KLA
The company whose machines inspect and measure silicon wafers between chipmaking steps, catching a defect before hundreds more steps are wasted on a ruined wafer.
The scores
Growth3/4 ↑
Revenue up 11% after 24% and a decline before that: it tracks wafer-fab equipment rather than beating it, and China contributed nothing.
Business quality3/4 →
A rival confirms concentration, and its tools go in before the process equipment. But in-situ metrology and Chinese entrants are untested against it.
Financial position4/4 →
Net debt under $1 billion against $5.7 billion of operating profit, nearest maturity in fiscal 2029, revolver undrawn, and the plan needs no outside money.
Execution record4/4 →
Revenue above the guided midpoint, earnings at the top of the range, margins up three years, and a segment note that reconciles.
Per share capture4/4 →
Share count down 3.1% in two years, buybacks 7 times compensation expense, no convertible, and earnings per share growing faster than net income.
Potential2/4 ↑
Backlog is conditional and the 2027 step-up is somebody else's capital decision - expected, not committed, and above the industry body's own forecast.
Risk profile2/4 →
Two concentrations resolving externally in single steps: one customer at 19%, and 30% of revenue where an export licence can void the order.
Valuation2/4 ↑
Buyers pay 42.0 times a year's earnings: the price needs the plan and the good case beyond it, 57% growth to reach its norm.
Evidence quality3/4 →
Everything central is filed, but the growth case runs through management's own equipment forecast and a service split that exists only on a call.
The Verdict
We rate KLA unattractive. Its machines inspect and measure silicon wafers between chipmaking steps, a job it dominates, keeping about 40 cents of profit from each sales dollar, so the business is solid. Orders are rising on demand for AI chips. The problem is the price: even after a sharp fall from the high, buyers pay nearly twice what the company has normally cost for each dollar of earnings, in an industry its own filings call cyclical. So it is full, not cheap. The view hinges on earnings rising, or the price falling, until that gap closes.
The long caseAI is pulling chipmakers into building advanced logic and memory factories, which need KLA's inspection; earnings must rise much faster for that to justify the price.
The evidence under this - 8 claims
- The 41.7% margin is the highest in the ten-year series in the research pack, which is what a cycle peak looks like from inside it.
- The vendor's operating income agrees with this figure while the repo's own XBRL reader returns $772,070 thousand for the same tag - recorded in data_notes; the filing's arithmetic settles it in the vendor's favour here.
- The same filing says backlog on any particular date does not provide meaningful information about the timing of future revenue recognition, and that there is increased risk of order modifications, pushouts or cancellations.
- It also says failure to obtain export licences has harmed backlog and could require returning substantial deposits received from customers in China.
- 2026-09-25editedRe-struck at the 24 September 2026 close of $187.11 in the refreshed research pack; the argument is unchanged and only the price moved, up 3.6% since the thesis.
- The six-year window spans a downcycle in fiscal 2024 in which earnings fell, so a median struck across it is depressed relative to a mid-cycle multiple.
- On the annualised September-quarter diluted earnings of $1.18 a share the multiple is about 32.6 times, and on consensus fiscal 2029 earnings it is about 19.1 times.
- 2026-09-25editedRe-struck at the 24 September 2026 close of $187.11 in the refreshed research pack; the argument is unchanged and only the price moved, up 3.6% since the thesis.
- 2026-10-28editedRe-struck at the 27 October 2026 close of $153.66 with the Q1 fiscal 2027 print; the argument is unchanged and only the price moved, down 17.9% from the $187.11 it was struck at.
- The six-year median spans the fiscal 2024 downcycle, so it is struck partly on depressed earnings and partly on a lower-growth industry than the one SEMI now forecasts.
- At the 30% compound rate the panel's fiscal 2029 figure implies, $5.76 arrives in about one and three-quarter years, so the requirement is a stretch of delivery rather than an unreachable number.
- 2026-09-25editedRe-struck at the 24 September 2026 close of $187.11 in the refreshed research pack; the argument is unchanged and only the price moved, up 3.6% since the thesis.
- 2026-10-28editedRe-struck at the 27 October 2026 close of $153.66 with the Q1 fiscal 2027 print; the argument is unchanged and only the price moved, down 17.9% from the $187.11 it was struck at.
- Seven analysts three fiscal years out is a thin panel, and the pack's own gradient warns that a consensus is a vendor aggregation methodology rather than a market expectation.
- A 30% compound rate for three years would be faster than any three-year stretch in the ten-year revenue series, which includes an outright decline in fiscal 2024.
- 2026-09-25editedRe-read at the 25 September 2026 pack: the panel's fiscal 2029 figure moved from $8.045 to $8.046 and the multiple was re-struck at $187.11.
