KLAC · Analyze — Financebot

Price: $185.04 · Market cap: $241.8B · Sector: Semiconductors (process control) Verdict: 🟠 HOLD — fully valued, do not add here · Conviction [7.5]

Prior file: analyze-2026-07-28.md — written the afternoon before the FQ4 FY2026 print, with no frontmatter and no recheck. It was therefore invisible to kb.py stale and to /thesis-drift for 56 days. It is used here as context, not as a baseline — this is a fresh pass. A claim-by-claim survival ledger is in §8. All per-share figures are post the 10-for-1 split effective 2026-06-11.

1. What actually happened at the print

The prior file was gated on one event. It has fired.

FQ4 FY2026 (quarter ended 2026-06-30, reported 2026-07-28)	Result
Revenue	$3.66B — +7% QoQ, +15% YoY, above guidance midpoint, beat $3.60B consensus
GAAP diluted EPS	$1.04 (upper end of guide)
Non-GAAP diluted EPS	$1.05 vs $1.00 consensus — beat
Gross margin	62.4% — expanded, on services mix and scale, despite memory pricing and tariffs
Free cash flow	$817M in the quarter

September-quarter guide (FQ1 FY2027)	Value	Implied
Revenue	$4.00B ± $0.20B	+9% QoQ, +24.6% YoY (FQ1 FY26 was $3.21B)
Non-GAAP gross margin	62.5% ± 1.0pt	still expanding
Non-GAAP diluted EPS	$1.16 ± $0.10	+10% QoQ

And yet the stock fell 10.8% the next day ($190.58 → $170.00 on 2026-07-29, on 20.4M shares). The reason given across coverage was tone, not numbers: the guide offered "only limited upside," and management said memory pricing remains a headwind that could persist through 2027 even while calling for broad-based 2027 investment across foundry/logic, DRAM, NAND and advanced packaging. CEO Rick Wallace: "trends driving our growth are strengthening, and we see momentum accelerating in the second half of calendar 2026 and continuing through 2027."

The verdict on the prior file's central bet: it said "the risk is the ESTIMATE, not the multiple." That framing was correct — and the estimate moved in the bull direction. FY2027 consensus non-GAAP EPS is now ~$5.43, revised up 7.7% in the last 30 days, versus the ~$5.13 the prior file was worried about. The cautious commentary hit the multiple, not the numbers. The prior file was right about the mechanism and wrong about the direction.

Price path since: $170.00 (7/29) → $209.13 (8/13) → $167.36 (9/16) → $185.04 today. A 25% round trip inside eight weeks on a name whose estimates only went up. This is a sentiment-driven tape, not an information-driven one.

2. General Health (§1) — the scorecard

FY ends June 30. Base year FY2020 for the 6-year window; FY2021 shown for the 5-year.

	FY2020	FY2021	FY2022	FY2023	FY2024	FY2025	FY2026
Revenue ($M)	5,806	6,919	9,212	10,496	9,812	12,156	13,579
Operating income ($M)	—	—	3,652	3,995	3,636	5,014	5,661
Net income ($M)	1,217	2,078	3,322	3,387	2,762	4,062	4,831
Operating cash flow ($M)	1,779	2,185	3,313	3,670	3,309	4,082	4,143
Capex ($M)	153	232	307	342	277	335	376
Free cash flow ($M)	1,626	1,953	3,005	3,328	3,031	3,747	3,767
Diluted shares (M, post-split)	1,580	1,554	~1,430	1,400	1,360	1,338	1,320
Diluted EPS ($, post-split)	0.770	1.337	2.19	2.42	2.03	3.04	3.66
FCF/share ($)	1.03	1.26	2.10	2.38	2.23	2.80	2.86
Gross margin	—	—	61.0%	59.8%	60.0%	60.9%	61.3%
Operating margin	—	—	39.6%	38.1%	37.1%	41.2%	41.7%

1.1 FCF and its CAGR

Metric	6-yr CAGR (FY20→FY26)	5-yr CAGR (FY21→FY26)
Revenue	15.2%	14.4%
Net income	25.8%	18.4%
Free cash flow	15.0%	14.0%
Revenue/share	18.7%	17.6%
FCF/share	18.5%	17.8%
Diluted EPS	29.7%	22.3%
Diluted share count	−3.0%/yr	−3.2%/yr

Per-share compounding of ~18%/yr over six years with a 61% gross margin is a genuinely excellent record. But the headline hides the one thing worth arguing about.

