KLA Corporation

Refresh delta (since 2026-06-17)

What changed

1. Backlog $7.86B → $12.57B (+60%) over FY26, "due to strong demand driven by the AI infrastructure buildout" [10-K FY26, Backlog]. Lead times are ~12 months company-wide and 18–24 months on some products. This is the single most important new fact. Growth through FY27 is now visible from orders already in hand.
2. FY26 closed at $13.58B revenue (+11.7%), $4.83B net income (+18.9%), GAAP EPS $3.66, non-GAAP EPS $3.76 [10-K FY26; 8-K Q4]. Q4 came in at $3.66B, above the midpoint of the $3.575B guide. Q1 FY27 is guided to $4.0B ±$0.2B (+9% QoQ at the midpoint) with non-GAAP EPS of $1.16 ±$0.10 [8-K Q4].
3. CY2026 WFE raised to "approximately the low $150 billion range", up from "$140 billion-plus". Advanced-packaging process control is now guided to ~$1.1B in CY26 (+70%+), against a prior "high 50s%". The CFO cited a street consensus of ~$190B for CY27 WFE (mid-20s% growth). He cited it; he did not guide to it.
4. Margins: the memory-cost headwind got longer. The ~100bps gross-margin hit from buying DRAM at higher prices "likely continues as we move forward through 2027". In June it was framed as "through CY2026". FY26 GAAP gross margin was 61.3% (FY25: 60.9%) [10-K FY26].
5. Share price: $237.33 on 2026-06-16 → ~$188 on 2026-09-22 (−21%). The fall came despite a Q4 beat. The day after the print the stock closed −6.2% at $190.80 and traded another −7.7% after hours, on margin mix and expectations. August added export-control worries and a rotation out of semi-cap stocks. September added a sector-wide AI-capex scare (about −6% on 2026-09-14).
6. June dossier errors, corrected here. (a) The June dossier's FY26 revenue base of $14.2–14.4B assumed "high-teens growth". That guide was calendar-year language, not fiscal, and FY26 actually came in at $13.58B. The EPS base of $3.75 was right. (b) The June 3Y price target of "$300" did not follow from its own arithmetic: $5.20 × 38–42 = $198–218. Lens 11 below rebuilds it.

What held. The structural thesis is unchanged. Wafer inspection still has 49% of revenue and grew +7% [10-K FY26]. Services are 23% of revenue, up +16.5%. Patterning grew +23%. KLA is described on the call as ~6× the size of its nearest process-control rival. The supply chain is unchanged and so is the moat.

Phase A — Understand the business

Lens 1 · Overview: carried, see the previous dossier. Updated facts from the 10-K:

• ~17,000 regular full-time employees at 2026-06-30 [10-K FY26].
• One customer was ~19% of total revenue in both FY26 and FY25 (13% in FY24) [10-K FY26]. The customer is not named. It is almost certainly TSMC [inference].
• International revenue was 87% of the total [10-K FY26].

Lens 2 · Supply chain: carried, unchanged; see the previous dossier.

Lens 3 · Moats: carried. One addition: services grew +16.5% while systems grew +10.4%. The installed-base annuity is compounding faster than the tool business [10-K FY26: service $3.126B vs $2.683B; product $10.454B vs $9.473B].

Segments (re-run: the mix moved)

By reportable segment, FY26 [10-K FY26]

By product line, FY26 [10-K FY26]: Wafer Inspection $6.631B (49%, +7.0%) · Patterning $2.707B (20%, +23.2%) · Service $3.126B (23%, +16.5%).

By geography, FY26 [10-K FY26]

The China read. China's share fell 43% (FY24) → 33% (FY25) → 30% (FY26) [10-K FY26 risk factors], but China dollars were flat at ~$4.05B. The share decline is growth happening elsewhere, not a China collapse. The Q4 quarterly mix was Taiwan 31%, China 26%, North America 18% and Korea 10%.

The memory mix is rising. Memory was 21% of semiconductor process-control systems revenue in Q4 (about 90% of that DRAM). The Sep-quarter guide assumes 27%. That is up from ~18% in Q3.

Phase B — Measure performance

Earnings result (latest print: Q4 FY26, quarter ended 2026-06-30, reported 2026-07-28)

[8-K Q4 unless noted]

• Revenue $3.658B: +15% YoY (Q4 FY25: $3.175B) and +7% QoQ (Q3: $3.415B). It beat the $3.575B guide midpoint and the ~$3.60B consensus.
• Non-GAAP EPS $1.05 against a ~$1.00 consensus. GAAP EPS was $1.04. GAAP net income was $1.363B.
• Non-GAAP gross margin ~62.4%, at the top of guidance. GAAP was ~61.4%, derived from cost of revenue of $1.413B. Non-GAAP operating margin was 43.7%.
• Revenue mix: product $2.837B and service $0.820B (+17% YoY). PCB & Component Inspection was $241M, +56% YoY. Patterning was $728M, +61% YoY.
• Cash: OCF $906M, capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. $89M, FCFFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. $817M. Returns were $876M ($571M buyback + $305M dividends).
• Guide for Q1 FY27: revenue $4.0B ±$0.2B, non-GAAP GM 62.5% ±1pt, non-GAAP EPS $1.16 ±$0.10, opex ~$690M, tax rate 14.5% [8-K Q4; web: Investing.com].
• Reaction: −6.18% in the regular session to $190.80, then −7.74% after hours to $176.04. The June dossier's warning played out: at this valuation, a beat is the baseline expectation. This time the beat-and-raise ran into a margin story, and the stock fell.