- 2026-10-29editedRe-read at 2026-10-29 after the print: the panel's fiscal 2029 figure moved from $8.046 to $8.177 and the multiple was re-struck at the $153.66 close.
- SEMI's own July 2026 forecast, published two weeks before the call, is $143.9 billion for 2026 and implies about $175 billion for 2027.
- The chief executive said on the same call, unprompted: 'We are not very good at forecasting.'
- 2026-10-29editedQ1 fiscal 2027 call: 2026 outlook raised to the mid-$150 billion range and 2027 above $200 billion
The company
A chip goes through hundreds of manufacturing steps, and a defect caught late means every later step was spent on scrap. KLA sells the machines that look at the wafer, the photomask or the finished package between steps, the software that sorts what they find, and the service that keeps them running.
Its tools go in before the production equipment for a new generation of chip, because they are used to set that equipment up. Its rivals are Applied Materials, ASML, Hitachi High-Tech, Lasertec and Onto Innovation.
The evidence under this - 3 claims
How it makes money
Mostly by selling machines: process control for chipmaking was 90.2% of fiscal 2026 revenue and 96.9% of segment profit. Once installed, each machine earns service fees; services were 23% of revenue, and management says 80% of that is under contract. Taiwan Semiconductor alone bought about 19%.
The evidence under this - 4 claims
- The 10-K discloses service revenue only in total and carries no contract-versus-transactional split anywhere, so the 80% cannot be checked against any filing.
- Management said service growth is running at the bottom of its own 13% to 15% target range this year, which is the weaker half of the same answer.
Where the money comes from, stream by stream
| Stream | Share of sales (FY2026) | How it earns it |
|---|---|---|
| Wafer Inspection | 49% | Optical and e-beam tools that find defects on wafers during chip making, sold mainly to foundries and memory makers. |
| Services | 23% | Maintenance and support on the installed base, mostly under contracts; it grows with the number of tools in the field. |
| Patterning | 20% | Metrology and reticle inspection that check the patterns printed on wafers and masks line up and measure right. |
| Specialty Semiconductor Process | 4% | Deposition and etch tools for specialty and power devices, sold outside the leading-edge logic and memory fabs. |
| PCB and Component Inspection | 3% | Direct imaging, inspection and printing systems for circuit boards and chip substrates, the former Orbotech business. |
| Other | 1% | Refurbished and remanufactured legacy systems, upgrades and enhancements for older process-control tools. |
Who runs it
Rick Wallace has run KLA since 2006 and joined it in 1988; Bren Higgins has run its finances since 2013. Insiders own less than 1% of the company. The yearly cash bonus rides on operating profit in dollars, which a busy year for the whole industry lifts, and the long-term shares on turning sales into cash better than rivals do. Nothing pays on profit per share.
The evidence under this - 3 claims
Unvested RSUs and PRSUs are excluded from the table, so the executives' exposure to the stock is larger than their owned shares; the CEO's fiscal 2026 PRSU target alone is 154,423 shares.
- Operating Margin Dollars is an absolute figure, so an equipment upturn funds the bonus whatever KLA's share; the 167.6% was paid in the year operating cash flow fell to 85.8% of net income (C07).
- Nothing pays on per-share results or return on capital; the cash test is the three-year PRSU, which is relative to peers.
Why it wins, and who could beat it
Why customers come back
A factory tunes its whole yield baseline to one vendor's measurements, and that vendor is chosen when a chip generation is defined, before production equipment arrives. Swapping midway means losing the baseline, so the choice holds for every expansion at that generation.
The stakes explain why buyers pay: over 500 steps, a per-step yield of 99.99% leaves above 95% of chips good, while 99.98% leaves about 90%. A Chinese rival's own annual report says the top five foreign vendors, led by KLA, hold above 70% of the market.
The evidence under this - 4 claims
It is a process-control vendor describing why process control matters, and the yield arithmetic is a compounding illustration rather than any named fab's measured loss.
- The market figures are Yole's and VLSI's estimates relayed by a filer, and the filer is the entrant with every reason to describe the incumbents as dominant and the domestic market as fast-growing.
- The summary states no process node at which Skyverse's tools are qualified in volume, so its growth says it is taking share in China without saying at which node.
The threat
Two threats, from opposite sides. ASML, Lam Research and Applied Materials build measurement into their own machines, so a wafer can be checked where it was just processed. And in China, export rules keep KLA out of some factories while local makers grow: Skyverse's revenue rose 48.75% in 2025, and KLA's own filings warn Chinese rivals may take share.
The evidence under this - 3 claims
- The market figures are Yole's and VLSI's estimates relayed by a filer, and the filer is the entrant with every reason to describe the incumbents as dominant and the domestic market as fast-growing.
