KLAC

We rate KLA Avoid. Our central estimate of its value is $106.54 a share, 0.64 times the $165.74 price and well under the $140.88 Hold line. The price already pays for FY2031 operating profit of about $10.8B at 20.3 times, above our 19.6x outside-view multiple, with nothing for waiting five years. To justify the price, wafer equipment spending in FY2031 would have to reach $352B against our $215B. Under 20% of our simulated scenarios get there (1.7% counted; subjective, not calibrated).

Terms used below. Each report values a company three ways: the central case is our best estimate, the bear case a severe but plausible bad outcome, and the bull case the matching good outcome. WFE (wafer fab equipment) is what chipmakers spend on the tools that make wafers, KLA's customers' budget. FY is KLA's fiscal year, which ends 30 June; CY is a calendar year. NOPAT is operating profit after tax. The exit price is our estimate of the share price at the end of the five years. The long fade values the company by letting its growth and its return on new capital fall gradually toward normal levels over 15 years; the outside view applies what comparable toolmakers have traded at. The cost of equity is the yearly return shareholders require, used to bring future cash back to today. "The Street" means the published estimates and price targets of Wall Street analysts. Alpha is Capfolio's own stock-ranking model. The repository's report checker (scripts/research-check.py) raises numbered flags when a report shows a pattern that needs an answer; the flag number is given in brackets next to what it means.

All figures are US dollars, millions unless marked, per diluted share after the ten-for-one split of 2026-06-11. The model is model.py with its saved output.txt; the blind record is assumptions.md, committed and pushed before this report (f1941c051, one figure fix in b8049e233). The analyst stopped at fair value; the market pass (price, Street, odds, Alpha and the chart read) was added afterwards by a second agent working for the chief investment officer (CIO). A review round (a model auditor and an opposite-side advocate, 22 findings) and the auditor's rerun then changed the WFE path, the operating cash, the margin path, the bridge, the case multiples and the odds ranges; section 11 answers every finding, and review.py with review-output.txt prints each change from the blind inputs.

1. What KLA sells, and why it earns what it earns

KLA sells the machines that find defects and measure patterns on wafers and masks, plus service contracts on the installed base. FY2026 revenue was 13,579.5: wafer inspection 6,630.8, patterning 2,706.8, other systems 153.1, services 3,125.9, specialty semiconductor process 502.5 and PCB and component inspection 460.3 (10-K FY2026, Note 17). The process-control segment earned 5,482.8 of segment profit on 12,244.7 of revenue (44.8%); the two non-semiconductor segments earned 172.6 on 1,334.5.

A chipmaker buys inspection to find yield losses faster. The CEO's example at the 2026-03-12 investor day: three points of yield on a leading-edge wafer can pay for that wafer. The spend rises with the number of new designs and nodes, not only with capacity, and each new tool needs algorithms and applications engineers (KLA has 1,600-1,700) that a new entrant does not have. That is why the gross margin sits 10 to 15 points above Applied Materials' and Lam Research's (section 5.2 of our semiconductor-equipment sector note, docs/research/sectors/semi-equipment/note.md).

What limits it: KLA's revenue is a share of what chipmakers spend on tools. Its systems revenue was 7.0% of KLA's own measure of wafer equipment spending (WFE including advanced packaging) in FY2026. When spending falls, KLA falls; it fell 62% from peak to trough in 2007-09, as far as anyone.

2. KLA has beaten its own revenue guide in each of the last eight quarters

KLA beat the midpoint of its own revenue guide in each of the last eight quarters, by +1.9% to +4.3% (median +2.3%), and was never outside the range; gross margin beat the guide in seven of the eight (output.txt). The June 2026 quarter: revenue 3,658 against a guide of 3,575 ± 200, GAAP gross margin 61.4% against 60.7%. FY2026: revenue +11.7%, GAAP operating margin 41.7%, backlog up from $7.86B to $12.57B.

The guide for the September 2026 quarter is revenue of $4.0B ± $0.2B (+25% on a year earlier), GAAP gross margin 61.6% ± 1 point, foundry/logic 73% and memory 27% of semiconductor process-control systems revenue (18% memory in the June quarter guide). Management raised its calendar 2026 WFE view from "$140 billion-plus" in April to the "low $150 billion range" in July.

3. Our forecast gives a central value of $106.54 a share

Systems revenue is FY WFE times KLA's share; services and the specialty and PCB lines have their own paths (assumptions.md gives each blind input and its source; the inputs changed in review are in model.py and section 11).

The WFE path, calendar years on KLA's definition: 152 in 2026; 185.1 in 2027 and 211.2 in 2028 (SEMI's 2026-07-14 growth of 21.8% and 14.1% on its $143.9B for 2026, scaled by KLA's 1.056); 231.6 in 2029 (SEMI's 300mm fab equipment outlook of 2026-10-13, $149B in 2026 to $227B in 2029, scaled the same way); then $215B in 2030 and 2031, the midpoint of management's 2030 model (investor day 2026-03-12). SEMI is the semiconductor equipment industry association. The 7% fall from 2029 to 2030 is our assumption about timing, disclosed: SEMI's outlook ends in 2029 and management's 2030 number is the only dated 2030 figure. It is not a forecast of a downturn.

FY2030 revenue of 24.1B is 7% under management's 2030 model of $26B ± 2.5B, and calendar 2030 (half of FY2030, half of FY2031) at 24.2B is 7% under and inside the range; we credit two thirds of its share claim. The blind model was 13.6% under year for year (section 11, finding 3). The Street's FY2030 EPS of $8.41 is management's number ($84 before the split).