1.2 ⚠️ The finding that matters: FCF conversion is decaying

	FY2024	FY2025	FY2026
Net income	$2,762M	$4,062M	$4,831M
Free cash flow	$3,031M	$3,747M	$3,767M
FCF / net income	110%	92%	78%
FCF YoY growth	—	+24.5%	+0.5%
Net income YoY growth	—	+47.1%	+18.9%

In FY2026 net income grew 18.9% and free cash flow grew 0.5%. The gap is working capital — the company's own disclosure attributes it to higher accounts receivable and inventory. That is the ordinary signature of an equipment maker building ahead of a shipment ramp, and it is not, on one year, evidence of earnings quality problems. But it is a three-year monotone decline (110 → 92 → 78) and it is the single reason the 15.0% FCF CAGR is so far below the 25.8% net-income CAGR. Flag it, watch it, don't panic on it. If FQ1 FY2027 shows conversion below ~75% again, the "FCF is the backbone" case weakens materially and the fair value in §4 comes down.

1.3 Capital allocation of FCF (FY2026)

Use	$M	% of FCF	Comment
Buybacks	2,495	66%	Bought heavily into a re-rating — see below
Dividends	1,058	28%	22% GAAP payout, 0.50% yield, raised again (FY26 DPS $0.81 vs $0.68 FY25, +18.6%)
Debt repaid	2	0%	No deleveraging needed
M&A	0	0%	No empire-building — third consecutive year of ~zero acquisitions
Total returned	3,553	94%	Funded partly from cash (cash $2.08B → $1.65B)
R&D (above the line)	1,532	—	11.3% of revenue
Capex (above the line)	376	—	2.8% of revenue — asset-light

Grade: A−. The deduction is for buyback timing, not for the policy. $2,495M of gross repurchase reduced the diluted count only 1.34% (1,337.5M → 1,319.6M) because the average purchase price was high and stock comp offset a large part of it. The six-year average reduction is −3.0%/yr; FY2026 achieved less than half of that for a record dollar amount. This is a standing risk for any buyback-heavy compounder at a cycle high, and it is worth saying plainly: the prior file's "share count −4.1%/yr" claim is not supported by the FY2026 filing.

1.4 Balance sheet

Metric	Value	Read
Total debt	$6.15B	Flat YoY
Cash	$1.65B	Down from $2.08B (funded returns)
Net debt	$4.50B	—
Debt / Assets	34.3%	Comfortable
Total liabilities / assets	64.6%	Elevated only because equity is small
Debt / Equity	96.9%	⚠️ Looks alarming, is not. Book equity is only $6.35B because ~$20B of cumulative buybacks has been charged against it. This is a capital-return artefact, not leverage.
Net debt / EBITDA	0.74x	The number that actually matters
Interest coverage (EBIT / interest)	19.9x	Very safe
Current ratio	2.88	Fine

No balance-sheet risk. The same buyback that compressed book equity also makes Graham's IV meaningless here — see §4.

3. Moat (§2)

3.1 Quantitative base — the moat is getting stronger, not eroding

	FY2024	FY2025	FY2026
Gross margin	60.0%	60.9%	61.3% (FQ4 alone 62.4%, FQ1 FY27 guided 62.5%)
Operating margin	37.1%	41.2%	41.7%
ROIC	—	39.4%	41.6% ↑
Return on capital	—	41.3%	43.6% ↑
ROE	—	100.8%	87.5% (falls only as equity rebuilds)
Incremental operating margin	—	58.8%	45.4% — above the 41.7% average

Two of these deserve emphasis:

ROIC rose while invested capital rose 16% ($10.58B → $12.24B). KLA is deploying more capital and earning more on it. Set this against the field's other finding: AMAT's ROIC fell 44% → ~33% on invested capital +72% against operating income +6.6% (Knowledge/Themes/semi-equipment-and-test.md). KLA is the opposite case inside the same field, and that is the cleanest quantitative argument for owning KLA rather than AMAT.
Incremental operating margin of 45.4% is above the 41.7% average, which means the next dollar of revenue arrives at a better margin than the average dollar. Margin expansion is not exhausted. (The inverse of this test is the repo's incremental-margin decay pattern; KLA passes it.)