- The summary states no process node at which Skyverse's tools are qualified in volume, so its growth says it is taking share in China without saying at which node.
The rivals: AMAT, LRCX, ASML, ONTO, and 2 more that are private, in-house or listed abroad.
Who else does this, rival by rival
| Ticker | Rival | How far along | Priced against it |
|---|---|---|---|
| AMAT · LRCX · ASML | Applied Materials, Lam Research and ASML integrated and in-situ metrologyPriced against it: One segment only | On sale | One segment only |
| Skyverse Technology (Shenzhen, 688361.SH) and other China-domestic inspection and metrology makersPriced against it: Listed abroad | On sale | Listed abroad | |
| Lasertec Corporation (Tokyo 6920)Priced against it: Listed abroad | On sale | Listed abroad | |
| ONTO | Onto InnovationPriced against it: Same product | On sale | Same product |
How big it could get
How big this could get
Chipmakers are forecast to spend $143.9B on factory equipment in 2026, up 23.1%, rising another 21.8% in 2027 and 14.1% in 2028 to $200B. An industry research firm puts inspection and measurement at about 14% of that, roughly $20.1B in 2026. KLA's process-control business sold $12.24B in fiscal 2026.
The evidence under this - 4 claims
- SEMI's own December 2025 forecast, six months earlier, put wafer-fab equipment at $126.1 billion for 2026 and $135.2 billion for 2027. The July revision raised 2026 by 14% and 2027 by about 30%, so the series is itself volatile and has recently moved a long way in one direction.
- An industry association's forecast is built from its members' own submissions, which are the same companies whose revenue it forecasts.
- Yole's figure is a paid vendor estimate reported by the trade press, not a statistical series, and the Skyverse filing quotes the same firm at $19.09 billion while quoting VLSI at $14.35 billion for 2024 - three figures from two estimators, a spread of a quarter.
- The perimeter is equipment. KLA's own segment includes service, so the market and the company are not struck on the same definition.
Two estimators' intensities differ: Yole's $18 billion against VLSI's $14.35 billion a year earlier would give a lower share, and neither is a census.
Who has signed
Backlog rose to $12.57B at 30 June 2026 from $7.86B a year earlier. No customer is named, and the annual report warns that bigger orders raise the risk of delays and cancellations, and that missing export licences could force it to return large deposits from Chinese customers.
The evidence under this - 2 claims
- The same filing says backlog on any particular date does not provide meaningful information about the timing of future revenue recognition, and that there is increased risk of order modifications, pushouts or cancellations.
- It also says failure to obtain export licences has harmed backlog and could require returning substantial deposits received from customers in China.
Where the growth is coming from, business by business
Where the growth is coming from
| Business | Share of revenue | Growth: last quarter | Last half-year | Last year |
|---|---|---|---|---|
| Defect Inspection | 48% | +0.7% | +6.1% | +25.3% |
| Service | 24% | +17.8% | +16.7% | +15.4% |
| Patterning | 21% | +31.2% | +23.2% | +12.1% |
| Specialty Semiconductor Process | 4% | -15.0% | -13.3% | +1.1% |
| PCB And Component Inspection | 2% | -13.9% | +18.9% | +25.4% |
| Other Revenue | 1% | -49.6% | -43.2% | -47.1% |
| Whole company | 100% | +7.2% | +9.9% | +17.5% |
The filed split through 2025-12-31. These rates are not carried forward: the reading below says why.
How we read it: The blend. Patterning's lead is one lumpy year, up 23% after 7%, with the March quarter 3% below a year earlier: a product line riding the same wafer-fab cycle as the whole, not a separate business.
The results
The numbers
Sales grew 12% in the year to Jun 2026, slower than the 24% in the year before.
It keeps 36% of sales as profit, up from 32% in the year to Jun 2023.
It keeps 61¢ of each sales dollar after making what it sells, and spends 11¢ of it on research, its biggest running cost.
Shareholders' money earned 87% a year, down from 139% in the year to Jun 2022.
It turned 28% of sales into free cash, $3.77B over the year.
It owes $4.24B more than the cash it holds, about 1.0 years of the cash its operations bring in.