The operating margin is held at the level of the latest dated guide, and no higher: 44.4%, the margin of the September-quarter guide (CFO, 2026-07-28: revenue $4.0B, gross margin 61.6%, opex $690M), flat in every year from FY2027. Gross margin is 61.6% throughout, so the memory-chip cost reversal and the end of intangible amortisation that management expects add nothing; opex is 17.2% of revenue (61.6% less 44.4%), which on our FY2027 revenue is $3,050M against the CFO's quarterly path of $2,865M, a choice that costs value and is stated. The incremental margin FY2026-31 is 48%. Management's 2030 model (a 63.5% gross margin less R&D 11% and SG&A 7%, about 46%; investor day 2026-03-12) is a claim we quote and do not adopt: at 46% value would be $110.56 (+$4.02). The remaining gap is disclosed: 44.4% is still above anything KLA has reported (best 41.7%, FY2026), and at 41.7% in every year value would be $99.75 (-$6.79); at 45.4%, which the CFO's FY2027 opex dollars on our revenue give, it would be $109.05 (+$2.51) (section 11, second-round item 3).

The exit takes the mean of two views of FY2031:

• Long fade, $114.43. NOPAT of 9,138, cut by 118 for depreciation that lags capex in a growth year, grows 7% in FY2032 and fades to 3% over 15 years at a weighted cost of capital (the blended cost of debt and equity) of 10.21%. That is 15.8x year-six NOPAT, or about 14.2x FY2031 operating profit.
• Outside view, $162.08. FY2031 operating profit at 19.6x, the median of four toolmakers' ten-year medians of enterprise value over trailing operating profit (Applied Materials AMAT 15.1x, Lam Research LRCX 16.6x, Onto Innovation ONTO 22.5x, Nova NVMI 31.0x). The bear and bull cases each have their own multiple, the peer median scaled by the case's fade value over FY2031 operating profit: 16.9x and 21.6x (section 11, finding 9). The bridge deducts the $4.5B of operating cash the business needs in the fade half but not here, because the peers' enterprise values net all their cash.

The outside view sits 42% above the fade. The gap is the discount rate: the peer medians come from years when the 10-year Treasury averaged about 2.5%, and the fade discounts at 10.2% on a 5.29% risk-free rate. The latest trailing peer median is 40.6x; used in the outside view it would give $160.68 today, which is a bet that the current boom is the normal level.

Which view we believe: the fade, for the level. It uses today's dated risk-free rate (Damodaran, 2026-10-01) and the cost of capital the rest of the report uses, while the peer medians come from years of lower rates and buyers in 2031 are not promised them. We keep the mean because the method fixes it (the owner's decision of 2026-10-16) and because the outside view is the only half that carries what the market pays for process control in a normal year. Fair value today is $91.58 on the fade alone and $121.49 on the outside view alone, against $106.54 on the mean; Avoid holds under all three (the Hold line is $140.88).

Value today discounts the exit and the payouts (90% of owner cash, as management targets) at 10.42%. Cost of equity is 5.29% (the risk-free rate) + 1.46 (beta) × 4.20% (the equity risk premium, the extra return investors require for holding stocks) − 1.0 point of calibration; the beta is the Blume-adjusted median of the four peers. Damodaran's semiconductor industry beta of 1.50 gives $105.02.

The three drivers that move value most

The multiple is a method input, not an operating forecast; among the operating drivers the next are opex as a share of revenue (18% / 15%: -$2.01 / +$5.53) and gross margin (60.5% / 64%: -$2.49 / +$5.43). FY2031 operating profit ±20%, which bundles all of them, moves value by ±$18.53.

4. Process control has been steadier than its peers in memory or capacity cycles, but not in a collapse

The question is whether KLA's business (process control, meaning inspection and measurement of wafers) has been steadier through the cycle than its peers'. Partly. KLA's quarterly revenue fell 21% from peak to trough in 2022-23, against 39% at Lam; in 2007-09 it fell 62%, against 75% at Lam and 60% at Applied (market data quarterly, output.txt). In the 2018-19 cycle our peak-to- trough count shows no fall at all, which is an artifact: Orbotech was consolidated from February 2019 and its revenue covers whatever organic decline there was.

In the model the steadiness has three sources, all measured, none assumed from management's word:

1. Services, 23% of revenue, never fell in any year FY2015-26 (sector note 5.1). The bear case keeps them growing, slowing to +4%, and they carry revenue while systems fall: bear FY2029 revenue falls 5.1% while FY WFE falls 8%.
2. Gross margin has a narrow history, 56.8-64.1% since FY2015, so the bear case only goes back to the pooled 60.5%. The one-year downturn test (WFE -18% with no China fill) gives revenue -12.1% and an operating margin of 35.7% at a 58% gross margin, 32.7% at 55%.
3. Opex is cut slowly, not at all in R&D. The bear case keeps growing opex at 5% a year, so its operating margin falls from 43.7% to 39.5% rather than holding.

The model does not credit share as a shock absorber: the bear case loses share (7.30% to 7.00%). The 2023 resilience came in a memory-led downturn, when KLA's logic-heavy mix protected it. That mix is moving the other way (memory 18% to 27% of semiconductor systems revenue between the June and September guides). A downturn led by foundry/logic has no clean precedent after 2009, and in 2009 KLA was not steadier. So: steadier against a memory or capacity cycle; no better than its peers against a collapse.

5. Answers to the open questions in our semiconductor-equipment sector note

Our sector note (docs/research/sectors/semi-equipment/note.md) left these questions open for the company report. The first seven carry the note's own numbers (10.1.1 to 10.1.7); the last five are specific to KLA.

10.1.1 Which WFE total, and KLA's share on one definition. KLA uses its own total, WFE including advanced packaging: about $120B in CY2025 and the "low $150 billion range" in CY2026 (CFO, 2026-07-28), roughly 3% above SEMI's $116.9B for 2025. On that definition KLA's systems revenue was 7.0% of fiscal-year WFE in FY2026. KLA's CY2024 total on the same definition is not in our sources, so we cannot state 2024 and 2025 shares on one basis; management's own claim is +150-160bp of share 2021-25. We model share only against KLA's total and never mix it with SEMI's.