3.2 Adversarial stress-test — "I am a well-funded rival. Can I take this?"

No, and the numbers now say so quantitatively. KLA holds ~58% share of process control and runs at roughly 6x its nearest competitor. To attack it you would need, simultaneously:

Decades of defect-library data. Process control is pattern recognition against a corpus of known defect signatures accumulated across every node at every customer. This is a data moat, and data moats compound with time rather than with spend. A rival with unlimited money still starts with an empty library.
Fab-floor trust at the leading edge. The tool tells the fab whether to scrap a $100k wafer lot. Nobody runs a second opinion from a new vendor on a leading-edge ramp.
Switching costs at the recipe level. Inspection recipes are qualified per-layer, per-node, per-fab.
Efficient scale. Process control is ~5-7% of WFE spend — large enough to be lucrative, too small to attract a full-scale assault from a rival that must build all four of the above from zero.

ASML does not compete here. Applied Materials and Hitachi High-Tech have tried for years without displacing KLA — and in 2025 AMAT lost served-market share while LRCX/KLAC/ASML gained. Moat sources, named explicitly: intangibles (defect-library IP + accumulated data), switching costs, efficient scale. Network effects and cost advantage are not the drivers.

Verdict: the moat is intact and quantitatively widening. The threat to this stock is cyclical, geographic and valuation-based. It is not competitive.

3.3 Revenue-stream map

Stream	Character	Durability	Risk
Semi process control systems — foundry/logic (62% of PC systems)	Capex-cyclical, node-driven	High — process-control intensity rises with every node shrink and with GAA/backside power	Leading-edge capex digestion
Semi process control systems — memory (38% of PC systems)	Capex-cyclical, far more volatile	Medium — HBM/DRAM complexity is genuinely raising PC intensity	Memory pricing headwind, guided to persist through 2027
Services	Recurring, installed-base-linked, growing	Highest — this is the ballast, and it drove the FQ4 margin beat	Erodes only if the installed base shrinks
Specialty semiconductor + EPC/components	Mature, low-growth	Medium	Non-core

Correction to the prior file: it stated memory was "15–25% of KLA revenue." The FY2026 disclosure puts memory at 38% of semiconductor process control systems revenue. Even after allowing for services and the non-PC segments diluting that toward the company total, memory exposure is materially larger than the prior file assumed. That cuts both ways — it is more exposure to the guided pricing headwind, and more exposure to the HBM complexity ramp that management calls out as a demand driver. It is the reason this report does not raise conviction despite the estimate revisions.

3.4 Disruption forecast, 5-10 years

Vector	Likelihood	Direction	Note
Node scaling stalls → PC intensity plateaus	Low	↓	GAA, backside power delivery, high-NA and advanced packaging all raise inspection intensity through 2030
Advanced packaging becomes a new PC market	High	↑	Explicitly named by management as a 2027 growth vector; KLA is already there
Chinese domestic process control matures	Medium-high	↓↓	The real long-term threat. Domestic toolmakers are years behind at the leading edge but not at legacy nodes, and this is where 30% of revenue sits
Export controls tighten again	Medium	↓	Already cost KLA share of China revenue: 41% (FY2024) → ~30% (FY2026)
AI-driven inline metrology displaces optical inspection	Low	↔	KLA is the party with the training data; more likely a sustaining innovation for KLA than a disruptive one

3.5 Evergreen assessment

Evergreen as a franchise; cyclical as a stock. As long as anybody makes leading-edge chips, somebody pays KLA's toll, and the toll rate rises with complexity. But ~93% of revenue is tied to a capex cycle that KLA does not control, and 30% of it sits in a jurisdiction the U.S. government adjusts by decree. Rate: 8/10 evergreen, 5/10 predictable.