The share count fell 3.1% in 2.0 years, and pay in shares costs 2.3% of sales.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-10-04. Fiscal years end in June.
| Jun 2022 | Jun 2023 | Jun 2024 | Jun 2025 | Jun 2026 | |
|---|---|---|---|---|---|
| Size and growth | |||||
| Sales | $9.21B | $10.5B | $9.81B | $12.2B | $13.6B |
| Sales growth | +33% | +14% | −6.5% | +24% | +12% |
| Profit | |||||
| Gross margin | 61% | 60% | 60% | 61% | 61% |
| Net margin | 36% | 32% | 28% | 33% | 36% |
| What growth costs, per sales dollar | |||||
| Research and development | 12% | 12% | 13% | 11% | 11% |
| Selling, general and admin | 9.3% | 9.4% | 9.9% | 8.5% | 8.3% |
| Pay in shares | 1.4% | 1.6% | 2.2% | 2.2% | 2.3% |
| Capital spending | 3.3% | 3.3% | 2.8% | 2.8% | 2.8% |
| Cash | |||||
| Cash from operations | $3.31B | $3.67B | $3.31B | $4.08B | $4.14B |
| Free cash flow | $3.01B | $3.33B | $3.03B | $3.75B | $3.77B |
| Balance sheet | |||||
| Cash and short-term investments | $1.58B | $1.93B | $1.98B | $2.08B | $1.65B |
| Debt | $6.66B | $5.89B | $5.88B | $5.88B | $5.89B |
| Net cash | −$5.08B | −$3.96B | −$3.90B | −$3.81B | −$4.24B |
| Returns and shares | |||||
| Shares (diluted) | - | - | 1.36B | 1.34B | 1.32B |
What growth costs
Growth has made KLA more profitable: research and office costs rise more slowly than sales, while the cost of building each tool stays about the same share. The last slow year showed the reverse, so the gain belongs to the upswing, not to the company alone.
The evidence under this - 2 claims
Capital spending fell from 3.4 to 2.8 cents per sales dollar, so part of the widening is spending that may have to return; and the ratios rose again in the one down year, so the leverage runs both ways.
- The 41.7% margin is the highest in the ten-year series in the research pack, which is what a cycle peak looks like from inside it.
- The vendor's operating income agrees with this figure while the repo's own XBRL reader returns $772,070 thousand for the same tag - recorded in data_notes; the filing's arithmetic settles it in the vendor's favour here.
The last report (28 Jul)
Quarter ended June 2026, filed figures.
In the September 2026 quarter KLA's sales grew 29% to $4.14B, above the middle of its guide, and it earned $1.22 a share on an adjusted basis. Cash from operations was $1.53B, almost all of its $1.55B profit, and customers paid in about 67.6 days, back under the 68 we watched.
It guided the December quarter to $4.7B, well above the $4.3B that would have eased our doubts. Inventories still rose by $434M, so the cash drag has eased but not gone.
On the call, management raised its 2026 chip-equipment market view to about $155B, expects more than $200B in 2027, and said memory costs hold gross margin near 62.5%.
What it did to the thesisIt answered the pre-registered test well: cash came back and the guide jumped. The call lifted the demand outlook, but buyers still pay far more per dollar of profit than KLA's norm, so the view stays Unattractive.
The quarter in detail
Sales growth from the previous quarter, filed figures.
Earnings per share $1.05 against $1.00 expected.
The evidence under this - 6 claims
Operating cash flow was 98.8% of net income, close to but not at or above it, and inventories grew by $434M in the quarter, so working capital is still absorbing cash.
- KLA sells some trade receivables: proceeds from sales rose to $86,020 thousand in fiscal 2026, so the reported receivable balance is after factoring and the underlying build is larger, not smaller.
- A quarter-end revenue denominator overstates seasonality effects; on full-year revenue the same calculation gives a smaller rise.
- 2026-10-28editedQ1 fiscal 2027 release (S20): receivables of $3,072M on revenue of $4,136M is about 67.6 days, back under the 68 line.
- Much of the gap is inventory and receivables built ahead of a guided 24.6% revenue increase, which is a use of cash that precedes revenue rather than a deterioration in earnings quality.
- Days of inventory actually fell slightly, from 242 to 235, so the inventory build tracked the business rather than outrunning it.
- The same guidance implies gross margin roughly flat on a 9.4% sequential revenue rise, which an analyst pressed on the call; management attributed it to memory-component pricing worth around 100 basis points and said it likely continues through 2027.
- Revenue growth of 24.6% would be more than double the 11.7% the company actually delivered in fiscal 2026.
- 2026-10-29editedguide period closed: the quarter landed above the midpoint; now a record of guide and result
Management gave no backlog figure on the call; one analyst noted the industry view that wafer equipment spending may peak near $300 billion, and management declined to guide 2030.
- SEMI's own July 2026 forecast, published two weeks before the call, is $143.9 billion for 2026 and implies about $175 billion for 2027.
- The chief executive said on the same call, unprompted: 'We are not very good at forecasting.'