10.1.2 End-market mix by quarter. KLA discloses foundry/logic against memory only as a guide on each call, for semiconductor process-control systems: 82/18 for the June 2026 quarter, 73/27 for September. We saved two of the eight quarters; the other six are a gap. China by quarter is in sector note 4.2 (26.2% in the June 2026 quarter; 29.8% of FY2026). KLA does not split leading-edge from mature-node revenue.

10.1.3 Revenue if WFE falls 15-20% with no China fill. Built from the segments with services at their own history: WFE -18% gives revenue 11,931 (-12.1%); operating margin 35.7% at a 58% gross margin.

10.1.4 China revenue by owner. Not disclosed. "China" counts shipments to fabs in China, including foreign-owned ones. The range depends on the foreign-owned fabs' share, which no filing gives; we do not estimate it. KLA's China revenue was flat at $4.05B in FY2026 while legacy buying offset export restrictions (sector note 4.2).

10.1.5 Where the gross-margin gain since FY2022 came from. There is almost none to split: 61.0% in FY2022, 61.3% in FY2026. Inside that, the memory-chip cost took about a point and falling amortization of acquired intangibles added back; mix and price did the rest. China fell from 43% to 30% of revenue over FY2024-26 while gross margin rose from 60.0% to 61.3%, so China revenue has not been carrying the margin (our inference).

10.1.6 Backlog and cancellation terms. Backlog rose to $12.57B from $7.86B. The 10-K says backlog "does not provide meaningful information about the timing of future revenue recognition" and gives no twelve-month split; contract liabilities of 1,775 have 1,537 due within twelve months. Cancellation terms are not disclosed beyond the risk-factor language on "order modifications, pushouts or cancellations".

10.1.7 Capex and headcount for the peak. Capex was 2.8% of revenue in FY2026; we model 3.0%. Headcount rose from about 15,000 to 17,000 regular full-time employees in FY2026 (+13%, against revenue +11.7%). Purchase commitments are $5.97B, mostly within twelve months. In a trough the cost is mostly people: 27% of staff are in R&D, which KLA keeps through downturns, and the bear case carries that as opex growing 5% a year.

KLA-specific question 1. Logic mix or share, and a logic-led downturn. Our evidence cannot split the 2023 resilience between mix and share; management claims both (+360bp of process-control share since 2021). Section 4 gives our answer for a logic-led downturn: the model treats it as the bear case with share falling, not as a case where KLA is spared.

KLA-specific question 2. The memory-cost headwind. The CFO put it at "somewhere around 100 basis points", "probably a little bit more", continuing into 2027 and then turning to a tailwind. The model holds the guided 61.6% flat to FY2031 and credits none of its reversal (management's 63.5% is a claim; at 63.5% value would be $110.83). If memory prices fall, the cost falls with them; the risk is that memory stays tight through 2028.

KLA-specific question 3. Backlog due within twelve months, and TSMC. Not disclosed (see 10.1.6). TSMC was about 19% of revenue in FY2025 and FY2026 and is the only customer above 10%. Its share of the backlog is not disclosed.

KLA-specific question 4. China against domestic entrants. An analyst on the 2026-07-28 call said Chinese memory makers spend "the bulk" of their process-control budget with local suppliers; KLA did not confirm or deny it. The CEO's answer was the high-mix, low-volume argument and the applications-engineer base. We take it that KLA is defended at the leading edge and exposed in Chinese memory and mature nodes. The bear case's share of 7.0% includes that loss.

KLA-specific question 5. Advanced packaging revenue. Process-control systems for packaging: about $635M in CY2025, about $1.1B expected in CY2026 (+70%; CEO, 2026-07-28). The investor day gave $950M of advanced-packaging revenue for 2025, which we read as including packaging-related sales in the other segments (our inference; KLA does not reconcile the two). Packaging is inside KLA's WFE total, so the model carries it as part of systems share. It has shorter lead times and newer customers, so it is likely more cyclical than front-end inspection; we have no downturn history for it.

6. Where we disagree with our sector note on three points

1. The 2019 downturn was not shown to be shallow for KLA. The note puts KLAC with AMAT at -5% to -23% peak to trough in 2019 and 2023. For 2019 the reported totals include Orbotech from February 2019, so the fall cannot be measured on them; our own count shows none. Only 2023 supports the claim.
2. "Logic-heavy" is a protection that is shrinking. The note gives the 73/27 split as a reason for steadiness. The same split was 82/18 a quarter earlier and the CFO expects memory to take a larger share of 2027 WFE. The note's test E7 (the claim that KLA is the least cyclical of the three companies compared) should be read as a claim about memory-led downturns; we would not expect it to hold in a logic-led one.
3. 2027 WFE: same direction, different base. The note's 2027 figure is SEMI's (+21.8%, about $175B, excluding packaging). Our central CY2027 is now SEMI's growth on its own 2026 base scaled to KLA's definition, which includes packaging and runs about 5.6% above SEMI: $185.1B (the blind model had $180B). It sits below the "around $190 billion" outside consensus the CFO cited. Not a disagreement, but the two numbers should not be compared directly.

Where we agree: China is the largest single risk and is in the bear case; services are the closest thing to recurring revenue; backlog is an order book, not a contract.

7. The bear case ($58.41) and what could make the central case wrong

Bear, $58.41. The 2027-28 capacity wave is followed by digestion, as every equipment peak since 2001 has been: WFE falls 10% in CY2028 (165 to 148) and 5% in CY2029 (to 140) and recovers only to $162B by CY2031. Share stays at 7.0% as Chinese suppliers take local memory and mature nodes; gross margin goes back to the 60.5% pooled average; working capital stays at 22% of revenue; the exit multiple is 16.9x. FY2031 operating profit is 6,552, against 10,751 central.