4. Valuation (§3)

4.1 ⚠️ Data trap — the vendor forward multiple is off by a fiscal year

Yahoo reports forward P/E 27.60 at $185.04, implying forward EPS of $6.71. Reconstructing from the company's own guide: FQ1 FY2027 non-GAAP EPS is $1.16; even at the FY2026 rate of sequential growth, FY2027 lands near $5.15–5.45. Zacks consensus is $5.43 (explicitly "+44.4% over fiscal 2026's $3.76"). $6.71 is the FY2028 number.

Basis	EPS	Multiple at $185.04
Yahoo "forward"	$6.71	27.6x ← wrong year, makes the stock look a rating step cheaper
FY2027E (true forward)	$5.43	34.1x
FY2026 non-GAAP (trailing)	$3.76	49.2x
FY2026 GAAP (trailing)	$3.66	50.6x

This is the third-plus repetition of Knowledge/Playbook/pitfall-vendor-forward-eps-is-the-wrong-fiscal-year, and KLAC is not yet on that note's ticker list. Every multiple below uses $5.43 (FY2027E).

4.2 Model suite

Model	Applied?	Result
Forward multiple (primary)	✅	Bear 28x × $5.15 = $144 · Base 33x × $5.43 = $179 · Bull 38x × $5.70 = $217
Own historical band	✅	3-yr average forward P/E 33.1x → $180. Sector median 29.6x → $161
FCF yield	✅	FY2026 FCF/share $2.86 → 1.54% yield — rich. FY2027E FCF/share ~$3.50 (assumes conversion recovers to ~85%) → ~1.9%. Still expensive on cash
Bogle expected return	✅	0.50% yield + ~15% normalised earnings growth − ~2.5%/yr multiple compression (34x → 30x over 5yr) = ~13%/yr
Graham IV	⚠️ Not meaningful	√(22.5 × 3.66 × 4.86) = $20.00. BVPS is $4.86 because buybacks destroyed book equity. Reporting this as a fair value would be false precision — it is discarded, not weighted down
DDM / DYT	❌ N/A	0.50% yield, 22% payout. Not a dividend name
Analyst consensus	—	29-44 analysts, mean target $233.77 (≈43x FY27E), range $150–325, consensus Buy

4.3 Fair value and the zones

Fair value: $165 – $205 (base ~$185, i.e. 30–37x FY2027E $5.43).

Spot $185.04 is sitting precisely on the base case. On the §0 first-principles grid this is "Great company, fair price" — which is neither the value quadrant nor the avoid quadrant. It is a HOLD.

Zone	Level	Reasoning
Entry (numeric)	$155 – $178	≤32.8x FY27E. The stock printed $167 on 9/16 and $170 on 7/29, so this band is live, not hypothetical
Strong add	< $160	≤29.5x — below the sector median, for a 58%-share asset with 41.6% ROIC
Trim	42x forward (FY2027E basis)	≈ $228 at today's $5.43 estimate; rises automatically as the estimate rises. ~43x is where the analyst mean target sits and where a peak-cycle equipment name has priced in the whole 2027 ramp
Hard cap (non-price)	Combined AVGO+KLAC > 8.5% of account	A cyclical bought at a cycle high deserves a size trigger as well as a price trigger — see §6

Trim caveat (cyclical): a multiple-based trim on a cyclical can invert at the peak, because the multiple compresses as peak earnings arrive. If FY2028 consensus keeps climbing toward $6.70+, re-set the trim multiple downward at the next pass rather than letting a rising EPS carry the trim dollar to a level that never gets hit.

5. Sentiment & intelligence (§ Sentiment analyst)

Tone: constructive on the business, unsupportive on the stock.

A portfolio-specific passage was removed from the public build.

On the insider selling: at a company with heavy equity comp, programmatic 10b5-1 selling is normal and is not on its own a signal. What raises an eyebrow is the CEO selling 72,019 shares at $171.88 on September 15 — within 2% of the local low, rather than pausing. It does not falsify anything. It is recorded because nobody in the C-suite has bought a share in the open market through a 40% drawdown.