- 2026-10-29editedQ1 fiscal 2027 call: 2026 outlook raised to the mid-$150 billion range and 2027 above $200 billion
What management did with the money
It returned $3.35B in fiscal 2026, 89% of free cash flow: $2.29B of buybacks and $1.06B of dividends. The share count fell 3.1% over two years. Debt of $5.89B is mostly covered by $4.90B of cash.
Insiders in 2026Q2: 5 purchases and 9 sales, from their Form 4 filings.
The evidence under this - 3 claims
The price, and where we disagree
In short. Full. A great business priced as if its record profits were the new floor, in an industry its own annual report calls cyclical.
The shares have fallen 39% from their June high, most of it in three sell-offs. Twice the reason was the same, after results that beat forecasts: rising memory-chip costs capping the margin. In January the finance chief called that temporary; by July management said it lasts through 2027.
The 28 October results and call showed cash catching up and a larger 2027, yet the shares fell again. The next test is the January results: whether cash and the guide hold.
The evidence under this - 5 claims
- 2026-09-25editedRe-struck at the 24 September 2026 close of $187.11 in the refreshed research pack; the argument is unchanged and only the price moved, up 3.6% since the thesis.
The same January answer held gross margin guidance at 62% plus or minus half a point and fiscal 2026 gross margin came in at 61.30%, so the transitory call was wrong about duration and close to right about size.
- Two analysts pressed on why KLA cannot price to offset it when ASML is raising prices on extreme-ultraviolet tools; management's answer was that it is hard to reprice orders already taken and that increases come at new-product introduction.
- The headwind was described as 'somewhere around 100 basis points' and then immediately widened to 'probably a little bit more', so the figure is not firm.
- 2026-10-29editedQ1 fiscal 2027 call: December gross margin guide 62.5% and the headwind still running through 2027
Management gave no backlog figure on the call; one analyst noted the industry view that wafer equipment spending may peak near $300 billion, and management declined to guide 2030.
How we read the multiple
Buyers pay about 42 times last year's profit per share, about 1.6 times KLA's six-year normal of about 27. For today's price to be merely normal:
- Profit per share must rise about 57%, from $3.66 to $5.76.
- Analysts expect $8.05 by fiscal 2029, which would do it.
- But only if this year's record profit, about 42 cents on each dollar of sales, is a floor rather than a peak.
Against its own history. Far dearer than its own normal on every measure, even after the fall. That price treats a record year as the new floor.
The evidence under this - 5 claims
- The six-year window spans a downcycle in fiscal 2024 in which earnings fell, so a median struck across it is depressed relative to a mid-cycle multiple.
- On the annualised September-quarter diluted earnings of $1.18 a share the multiple is about 32.6 times, and on consensus fiscal 2029 earnings it is about 19.1 times.
- 2026-09-25editedRe-struck at the 24 September 2026 close of $187.11 in the refreshed research pack; the argument is unchanged and only the price moved, up 3.6% since the thesis.
- 2026-10-28editedRe-struck at the 27 October 2026 close of $153.66 with the Q1 fiscal 2027 print; the argument is unchanged and only the price moved, down 17.9% from the $187.11 it was struck at.
- The six-year median spans the fiscal 2024 downcycle, so it is struck partly on depressed earnings and partly on a lower-growth industry than the one SEMI now forecasts.
- At the 30% compound rate the panel's fiscal 2029 figure implies, $5.76 arrives in about one and three-quarter years, so the requirement is a stretch of delivery rather than an unreachable number.
- 2026-09-25editedRe-struck at the 24 September 2026 close of $187.11 in the refreshed research pack; the argument is unchanged and only the price moved, up 3.6% since the thesis.
- 2026-10-28editedRe-struck at the 27 October 2026 close of $153.66 with the Q1 fiscal 2027 print; the argument is unchanged and only the price moved, down 17.9% from the $187.11 it was struck at.
- Seven analysts three fiscal years out is a thin panel, and the pack's own gradient warns that a consensus is a vendor aggregation methodology rather than a market expectation.
- A 30% compound rate for three years would be faster than any three-year stretch in the ten-year revenue series, which includes an outright decline in fiscal 2024.
- 2026-09-25editedRe-read at the 25 September 2026 pack: the panel's fiscal 2029 figure moved from $8.045 to $8.046 and the multiple was re-struck at $187.11.
- 2026-10-29editedRe-read at 2026-10-29 after the print: the panel's fiscal 2029 figure moved from $8.046 to $8.177 and the multiple was re-struck at the $153.66 close.
- The 41.7% margin is the highest in the ten-year series in the research pack, which is what a cycle peak looks like from inside it.
- The vendor's operating income agrees with this figure while the repo's own XBRL reader returns $772,070 thousand for the same tag - recorded in data_notes; the filing's arithmetic settles it in the vendor's favour here.