Bull, $154.11. The build-out runs at the outside consensus ($195B in CY2027) and keeps growing; KLA takes management's full +150bp of share and a 64% gross margin; the exit multiple is 21.6x. FY2031 revenue of 29.8B is above management's own 2030 model.

What could make the central case too high: the central path has no downturn after 2028 because none is dated, and history says one will come within the five years. The case weights (25/50/25) are ours; weighted, the value is $106.40. What could make it too low: KLA beats its own guide every quarter, and we did not add that beat to the level of revenue (as a level shift in every year it adds $3.43, for FY2027 alone $0.12; section 11, finding 8).

8. The September 2026 quarter, reported in late October, could change five things

KLA reports the September 2026 quarter after this report. It could change:

• FY2027 revenue. The central case reads the guide plainly for the full year; the forecast point for the September quarter is $4.09B, the guide's $4.0B plus the median beat (section 10, forecasts), and December about $4.4B. A December guide below $4.2B lowers FY2027-28; one above $4.6B says the beat pattern continues.
• Gross margin against the 61.6% guide, and any update on the memory cost.
• Backlog (the CFO expected it to keep rising from $12.57B) and the memory share of systems revenue.
• China (26.2% in June) and anything said about the Affiliates Rule (a US export-control rule), whose suspension ends on 2026-11-10.
• The CY2027 WFE view. A management number above $190B moves us toward the bull path for CY2027; it does not change the FY2031 normal level by itself.

9. Dates and results that would show the central case is wrong

10. What the $165.74 price pays for, and how the Street and Alpha see it

Added after the analyst stopped at fair value, by the market pass (the step that reads the price, the Street and Alpha). The market pass as first written changed no assumption; the review round that followed did, and this section was rewritten to the revised model (section 11). Its scripts are market.py with market-output.txt and street_path.py with street_path.txt; the price is in data/quote.json and the Street's estimates in data/analyst-estimates.json.

The price and how it moved

KLAC closed at $165.74 on Friday 2026-10-16 (Alpaca SIP, a consolidated US market data feed, daily bar, split adjusted). Robinhood's last regular-session trade was $165.75 at 19:59:59 UTC, one cent or 0.006% apart; the 2026-10-15 close of $174.46 is the same in both, which confirms the post-split basis. Both were read on Sunday 2026-10-18. After-hours trade was $166.20 (Robinhood). The price is 16% below its 2026-10-06 close of $197.46 and 13% below the $190.80 close of 2026-07-28, the day of the results. The next close, 2026-07-29, was $170.19 (-10.8%); the price was $195.45 by 2026-08-04 and $210.52 on 2026-07-24 before the results. The last session alone fell 5.0% ($174.46 to $165.74). We did not research why any of these moves happened.

What the price pays for

Market value is $217.5B on 1,312M diluted shares, and enterprise value is $218.5B (notes of $5.95B less $4.9B of cash and securities at June 30). That is 38.6x FY2026 operating profit and 27.7x our FY2027 operating profit of $7.9B. Against our FY2031 operating profit of $10.8B it is 20.3x, above our own 19.6x outside-view multiple: the price already pays today for profit five years out and takes no discount for waiting.

Each row changes one input and leaves the rest at central:

Management's 2030 model is $215B ± $20B of wafer equipment spending (calendar 2030), which is now our central, so $352B is 64% above it and 50% above the top of its range. The multiple of 42.6x is above every peer's ten-year median (15.1x to 31.0x) and above the peers' latest trailing median of 40.6x, which prices a boom in progress. A Buy needs $376B or 1.75x our operating profit; no multiple inside its bracket gives one. Using KLA's own beta (market data profile 1.447, Blume 1.30) instead of the peers' 1.46 would lift central fair value to $113.23 and the purchase price 3.5% to $103.83; we did not use it and the view does not change (section 11, advocate finding 6).

The chance the price is justified is 1.7%; the world in which it is right (checker flag 16)

In 20,000 simulated scenarios (odds.json) the chance that value reaches the price is 1.7% (6.6% in the bull world, under 1% in the others). The 10th to 90th range is $76 to $149 and the median is $111, 4.1% above the central $106.54 (the checker's flag for a median far from central, flag 14, does not apply). We write the odds only as a band: under 20% of our scenarios, subjective and not calibrated.

The simulation carries the dated drivers only. The ranges are the ones registered before the blind commit (WFE up to 260, bull 290; multiple up to the highest peer ten-year median, 31.0x), with each case's own multiple and the cost of equity, services, working capital and tax added. The first revision had widened the ends to a Street-implied spending figure and to the peers' latest trailing multiple of 40.6x, both read after the price and the second price-dependent; that raised P to 11.6% and was reversed (section 11, second-round items 1 and 2). The intermediate steps quoted in the first revision came from scratch runs outside this folder and are not reproducible; they are withdrawn.

Two of the report checker's flags fire: the odds flag (flag 16), because P is 1.7%, under the 10% line, and the price-outside-our-range flag (flag 18), because the price of $165.74 is above our 90th percentile of $149, at about the 98th. A price outside nearly all our scenarios means our ranges or our model may be what is wrong, which is why the next paragraph prices the world in which the price is right; we did not widen any range toward the price.

The coherent world in which $165.74 is right uses only dated inputs: the Street's own revenue and operating profit through FY2031 ($31.8B and $15.4B) with the exit at 24.6x FY2031 operating profit, inside the peers' ten-year medians (ONTO 22.5x, NVMI 31.0x) though above their 19.6x median, or 19.4x all-in (street_path.txt). Two things weaken it. The Street's FY2031 figures come from 8 revenue and 3 earnings estimates and jump 18% in revenue after +6% the year before; growing the Street's FY2030 at our FY2031 rates instead gives $126.27 and then needs 32.4x (23.3x all-in) to reach the price. And 24.6x is a multiple for a normal year while FY2031 is already bigger than any year KLA has had. We did not move any assumption toward this world.

