2026 10-Q shows approximately 20,600 at 2026-03-29; we use 20,600. Segment structure — the vendor labels "System" and "Customer Support and Other" as segments; the 10-K states Lam has ONE reportable segment and these are Note 4 revenue-disaggregation categories; we use them, relabelled (the figures themselves tie exactly to reported revenue in every year). Goodwill — the vendor reports $0 for FY2026 having reported $1.626B for FY2025; the 10-K's goodwill note confirms goodwill exists; the vendor has folded it into intangibles, the aggregate is preserved, and the FY2026 goodwill line is rejected. - Data rejections:
estniAvg,ebitAvgandebitdaAvgrows rejected for FY2027–FY2031 (net income exceeds EBIT in every year) and for FY2026 (EBIT $7.422B against reported operating income of $8.200B) — all forward valuation runs onepsAvg, which is validated against the FY2026 actual to within 1.1%;cashConversionCycleTTM(219.5 days) rejected becausepayablesTurnoverTTManddaysOfPayablesOutstandingTTMare both zero;priceToFairValueTTM(31.977494) rejected as identical topriceToBookRatioTTMand containing no valuation information; FY2026 goodwill rejected; FY2026seg_prodandseg_geomissing and stated as gaps; the memory/foundry/logic market-mix percentage table not extractable from any filing in this folder and stated as a gap rather than estimated. - Basis notes: consensus EPS for Lam is effectively GAAP — the FY2026 estimate came within 1.1% of the reported GAAP figure — so trailing and forward multiples in Section 6 are directly comparable. The
techblock's closing-basis 52-week range ($433.33 / $95.94) is used rather than thequoteblock's intraday range ($438.50 / $94.11), because the moving averages, RSI, MACD and drawdown are computed from the same closing series; drawdown from peak is −26.7% on the basis used and would be −27.5% on the intraday basis. FY2026 balance-sheet and cash-flow figures come from an unaudited earnings release and are uncorroborated by any filing in this folder. - Estimate coverage: 20 analysts on FY2027 revenue and EPS, 19 and 16 on FY2028, 13 and 9 on FY2029, dropping to 5 and 4 by FY2031. The FY2027 EPS range is $7.64 to $10.36 and the FY2028 range is $9.98 to $12.70 — spreads of 26% and 23% of the midpoint, so consensus itself is not confident. Fair values are anchored on the FY2028 mean of $11.77 (16 analysts) with the FY2027 mean of $9.38 (20 analysts) as cross-check; no conclusion rests on the FY2029–FY2031 rows.
- Peer note: the vendor peer set (AMAT, ANET, ARM, INTC, INTU, KLAC, MU, QCOM, TXN, UBER) is only partly relevant — AMAT and KLAC are genuine WFE comparables and MU is the key customer-side read-across; ANET, INTU, QCOM, ARM, TXN and UBER are not comparable to a capital-equipment maker. No peer earnings or multiples are supplied in this file, so no peer-multiple comparison is drawn; the only cross-name figures used are the batch price-target gaps (ASML +34.7%, LRCX +16.9%, AMAT +15.9%, KLAC +15.2%) and today's closing moves.
- Half-weight management framing: the three
LRCX_mgmtknowledge-base claims, the September-quarter revenue and EPS consensus of $8.085B / $2.16 (updated 2026-08-04, i.e. anchored on guidance issued 2026-07-29), and the 10-Q's explanation of the deferred-revenue decline as offset by "increases across other components... associated with growing business levels" are all company-sourced and are discounted and labelled wherever used. - Fair-value caveat: the $205 / $353 / $430 anchors are multiples of consensus FY2028E diluted EPS of $11.77 (17.4x / 30.0x / 36.5x), cross-checked against FY2027E EPS of $9.38 (21.9x / 37.6x / 45.8x), FY2029E EPS of $13.51, the 200-day moving average of $242.72 and the 52-week closing high of $433.33. Stated arithmetic, not a discounted cash flow. All three scenarios assume the market pays LESS than the 55.2x it pays today; none requires multiple expansion.
- Non-equity tripwire: checked and passed. LRCX is NASDAQ-listed common stock (IPO 1984-05-04), not an ETF, fund or ADR, with a 366% 52-week price band, a beta of 1.805, 14.73M shares of daily volume and a rising rather than fixed dividend.
- Timing: FY2026 Q4 and full-year results were released 2026-07-29, six days before this dive, so the most recent quarter is fully reflected. The next print is 2026-10-21, 78 days away. There is no near-term catalyst, which is explicitly part of the case for waiting rather than an oversight. Prices are the 2026-08-04 16:00 ET close, on a day LRCX rose 7.85%.
- Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
- Version: 2026-08-04-full.