6. ⚠️ The AVGO-vs-KLAC bucket conflict — the decision that matters

Both the KLAC and AVGO files have said to own ONE of these at size, not both. The account now holds both: KLAC 5.28% + AVGO 2.21% = 7.49% in one AI-capex bucket, with KLAC the larger. The KLAC build was funded by selling QQQ — an active de-indexing decision, not drift, and the largest capital deployment of the month.

6.1 They are not actually the same exposure

The "own one, not both" rule deserves re-examination, because these two names are less correlated than the shared "AI capex" label suggests:

	KLAC	AVGO
What it sells	Process control tools — a toll on wafers fabricated, any customer, any chip	Custom AI ASICs + networking — a toll on specific hyperscalers' silicon programs
Customer concentration	Diversified across every leading-edge fab	6 named/rumored relationships; Google, Meta, OpenAI, Anthropic carry the thesis
Revenue driver	WFE capex, +7.3% in 2027 (SEMI)	AI semi revenue: $58B FY26 → $115B FY27 → $230B FY28 target
Growth rate	Mid-teens	AI segment +221% YoY in Q3, guided +236%
Balance-sheet risk	None (0.74x net debt/EBITDA, no M&A)	$29B disclosed max credit exposure backstopping Anthropic's SPV lease obligations, no first lien; a further $60-70B facility reported, unsized
Cyclicality	High — capex cycle, memory pricing, 30% China	Lower near-term, but concentration risk substitutes for cycle risk
ROIC	41.6%, rising	~47% ex-goodwill (untested two passes)
FCF quality	Deteriorating — 78% of net income, FCF +0.5% in FY26	Excellent — $13.7B in Q3 alone, 46% FCF margin, a record

They share the "AI capex" factor but the failure modes are different: KLAC breaks on a capex digestion or a China action; AVGO breaks on a hyperscaler in-sourcing or a credit event at a leveraged AI customer. Holding both is defensible diversification within the factor — it is the combined size, not the pairing, that needs a cap.

6.2 But at today's prices, the answer is not close

	KLAC	AVGO
Spot	$185.04	$362.93
Fair value	$165–205	$425–500
Position vs. fair value	~0% — dead on the base case	−15% to −27% — below the range
Entry zone	$155–178 (above spot — not in zone)	$360–420 (in zone)
Verdict	HOLD [7.5]	ACCUMULATE [8.3]
Forward multiple	34.1x FY27E	25x-fwd trim basis; trading well inside it
Estimate direction	↑ (+7.7% / 30d)	↑ (mean target rose $525.97 → $533.41 while the price fell)

AVGO is the better vehicle at today's prices, and it is not a close call. Not because it is the better business — that is arguable both ways — but because of §0: it is not just how good the company is, it is how well the market has priced it. AVGO is a great business trading below its fair-value range with rising estimates. KLAC is a great business trading exactly at its fair-value midpoint. One of those is the value quadrant; the other is the "good company, fair entry" quadrant.

6.3 Recommendation — advisory only, execute in your own broker

Do NOT reverse the KLAC build. There is no thesis break, spot is inside fair value, and selling a position bought at ~$184 at $185 is a round trip that pays frictions to fix a weighting problem with a position decision. KLAC stays. It just stops growing.
KLAC is closed to new money above $178. Re-open it only in the $155–178 band, and only if the combined cap in (4) still has room.
Direct the next AI-capex dollar to AVGO, which is in its zone and ~20% below fair value, until AVGO is at least as large as KLAC. Target inside the bucket: roughly 4.5% AVGO / 3.5% KLAC.
Combined AVGO + KLAC cap: 8.5% of the account. Currently 7.49%, so there is ~1.0pt of headroom — enough to take AVGO from 2.21% to ~3.2% without touching KLAC. To get AVGO above that, fund it from the wider semi sleeve, not from KLAC — the prior file already named INTC (+204%, Alert-rated) and AMD (+334%, above trim) as the funding candidates, and that logic still holds.
Note the sleeve, not just the pair. AMD, INTC, QCOM, ANET and SNPS are also AI-semi factor exposure. The pair cap is a sub-limit inside a wider semi sleeve that should stay under ~20%.
On the QQQ funding: swapping index core for a 34x-forward cyclical at 5.28% is a real step up in idiosyncratic risk. It is not wrong — KLAC is a higher-quality asset than the average QQQ constituent — but it should be the last such swap until the pair is rebalanced toward AVGO.