Our value is 0.48x the Street's average target (checker flag 17)

Our $106.54 is 0.48x the Street's average target of $223.88 (market data price-target-consensus: median $220, range $180 to $325, 13 targets in the last quarter, 73 in the last year; horizons and update dates absent, captured 2026-10-18T14:46:53Z). The Street's estimates by fiscal year, with the targets as a multiple of its own earnings:

Street operating profit is the vendor's ebitAvg, whose basis (GAAP or adjusted) is not stated; KLA's non-GAAP EPS ran 3% above GAAP in FY2026 ($3.76 against $3.66). The Street's FY2030 EPS of $8.41 is management's 2030 model ($84 before the split); five analysts. Putting the Street's revenue and operating profit through our engine, with our capex, working capital and fade unchanged, the gap to the average target ($117.34) splits in the order smooth level, FY2031 jump, exit multiple:

The earnings level explains a third of the gap to the target ($40.58) and the exit multiple two thirds. Through FY2030 alone, with the exit year left at our FY2031 profit, the Street adds only $2.45 (Street through FY2030 then ours, $108.99), because the exit capitalises one year; that is not the earnings level's worth. The exit multiple in the last row is the outside-view part only. The exit is the mean of the fade and the outside view, so on FY2031 operating profit the all-in exit multiple is 16.9x in our central, 19.4x on the Street's path at the price (23.3x on the smooth level) and 27.3x on the Street's path at the target. We do not adopt 40.5x; it is the peers' latest trailing 40.6x held as normal, the bet that a boom multiple is a normal multiple, and our report already prints it ($160.68 with our own profit, section 3). If the target is a twelve-month price (market data gives no horizon), $223.88 discounted one year at our cost of equity is $202.75 today.

Our 10th to 90th range is $76 to $149, 50% as wide as the Street's target spread ($73 against $145), and the price is above our 90th percentile (checker flag 18). We did not widen it: the price being outside is what the odds already say, and the first revision's widening to a trailing multiple and a Street-implied ceiling was reversed because it fitted the price (section 11, second-round items 1 and 2).

Capfolio's Alpha model also says avoid, for a different reason

Alpha's dated read (alpha ticker analyze KLAC, data as of 2026-10-17, run 2026-10-18) is against the stock: core score 17.6 of 100 (Weak), momentum lens "breakdown, avoid or reversal only" (score 22), valuation lens "expensive, richer than its own history on all four multiples", growth moderate and quality supported (score 50). The factor percentiles are valuation 17, momentum 35, growth 42, quality 53 and expectations 43. Its rank is not available: the rank cache is stale (rank_cache_stale) and data health is limited, so no position in the model's list is claimed.

We agree on direction and differ on the reason. Alpha's weakness comes from price against history and a falling trend; ours comes from what the business earns by FY2031 against what the price pays for it. Alpha scores growth and quality as middling where we see a franchise earning a 44% operating margin, and it says nothing about FY2031. If the trend turned and the price rose, Alpha's read would improve and ours would not; if the price fell to $100.31 or lower ours would turn to Buy. The two lenses are shown side by side; neither is a veto of the other.

Technical read of the chart: a D grade (timing, beside the view and not an input to it)

levels.py KLAC --group SMH --brief, Alpaca SIP daily bars to 2026-10-16 (sha256 prefix 61ad48c379289bea): grade D. Stage is topping: the close of $165.74 is below the 30-week average of $195.10, which is still rising. Structure is damaged: the close is below the 50-day average ($186.91) and the swing low of $165.89 from 2026-09-16 is broken. The close is 2.6 ATR (average daily trading ranges) below the 21-day average, the ratio to QQQ is 26% under its 63-session high, and both QQQ and SMH are below their 21- and 50-day averages. KLA's results are expected on 2026-10-28 (vendor date, not company-confirmed), eight sessions away. The chart improves only on a weekly close above $195.10 while that average is rising. A D grade is poor timing for a new purchase and agrees with the Avoid, but it says nothing about the $106.54, which a price of $166 or $120 leaves unchanged.

The forecast we committed for later grading

One, for the release expected 2026-10-28: September-quarter revenue of $4.09B, within $4.00B to $4.20B. It is the guide midpoint of $4.0B plus the median beat of the last eight quarters (2.3%; 8 of 8 above the midpoint, +1.9% to +4.3%, output.txt); the low end is no beat and the high end the top of the guide's $0.2B band. The blind assumptions committed $4.0B within $3.8B to $4.2B, reading the guide plainly; the point was revised in review (section 11, finding 7) from a record that was in this folder before the blind commit. We add no gross-margin or December-guide forecast: assumptions.md commits a lower bound for the December guide ($4.2B) and a point of about $4.4B, but no upper bound before the blind commit, and its 62.0% gross margin is a full-year figure, not the quarter's.

Checker flags and how each was answered

11. A review round moved the value from $105.41 to $106.54, and every finding is answered

A model auditor and an opposite-side advocate reviewed this report after the market pass (11 findings each), and the auditor then reran it and raised further points (the second round, below; it is listed first because it has the latest answers). In the table below, "Blind" is the analyst's model before any market data, "First revision" is the model after the first round, and "Final" is the model after the second round. The auditor's findings carry the auditor's own numbers (F1 to F13) and the advocate's are numbered 1 to 11. The reviewers' dollar effects are theirs, run on the model as it stood; ours are rerun on the final model in review.py (review-output.txt) and street_path.py. An assumption changed only where the evidence is dated and does not depend on the KLAC price; nothing was moved toward the price.