Priced against its rivals
Only Onto Innovation sells the same thing to the same buyers as a large part of what it does. A price comparison needs at least two, so the price is judged against its own history instead.
Private or foreign rivals that cannot be priced: Hitachi High-Tech Corporation, Lasertec Corporation.
Where we disagree with the market
An AI winner on the critical path, sold off too hard; the fall is an opportunity in a compounding business.
The business is excellent and accelerating, but three years of that is already in the price, in an industry its own filings call cyclical.
A 39% fall reads as risk removed. It is not: buyers still pay nearly twice what this company has normally cost for each dollar of a year's earnings.
We are wrong if equipment spending reaches $190B in 2027, the backlog converts without Chinese cancellations, and earnings get near $8 a share by fiscal 2029.
What analysts say, and the evidence
The evidence under this - 5 claims
- 2026-09-25editedRe-struck at the 24 September 2026 close of $187.11 in the refreshed research pack; the argument is unchanged and only the price moved, up 3.6% since the thesis.
- The six-year window spans a downcycle in fiscal 2024 in which earnings fell, so a median struck across it is depressed relative to a mid-cycle multiple.
- On the annualised September-quarter diluted earnings of $1.18 a share the multiple is about 32.6 times, and on consensus fiscal 2029 earnings it is about 19.1 times.
- 2026-09-25editedRe-struck at the 24 September 2026 close of $187.11 in the refreshed research pack; the argument is unchanged and only the price moved, up 3.6% since the thesis.
- 2026-10-28editedRe-struck at the 27 October 2026 close of $153.66 with the Q1 fiscal 2027 print; the argument is unchanged and only the price moved, down 17.9% from the $187.11 it was struck at.
- SEMI's own July 2026 forecast, published two weeks before the call, is $143.9 billion for 2026 and implies about $175 billion for 2027.
- The chief executive said on the same call, unprompted: 'We are not very good at forecasting.'
- 2026-10-29editedQ1 fiscal 2027 call: 2026 outlook raised to the mid-$150 billion range and 2027 above $200 billion
- Seven analysts three fiscal years out is a thin panel, and the pack's own gradient warns that a consensus is a vendor aggregation methodology rather than a market expectation.
- A 30% compound rate for three years would be faster than any three-year stretch in the ten-year revenue series, which includes an outright decline in fiscal 2024.
- 2026-09-25editedRe-read at the 25 September 2026 pack: the panel's fiscal 2029 figure moved from $8.045 to $8.046 and the multiple was re-struck at $187.11.
- 2026-10-29editedRe-read at 2026-10-29 after the print: the panel's fiscal 2029 figure moved from $8.046 to $8.177 and the multiple was re-struck at the $153.66 close.
What could go wrong, and how we would know
The risks
- A cyclical peak. Sales depend on chipmakers' equipment budgets, which KLA's own annual report calls historically cyclical, and record margins are where cycles peak.
- China. China was 29.8% of revenue, down from 42.8% in fiscal 2024, and export rules help local rivals take share.
- Costs outrunning sales. In the main business, costs, research and selling all grew faster than the 11.9% rise in sales, so its margin slipped.
- Pricier memory chips. Management expects dearer memory components to cut gross margin by around 1 percentage point, probably a little more, through 2027.
The evidence under this - 6 claims
- The 41.7% margin is the highest in the ten-year series in the research pack, which is what a cycle peak looks like from inside it.
- The vendor's operating income agrees with this figure while the repo's own XBRL reader returns $772,070 thousand for the same tag - recorded in data_notes; the filing's arithmetic settles it in the vendor's favour here.
- Consolidated gross margin rose over the same period, from 60.91% to 61.30%, because the PCB segment improved - so the company-level trend and the core-segment trend point in opposite directions.
- The segment figures are GAAP as tagged and exclude the unallocated items management discusses its business on.
- Two analysts pressed on why KLA cannot price to offset it when ASML is raising prices on extreme-ultraviolet tools; management's answer was that it is hard to reprice orders already taken and that increases come at new-product introduction.
- The headwind was described as 'somewhere around 100 basis points' and then immediately widened to 'probably a little bit more', so the figure is not firm.
- 2026-10-29editedQ1 fiscal 2027 call: December gross margin guide 62.5% and the headwind still running through 2027
What we still do not know
- How big 2027 is. Management's $190B for equipment spending sits about 8% above the industry forecast, and nothing filed settles which is right.
- How much of the $12.57B backlog is Chinese orders that a refused export licence could still cancel.
The evidence under this - 4 claims
- SEMI's own July 2026 forecast, published two weeks before the call, is $143.9 billion for 2026 and implies about $175 billion for 2027.
- The chief executive said on the same call, unprompted: 'We are not very good at forecasting.'