Full year FY26 [10-K FY26]

• Revenue $13.579B (+11.7%).
• GAAP GM 61.3% (FY25: 60.9%, FY24: 60.0%).
• GAAP operating margin ~41.7% [derived: revenue − COGS $5.255B − R&D $1.532B − SG&A $1.132B].
• Net income $4.831B (+18.9%).
• GAAP EPS $3.66 (+20.4%); non-GAAP EPS $3.76 (FY25: $3.33).
• R&D $1.532B (11.3% of revenue, +12.6%).

Earnings calls (the tone is still confident, and supply is the constraint)

• Wallace (verbatim, prepared remarks): "We expect momentum across our business to accelerate through the second half of calendar 2026 and for significant growth to continue in calendar 2027."
• Wallace (verbatim): leading-edge designs "consuming a growing percentage of new wafer starts are driving greater demand for process control." This restates the complexity thesis.
• Higgins on 2027 WFE: "There's a consensus view out there … somewhere in and around the $190 billion range." He added "we feel pretty good about the company's ability to execute and deliver on our piece of that."
• Higgins on memory costs: the hit was "somewhere around 100 basis points" and "likely continues as we move forward through 2027."
• Wallace (per a low-reliability summary): "I have very few concerns about what's going to happen in 2027."

Tone trend, two calls: April was "WFE is literally just as fast as we can go". July was "accelerating … significant growth to continue in 2027". Both calls frame the constraint as supply, not demand. The one thing that got worse is the timeline on the margin headwind. A proper multi-call series still needs transcripts on the shelf (see Open items).

Lens 7 · Comps: not re-run. One snapshot update: fwd P/E ~34 and trailing ~50 at ~$184–188. In June the figures were ~35 fwd and ~72 trailing. Peer multiples were not re-pulled. A consistent dated peer snapshot is still an open item.

Stock-price catalysts (moves of 5% or more since June)

The pattern the June dossier named still holds. In order of weight, the stock moves on (1) China policy headlines, (2) the AI-capex narrative and (3) the size of guidance raises. This quarter added (4) the gross-margin trajectory. The market punished a beat because the memory-cost drag was extended.

Phase C — Judge people & books

Management and capital allocation (re-run: new numbers)

• The CFO is Bren Higgins, confirmed from the call transcript. The June dossier's "Brent" was a typo.
• FY26 capital returns: $3.348B ($2.290B buybacks + $1.058B dividends) [8-K Q4; 10-K FY26 cash flow]. That is 89% of $3.77B FCF, just under the ">90%" target.
• Share count: 1,320.2M → 1,307.0M outstanding (−1.0%) [10-K FY26]. At these prices the buyback mainly offsets dilutionDilutionIssuing new shares, so each existing share owns a smaller slice of the same company. plus a little.
• $9.74B authorization remains at 2026-06-30, including the $7.0B added 2026-03-11 [10-K FY26].
• Dividend: $0.23 per quarter, the 17th consecutive annual increase. Another $0.23 was declared 2026-08-06, payable 2026-09-01 [8-K 2026-08-06].
• Insider selling: Wallace sold 87,568 shares at ~$198.95 in August. That is about $17M. It is consistent with the routine 10b5-1Rule 10b5-1 planA schedule an insider sets in advance saying when their shares will be sold, so the sales cannot be timed on private news. A sale made OUTSIDE such a plan is a live decision, which is why it carries more information. selling seen before. Treat it as a flag to verify, not a signal.

Forensic red flags (re-run: filings now on the shelf)

• Cash conversion lagged earnings. Net income rose +19% but OCF rose only +1.5% ($4.082B → $4.143B), and FCF was flat ($3.747B → $3.767B) [10-K FY26]. The FCF margin fell from ~30.8% to ~27.7%.
• Working capital absorbed the difference:
• Receivables +$625M (+28%)
• Inventory +$436M (+14%)
• Other current assets +$214M
• Rough DSO on quarterly revenue: ~65 days → ~72 days [derived: AR / Q4 revenue × 91]

The benign read is the more likely one. Q4 revenue was +15% YoY and back-end-loaded, backlog grew $4.7B, and deferred system revenue also rose (+$116M). That is a book-and-ship ramp, not channel stuffing. Still, this is the one number to watch next quarter. If DSO keeps rising while Q1 revenue steps up to $4.0B, the flag becomes real.