The central bridge, in order: SEMI's July path for CY2027-28 +$0.51, SEMI's 300mm outlook for CY2029 +$0.73, management's $215B for CY2030-31 (the dip) +$2.83, operating cash at the CFO's midpoint -$0.24 (that is the first revision, $109.23); then the margin held at the September-quarter guide's own 44.4% instead of the blind margin path -$3.75, no operating cash deducted in the outside view +$1.08, the pension liability -$0.02: $106.54. Bear: operating cash -$0.24, its own multiple -$4.21, the outside-view cash +$1.08, pension -$0.02. Bull: operating cash -$0.24, its own multiple +$6.90, the 2029 WFE path +$0.38, the outside-view cash +$1.08, pension -$0.02. From the $109.17 of the previous draft, the one change is the margin: 45.4% to 44.4%, -$2.63 (the draft took the CFO's FY2027 opex dollars, $2,865M, and held their share; the held margin takes $3,050M in FY2027, which is why this is more than the -$2.29 of a FY2031-only reading).

Second round, after the auditor's rerun

1. Odds tail (auditor F1, F2): fixed. The bull world's multiple maximum was 40.6x, the peers' trailing EV/EBIT at today's prices: price-dependent, circular with the price being tested, and read after the price. It is back at the registered 31.0x (the highest peer ten-year median, set before the blind commit). Before P 11.6%; after 1.7%, with the 10th/50th/90th percentiles $76/$111/$149 (before $66/$113/$169). The fall is not the ranges alone: the margin hold and the narrower gross-margin and opex ranges below also lower it. The intermediate P values quoted in the first revision (2.5%, 3.2%, 5.2%, 11.2%) came from a scratch script outside this folder and are not reproducible; they are withdrawn. Evidence: odds.json, drivers.json.
2. Spending ceilings (auditor F4): fixed. WFE maxima 285 (base) and 300 (bull) cited the Street-implied FY2031 spending read after the price. Back at the registered 260 and 290. No dated, non-Street source supports higher ceilings: SEMI's outlook ends at $227B in 2029 for 300mm fab equipment. Effect is inside the P above.
3. Operating margin (auditor F5): fixed at the September-quarter guide's own 44.4%, the rule used for the Lam Research report. Before: gross margin 62.0% rising to 63.0% and opex growing 10% a year, FY2031 operating margin 45.7%, supported only by management's 2030 model, a claim. First answer: gross margin 61.6% and the CFO's FY2027 opex dollars then held at their share, 45.4%, value $109.17. The latest dated guide gives no opex leverage, so the margin is now the guide's own: revenue $4.0B, gross margin 61.6%, opex $690M, an operating margin of 44.4%, flat in every year from FY2027 (opex 17.2% of revenue, $3,050M in FY2027 against the CFO's quarterly path of $2,865M, a conservative choice that is stated). Value $106.54. Management's 46% (63.5% gross margin less R&D 11% and SG&A 7%, investor day 2026-03-12) is quoted as a claim: at 46% value would be $110.56 (+$4.02). The remaining gap is disclosed: 44.4% is above anything KLA has reported (best 41.7%, FY2026); at 41.7% in every year value is $99.75 (-$6.79), at 45.4% $109.05 (+$2.51). The auditor's -$8.30 at 41.7% was on the first revision. The simulation carries opex as a share of revenue (15% to 18%, mode 17.2%) and gross margin 60.5% to 63.5%.
4. Growth after year five (auditor F6): kept, with a dated reason, disclosed. The fade starts at 7% growth in FY2032 and reaches 3% over 15 years; it follows a flat FY2030-31 (the pause at management's $215B). Seven percent is below every dated long-run measure of equipment spending in the sector note (section 3.1, SEMI total equipment spending): $29.5B in 2008 to $135.1B in 2025 is 9.4% a year and $64.5B in 2018 to $135.1B is 11.1% (output.txt). Value $106.54; at 5% it is $102.76 (-$3.78) and at 3% $99.38 (-$7.16) (the auditor's -$3.89 and -$7.37). Seven percent is still a judgement; the simulation draws 5% to 8.5%.
5. Cash in the outside view (auditor F7): fixed. Peers' enterprise values net all their cash, so the operating cash the business needs ($4.5B) is no longer deducted in the outside half (the fade half still deducts it). Outside view $158.65 to $162.08 on the final model; central +$1.08 (the auditor's +$1.08); the same +$1.08 in the bear and bull cases.
6. Pension (auditor F9): fixed. The $43.1M pension liability (10-K FY2026, other non-current liabilities, 2026-06-30) is in the exit bridge: -$0.02.
7. Bear text (auditor F8): fixed. Section 7 said bear spending "falls 9% in CY2028"; the path is 165 to 148, which is -10.3%. It now says 10% (and 5%, to 140, in CY2029).
8. Fade against outside view (auditor F13): answered in section 3. The outside view is 42% above the fade ($162.08 against $114.43 at the exit). We believe the fade more for the level, because it uses today's dated risk-free rate (Damodaran 2026-10-01) and the report's own cost of capital, while the peer medians come from years of lower rates; we keep the mean because the method fixes it. Fair value today is $91.58 on the fade alone, $121.49 on the outside view alone and $106.54 on the mean; Avoid holds under all three (Hold line $140.88).
9. Unverified figures (auditor F11, F12): labelled. Four of the auditor's small items (finding 11 below) and the advocate's exit-multiple history (advocate 8) are the reviewers' figures and have not been rerun here; they are marked unverified in their lines. The first revision's P steps are not reproducible (item 1).
10. Folder: done. The report folder is moved from docs/research/2026-10-18-klac to docs/research/2026-10-16-klac with git mv; it shows as renamed in git status.