- 2026-10-29editedQ1 fiscal 2027 call: 2026 outlook raised to the mid-$150 billion range and 2027 above $200 billion
- Management described $190 billion as a consensus view held by others rather than as its own forecast, and noted more bullish views exist.
- The two figures may not have identical definitions: SEMI's wafer-fab equipment includes mask/reticle and fab facilities equipment, and no definition was stated on the call.
- The same filing says backlog on any particular date does not provide meaningful information about the timing of future revenue recognition, and that there is increased risk of order modifications, pushouts or cancellations.
- It also says failure to obtain export licences has harmed backlog and could require returning substantial deposits received from customers in China.
What would change our view
- Unattractive→AttractiveBack to its own normal multiple
It needs:
- ○The shares fall to about 26.7 times earnings, their own six-year norm, from 42.0; rising earnings get there as surely as a falling price.about $97.68 today, 48% below the last close
- Reopens the researchThe cycle proves it is not a peak
It needs:
- ○Operating margin holds above 37% as sales rise, backlog stays above $10B, and process control keeps outgrowing equipment spending.checked at the next results, due by 27 Jan
What is coming
- 28 October 2026 Q1 fiscal 2027 results: Whether operating cash flow catches up with net income and receivable days fall, against the $4.0 billion revenue guide, and what December is guided to.
Every dated event ahead
- 28 October 2026 Q1 fiscal 2027 results: Whether operating cash flow catches up with net income and receivable days fall, against the $4.0 billion revenue guide, and what December is guided to.
- 4 November 2026 Annual meeting and say-on-pay vote: How holders vote on a pay plan whose bonus paid 167.6% in a year cash conversion fell; last year's support was about 92%.
- January 2027 Q2 fiscal 2027 results: Whether memory costs keep weighing on gross margin, and management's first full read on 2027 wafer-fab equipment against the roughly $190 billion it cited.
Since we wrote this
- 29 OctEarningsThe call lifted the 2027 outlook; Unattractive heldManagement raised its equipment market view and expects more than $200B in 2027, but the shares still cost 42.0 times earnings against a norm of 26.7, so the view held.
- 28 OctEarningsCash flow recovered, December guide $4.7B; price re-struckThe working-capital drag eased and the guide rose, but the shares still cost 42.0 times earnings against a norm of 26.7, so Unattractive and Medium hold.
- 5 OctThesis reviewThe long case now sits under the verdictAfter the five-name pilot, the owner asked for the long case on every thesis.
- 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the proxy and filings.
- 25 SepThesis reviewIts market sized at about $20 billion. View heldProcess control came to roughly $20.1B of 2026 equipment spending, and both sell-offs after beats had one cause, memory costs. No results landed, and earnings now need to rise 92%.
- 11 SepThesis writtenUnattractive, medium convictionWritten six weeks after June-quarter results that beat forecasts still sent the shares down 10.8% in a session.
Also checked 1 time with nothing new to act on.
Sources 21
Sources - 21
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S01primarycompany filing2026-08-06the statements, the segment note and its reconciliation, backlog, customer and geographic concentration, the debt schedule, the buyback and dividend detail, the export-control risk factors and the Competition sectionbacksC01C02C03C04C05C06C08C12C19C20C22C24C30C34C35C37C38C39C40C42C49C50C54C56C57
- S02primarycompany filing2026-07-28September-quarter guidance, the quarterly statements, the GAAP-to-non-GAAP gap and the capital-returns totalbacksC06C11C38C40
- S03primarytranscript2026-07-28KLA Q4 fiscal 2026 earnings call transcriptmanagement's wafer-fab equipment numbers, the memory-component margin headwind, lead times, the service contract share and the advanced-packaging figure; separately, eleven analysts' questions, of which four pressed on why price cannot offset cost - a different origin from the answers
our view of 2027, there's a consensus view out there, I would say, that's somewhere in and around the $190 billion range ... more or less, you're in and around $190 billion or so translates into a mid-20s type growth rate ... 80% of revenue is contract-based, so it provides a nice anchor for the company in terms of visibility moving forward ... we have a new long-term target model for service growth of 13% to 15%
backsC14C21C31C32C41C44 - S04secondaryvendor data2026-09-25Research pack, frozen at content/KLAC/research-pack.jsonprice, the 252-day range and returns, the six-year median multiples, the analyst split and estimates, FINRA short interest and the Form 4 summary; refreshed 2026-09-25 at the 24 September close, with the daily closing series behind the repricing episodesbacksC10C27C28C43