• SBC: $310M (+17%), 2.3% of revenue [10-K FY26]. Modest. The GAAP-to-non-GAAP EPS gap is small ($3.66 vs $3.76), mostly acquisition-related amortization. Purchased intangibles fell to $256M from $445M.
• Balance sheet: $5.89B long-term debt against $4.90B cash plus marketable securities, so net debt is ~$1.0B. Equity rose to $6.35B from $4.69B [10-K FY26]. Goodwill of $1.79B is stable, with no impairment.
• Legal: 10-K Item 3 incorporates Note 14. It describes only ordinary-course commercial, IP and employment claims, names no specific material matter, and says the amounts provided are adequate [10-K FY26].
• Regulatory: re-run 2026-09-25 (regulatory/regulatory-findings.md). No SEC Litigation Releases and no AAERs naming KLA since 2021-09-25. The AAER query ran cleanly this time, which closes the June open item.

Phase D — Project & stress-test

Forward projection (post-split non-GAAP EPS, June fiscal year)

Anchors:

• FY26 actual: $3.76 [8-K Q4].
• Q1 FY27 guide: $1.16 ±$0.10 on $4.0B [8-K Q4].
• Backlog of $12.57B covers ~3 quarters of FY27 revenue at the Q1 run-rate [10-K FY26].
• Management says H2 CY26 accelerates and 2027 sees "significant growth".
• Consensus FY27 revenue is ~$17.85B, from a Zacks figure relayed through an aggregator, so low reliability.

Against June: the FY27 base moves $4.50 → ~$4.90 (+9%) and FY28 moves $5.20 → ~$5.70. The bear case is now floored higher, because backlog locks in most of FY27.

Forecast update (not logged, unattended run). June's candidate claim was "FY27 non-GAAP EPS ≥ $4.40, p≈0.62". Given the Q1 guide (a $4.64 annualized run-rate before any growth) and the backlog, p is now ~0.85. Log it via our model in an attended pass.

Bull vs bear

Bull, updated. Everything the June bull case needed has happened or been guided:

• Process control keeps taking a rising share of the equipment dollar.
• Advanced packaging became a ~$1.1B leg (+70%).
• Memory is broadening the mix (21% → 27%).
• A $12.6B backlog removes most FY27 demand risk.

Meanwhile the price fell 21%. Before, the valuation rested on growth that had to land. Now the growth is largely in the backlog. On FY27E base EPS, KLAC is ~38× at $188, where in June it was ~53× on the old FY27E at $237.

Bear, updated.

1. The margin story is the new crack. The memory-cost drag extends through 2027, and the mix shift toward memory customers (27%) is margin-dilutive. The market has already shown it will punish a beat on margin alone.
2. The CY27 debate is live. The street sits at ~$190B WFE. Management cited the number but won't own it. An AI-capex wobble like September 14th's hits exactly this assumption.
3. China is flat, not shrinking. Flat looks fine today, but $4.05B, 30% of revenue, is still exposed to a BIS rule change. April's peer disclosure of an export-control headwind shows the mechanism is still working.
4. Cash conversion lagged earnings in FY26 (Lens 10).

Pre-mortem, refreshed. The most likely route to a broken thesis within 18 months: CY27 WFE comes in near $160–170B instead of $190B, and FY28 guidance goes flat. Memory costs hold GM near 61%. A stock priced at ~38× a FY27 number that then stalls re-rates toward the ~30× sector multiple, which puts the bear value near ~$130.

Devil's advocate (short)

• "The backlog makes FY27 safe." Backlog is orders, not revenue. Lead times of 12–24 months mean customers are reserving slots. In a digestion year, reservations get pushed out, and the 10-K itself lists "the readiness of customer fabs" and "changes in the estimated versus actual start time of customers' projects" as timing risks.
• "It's cheaper now." The stock is still ~50× trailing and ~34× forward, above AMAT and below ASML. Cheaper than June is not the same as cheap.
• "Services compound." They do, at +16.5%, but they are still 23% of revenue. 77% is still tool sales.
• The single permanent-impairment scenario is unchanged: China localization over 7–10 years, with a Taiwan disruption as the low-probability cliff.

Management questions

Carried from the previous dossier. Three new questions go to the front:

1. FY26 FCF was flat while net income rose 19%. How much of the $625M receivables build is Q4 timing, and what DSO should we expect as revenue steps to $4B+?
2. Memory costs are now a ~100bps drag through 2027. Is there a contractual or sourcing fix, or is this simply the price of the DRAM super-cycle until it turns?
3. Of the $12.57B backlog, how much is scheduled beyond 12 months, and what share sits with the one ~19% customer?