The checker's odds flag (16, P under 10%) and its price-above-our-90th-percentile flag (18) fire again. Flag 16: P is 1.7% (6.6% in the bull world, under 1% in the others). Flag 18: the price of $165.74 is above the 90th percentile of $149, at about the 98th. They are answered in section 10 by the world in which the price is right and by the gap split; no range was widened toward the price.

First round

The model auditor

1. The story behind the Street-gap flag (flag 17): fixed (the story; no value changed). Before: the Street's FY2027-30 "adds only $2.70", the rest FY2031 and the multiple. After: of the $117.34 gap to the average target, the smooth earnings level is +$19.73 (17%), the Street's FY2031 jump +$20.85 (18%) and the exit multiple +$76.76 (65%) (section 10). Evidence: Street analyst estimates and price targets read 2026-10-18 (street_path.txt); the auditor's split on the blind model was +$40.89 earnings, +$77.58 multiple.
2. "40.7x" overstates it: fixed (framing). Before: the Street's path needs a 40.7x exit. After: that is the outside-view part; all-in over FY2031 profit the exit is 16.9x for us, 19.4x on the Street's path at the price (23.3x on the smooth level) and 27.3x at the target (street_path.txt). No value changed.
3. Management's 2030 model: fixed (comparison and path). Before: "FY2031 revenue 9% under"; year for year FY2030 was 13.6% under. After: FY2030 revenue 24,124 is 7.2% under $26B, calendar 2030 24,169 is 7.0% under and inside $26B ± 2.5B; the Street's FY2030 EPS of $8.41 is management's $84 before the split. Management's midpoint at a 46% margin grown at our FY2031 rate is $117.34 (+$10.81; the auditor's +$15.03 was on the blind path, whose FY2031 growth was +4.3%), its bottom $107.39 (+$0.86), not adopted because we credit two thirds of its share claim and none of its margin claim. Evidence: investor day 2026-03-12.
4. The 45.7% margin comes from 10% opex growth: disclosed, and changed under second-round item 3. Before 45.7% with a 50% incremental margin; after 44.4% and 48% with the margin held at the guide's own. At a 45% incremental margin (opex at 18.5% of revenue) value is $103.37 (-$3.16); at 40% (20.7%) $97.85 (-$8.68). Management's 2030 model nets to about 45.5% (63.5% gross margin less 18% opex); ours is 44.4% with a lower gross margin and opex of 17.2%, a mix that is disclosed rather than defended.
5. Operating cash: fixed. Before $4,000 (the low end of the CFO's "$4-5 billion to run the company"); after $4,500, the midpoint: -$0.24. The top, $5,000, is a further -$0.12 on the final model. Evidence: investor day 2026-03-12; June cash of $4.9B is a balance, not a need.
6. Confidence 65% too high: fixed. Before 65%, defined on FY2031 operating profit; after 48%, defined as the guide does, on value: 47.8% of the simulation lands within 20% of $106.54 ($85.23 to $127.84). On operating profit it is 58.0%. The first revision gave 40%; the figure rose because the spending and multiple tails were put back. Evidence: confidence.py, confidence-output.txt.
7. Q1 forecast point: fixed, disclosed as revised in review. Before $4.0B (3.8 to 4.2); after $4.09B (4.00 to 4.20), the guide midpoint plus the median beat of 2.3% (8 of 8 quarters beat, +1.9% to +4.3%; releases in data/releases/, output.txt); the low end is no beat. The blind assumptions.md still commits $4.0B.
8. The guide beat as a level shift: rejected. A 2.3% revenue beat at the same gross margin and opex adds $0.12 in FY2027 only, $0.28 in FY2027-28 and $3.43 as a level shift in every year to FY2031 (the auditor's +$2.96 was on the blind model). Eight quarters measure how conservative the guide is, not the FY2031 level; there is no dated evidence that the beat persists five years.
9. One multiple across all cases: fixed. Before 19.57x everywhere; after bear 16.89x, central 19.57x, bull 21.56x (the peer median scaled by each case's fade value over FY2031 profit: 12.23x, 14.18x, 15.62x). Bear $61.80 to $58.41 (multiple alone -$4.21), bull $146.02 to $154.11 (multiple alone +$6.90), weighted $104.66 to $106.40. The peer extremes (15.1x and 31.0x) give $55.61 / $186.93 / $113.90, not adopted: they pair the lowest and highest of four peers' medians with a case, a difference of business, not of the case. The auditor's coherent scaling on the blind model gave $57.56 / $152.83 / $105.30.
10. The odds simulation is truncated: fixed in part. Added to the simulation, none price-dependent: each case's own multiple, the cost of equity (9.75% at KLA's own beta to 10.59% at Damodaran's), services growth, working capital (16% to 22%) and tax (14.5% to 16%). Not added: the 40.6x multiple tail and the Street-implied spending ceilings, which were put back (second-round items 1 and 2). P is 1.7% with $76/$111/$149 (blind ranges 2.2% with $74/$107/$147). Evidence: odds.json, drivers.json.
11. Small items: disclosed, none applied except as noted. Rerun on the final model: 1,317M shares if all RSUs count -$0.40 (the auditor accepts the treasury method), cash yield 3.77% +$0.04, amortisation schedule 129.0/49.1/35.6/14.8/1.5 -$0.02. Unverified, the auditor's figures and not rerun here: debt at fair value +$0.27, a lease double count in the fade half +$0.06, the FY2027 stub payout date -$0.04, and peers' EV netting all cash +$0.96 (superseded by second-round item 5, +$1.08 on our model). Each is under 1%.