- S05primarycompany filing2026-02-17a competitor's own statement that process control is concentrated among few participants, and its naming of in-situ metrology from the process-tool makers and of local Chinese entrantsbacksC16C17
- S06primarycompany filing2026-02-24the sequencing fact that process control goes in before the process equipment at a new node, the only listed pure-play peer's scale, and KLA named as its principal competitorbacksC18C30
- S07primaryindustry data2026-07-14the externally measured base and rate behind machine.prize and machine.current, and the 2027 figure that management's own number exceeds. Read on the wire that distributed it because semi.org returns 403 to curl and to an agent fetch alike
the Wafer Fab Equipment (WFE) segment...is projected to increase 23.1% to $143.9 billion in 2026 ... WFE segment sales are projected to expand 21.8% in 2027 and 14.1% in 2028 to reach the $200 billion mark ... DRAM equipment sales are projected to rise 39.0% to $38.8 billion in 2026, followed by 27.4% growth in 2027 and 15.0% growth in 2028, reaching $56.9 billion
backsC13 - S08primarycompany filing2025-12-12Applied Materials Form 10-K for the fiscal year ended October 26, 2025 AMATthat process control is not a reported line in any of its three segments, and that its Competition section does not name KLA backbacksC30
- S09primarycompany filing2026-02-25that its metrology and inspection systems are framed as a component of holistic lithography and carry no separate reported revenue - the perimeter judgement that kept it out of the cohortbacksC30
- S10tertiarynews2026-07-29Barron's - KLA Stock Heads for Worst Month Since 1987. Earnings Aren't Doing It Any Favors.evidence that the drawdown followed results that beat guidance - used only for the observed common view, never for a machine claimread, but no claim rests on it
- S11primarycompany filing2026-04-24the China entrant's own revenue and growth, and its relay of Yole's and VLSI's process-control market sizes, the top-five share and the yield arithmetic. Read on cninfo, the exchange's disclosure platform, after the company's own site failed to resolve at the thesisbacksC46C48
- S12secondaryindustry data2026-06-22Electronics Weekly - Metrology and Inspection equipment market tops $18bn (Yole Group analysis) Yole Groupan outside estimate of process control's share of wafer-fab equipment, the figure that composes the prize for the first time. Read directly from the page, not through an agent's summary
The metrology and inspection equipment market exceeds $18bn as AI, HBM and advanced packaging drive demand for process control, says Yole Group. The market grew 15% in 2025 and accounts for approximately 14% of total wafer fab equipment spending. ... The market is expected to grow to about $25bn by 2031. ... KLA, Applied Materials and Lasertec remain market leaders, while competition intensifies across specialised process-control segments.
backsC45 - S13primarytranscript2026-01-29KLA Q2 fiscal 2026 earnings call transcript, question sectionthe January objection in the analysts' own questions - whether higher memory-chip costs can be passed on - and the finance chief's answer that it was transitory
We think the memory situation is transitory And and really as it relates to how we we price our products, it's much more of a a value oriented pricing model
backsC44 - S14tertiarynews2026-07-01the stated reason for the 1 July fall - group profit-taking, with no company news - used only in prior_repricing
fell 12.3% in the afternoon session after the company was caught in a broader sell-off among semiconductor stocks as investors took profits following a record-breaking first half for the sector
read, but no claim rests on it - S15primarycompany filing2026-04-30the March-quarter revenue by major product category, used to check the vendor's product split, which stops at the December 2025 quarterbacksC49
- S16primarycompany filing2026-09-22executive officers and tenure, beneficial ownership as of September 9, 2026, the bonus plan's Operating Margin Dollar grid and payout, the PRSU metric and its percentiles, the meeting date and its proposalsbacksC51C52C53C58
- S17primarycompany filing2022-08-05fiscal 2021 and 2022 statements, the June 2022 accelerated share repurchase and the debt raised alongside itbacksC54C57
- S18primarycompany filing2024-08-05fiscal 2023 and 2024 statements, read through the frozen numbers cardbacksC54C57
- S19primarycompany filing2019-08-16the Orbotech acquisition's closing date and purchase considerationbacksC56
- S20primarycompany filing2026-10-28the September-quarter statements and cash flow, the December-quarter guidance and capital returnsbacksC11C59
- S21primarytranscript2026-10-28KLA Q1 fiscal 2027 earnings call transcript (FMP), held 2026-10-28data/cache/transcripts/KLAC-2027Q1.jsonthe calendar 2026 and 2027 wafer equipment outlook, the gross margin guide and memory-cost headwind, lead times, China share, capital spending, memory mix; analysts from Goldman Sachs, JPMorgan, Cantor, UBS, TD Cowen, Bernstein, Wells Fargo, Deutsche Bank, Wolfe, Morgan Stanley and Redburn
We currently expect the wafer equipment market to exceed $200 billion in calendar 2027.
backsC14C31C60