The opposite-side advocate

1. CY2027-28 WFE on SEMI's July path: fixed. Before 180 and 196; after 185.1 and 211.2 ($143.9B for 2026, +21.8% and +14.1%, times KLA's 1.056); +$0.51 (the advocate's +$0.50). Evidence: SEMI 2026-07-14; KLA's $152B against SEMI's $143.9B for 2026.
2. CY2029-31 against SEMI's 300mm outlook: fixed, with a disclosed dip. Before 200, 205, 212; after 231.6, 215, 215: 2029 +$0.73, 2030-31 +$2.83. Flat 231.6 for 2029-31 would give $113.48 (+$6.95) and a 2019-style 231/215/228 $109.19 (+$2.65), neither adopted: the outlook ends in 2029, management's $215B ± $20B (investor day 2026-03-12) is the only dated 2030 figure, and the fall to it is our timing assumption. SEMI's release returned 403 to the advocate; the text we hold is in data/sector/semi-2026-10-13-300mm-fab-outlook.txt.
3. Share 8.5% or 9.0%: rejected. Central 8.00% stays; 8.5% gives $112.50 (+$5.96) and 9.0% $118.46 (+$11.93) (the advocate's +$5.89 and +$11.77 on the blind model). KLA's 5.8% to 7.4% over 2021-25 and its +150bp claim are management's own series; we credit two thirds because memory rises from 18% to 27% of systems revenue between the June and September guides and an analyst on the same call said Chinese memory makers buy mostly from local suppliers (call of 2026-07-28; KLA did not confirm it). The simulation reaches 9.2% in the bull world.
4. Services 14% every year: rejected. The central starts at 14% and fades to 10%; 14% flat adds $2.74 (the advocate's +$2.80). The bear and bull paths bracket it and the simulation carries services (-4.5 to +2.5 points). Services never fell FY2015-26 (market data), which the bear case already uses.
5. Gross margin 63.5%: rejected. 63.5% adds $4.29 on the final model, whose gross margin is the guided 61.6% held flat; management's 63.5% ± 0.5 is its own 2030 target, a claim, and the rule is no margin gain beyond what dated guidance supports (second-round item 3).
6. KLA's own beta of 1.30: rejected as the central, shown. The peer-median Blume beta of 1.46 was fixed before the blind commit; the own beta (market data profile raw 1.447, Blume 1.299, 2026-10-18) gives $113.23 (+$6.69; the advocate's +$6.62) and is the low end of the simulated cost of equity. The advocate gave no dated reason to prefer one vendor's raw figure over the peer set, and the method prints it beside the central; choosing it after seeing the price would move the model toward it.
7. Combined world: rejected as a central case, priced. Flat 231.6 for 2029-31, share 9.0%, gross margin 63.5% and services 14% give $134.01 (+$27.47), with beta 1.30 $142.59; with our dip kept $125.95 and $133.96 (the advocate's $133.46 and $141.98 on the blind model). The price then needs a 29.2x outside multiple, 26.4x with the own beta. Stacking each lenient input is not evidence for the stack; even so the price is not reached.
8. Exit-multiple history: disclosed, no change; unverified. The advocate's own check, not rerun here: KLA's 49 quarter-ends have a median of 16.8x, a 10th to 90th percentile of 11.2x to 27.6x and a post-2022 median of 21.8x, so 19.6x is not too low.
9. Five-year profit growth: disclosed, no change. FY2026-31 operating profit is 1.90x (blind 1.90x), against KLA's windows of 2.27x to 2.86x in FY2015-26 and a 20-year median of 2.1x (the advocate's figures). The window starts from FY2026's 41.7% operating margin, the highest of FY2015-26 (23.5% to 41.7%, output.txt), and ends in a normal year, and the bull case is 2.57x, inside the historical range.
10. Year mismatch: fixed (same as auditor 3). Before FY2031 $23.7B against management's calendar 2030 $26B, 11% under like for like; after calendar 2030 is $24.2B, 7.0% under and inside the range.
11. Opex 6%: rejected. Opex growing 6% a year from the guided FY2027 level instead of held at the guide's margin adds $5.84 on the final model (the advocate's +$5.36 on the blind model) and gives a 53% incremental margin. The advocate does not recommend it without a dated cost plan and there is none; opex grew 12% a year FY2016-26 (market data).

12. Sources

• KLA 10-K for FY2026 (filed 2026-08-06) and earlier 10-K and 10-Q copies, masked, from the sector note's data/edgar/.
• KLA 8-Ks of 2026-03-12 and 2026-08-06 (EDGAR, read 2026-10-18).
• KLA earnings releases, 2024-07-24 to 2026-07-28 (data/releases/, masked).
• Call transcripts: Q3 FY2026 (2026-04-29), Q4 FY2026 (2026-07-28) and the investor day (2026-03-12), via TradingView (Quartr), masked (data/calls/).
• Market data annual and quarterly statements for KLAC, fetched 2026-10-18 (data/); peer key metrics, statements and betas for AMAT, LRCX, ONTO, NVMI and ASML, fetched 2026-10-18 (data/peers.json).
• Sector notes: docs/research/sectors/semi-equipment/note.md, memory and semis-ai, as of 2026-10-18.
• Review round, read 2026-10-18: SEMI's 2026-07-14 wafer equipment forecast (through the sector note, section 5.4) and its 300mm fab equipment outlook of 2026-10-13 (data/sector/semi-2026-10-13-300mm-fab-outlook.txt); the model auditor's and the advocate's findings, listed in section 11.
• Damodaran implied equity risk premium and risk-free rate, 2026-10-01 (method.MARKET); semiconductor industry beta, October 2026.
• Market pass, read 2026-10-18: Alpaca SIP daily bars (the technical-analysis levels.py loader); Robinhood get_equity_quotes; market data quote, price-target-consensus, price-target-summary, analyst-estimates (annual) and profile beta for KLAC (data/quote.json, data/analyst-estimates.json, street.json); alpha ticker analyze KLAC for Alpha's read.