7. Synthesis — where the agents disagree

Fundamentals vs. Moat. Fundamentals flags the FCF-conversion decay (110% → 92% → 78%) and the buyback that reduced the share count only 1.34% for a record $2.5B. Moat counters that ROIC rose to 41.6% on 16% more invested capital, gross margin expanded for a third straight year, and incremental operating margin (45.4%) sits above the average — none of which is what deteriorating economics looks like.

Resolution: Moat wins on the structural question, Fundamentals wins on the timing question. The economics are getting better; the cash timing is getting worse because of a receivables and inventory build ahead of a shipment ramp. These are compatible. Fundamentals' point is the one that belongs in the trigger list, and it is there: FCF conversion must stop falling.

Valuation vs. Sentiment. Sentiment points to estimates revised up 7.7% in 30 days, a 2027 WFE forecast of +7.3%, and a $233.77 mean target. Valuation points out that at $185.04 the stock already trades at 34.1x those revised estimates and 1.54% FCF yield, and that the vendor 27.6x that makes it look reasonable is built on the wrong fiscal year.

Resolution — weighted for company type. This is a cyclical with a genuine moat, so per §5 the weighting is Fundamentals + Valuation over Sentiment momentum. Sentiment's evidence is real and it is why this is HOLD rather than TRIM. Valuation's evidence is why it is not ACCUMULATE.

A portfolio-specific passage was removed from the public build.

8. Survival ledger — the 2026-07-28 pre-print file

Prior claim	Status	Evidence
Moat intact; process control is a toll booth no rival can attack	✅ SURVIVED, strengthened	58% share, ~6x nearest rival, GM 60.9% → 61.3% (FQ4 62.4%), ROIC 39.4% → 41.6%
"The multiple isn't the fragile part — the estimate is"	✅ SURVIVED in mechanism, wrong in direction	The estimate moved — up 7.7%, to $5.43. The multiple took the hit instead
"Do not buy today. Let the print land."	✅ SURVIVED — and was correct	Stock fell 10.8% the next session, to $170.00
Operating margin pinned at 41–42%	✅ SURVIVED	FY2026 41.7%
Process control to grow >20% in 2026, outpacing WFE	✅ SURVIVED (directionally)	FQ1 FY27 guide +24.6% YoY vs WFE +9.0%
Management would face cautious FY27 commentary	✅ SURVIVED	Memory pricing headwind guided to persist through 2027
"Memory is 15–25% of KLA revenue"	❌ SUPERSEDED	38% of semiconductor process control systems revenue
"Share count −4.1%/yr"	❌ SUPERSEDED	FY2026 diluted count fell 1.34%; 6-yr average −3.0%/yr
Fair value $155–185	❌ SUPERSEDED — too low	Built on $5.13 FY27E; consensus is now $5.43. New range $165–205
"Trim $260+" (a frozen dollar)	❌ REPLACED	Now 42x fwd FY27E ≈ $228, and it rises with the estimate
Accumulate zone written as prose (`<$190`, `$165–190`)	❌ REPLACED	Now a numeric `entry: 155-178` field, machine-readable by the scanners
Korea 15–20% of geography	⏳ UNTESTED	Not re-verified this pass; China (30%) is the disclosure that is available and is the larger exposure
"You already own AI-semi capex via AMD, AVGO, INTC, QCOM, ANET, SNPS — treat KLAC as a quality upgrade, not a new incremental bet"	⚠️ SURVIVED and was not followed	KLAC went 2.2% → 5.28% funded from QQQ, i.e. as a new incremental bet. §6 addresses it

9. Verdict

🟠 HOLD — Conviction [7.5]

Field	Value
Fair value	$165 – $205 (base ~$185)
Entry	$155 – $178
Strong add	< $160
Trim	42x forward (FY2027E) ≈ $228 today
Recheck	2026-10-30 (day after the FQ1 FY2027 print)

KLA is one of the highest-quality assets in the semiconductor complex — 58% share of a market it has held for decades, 41.6% ROIC that is rising, a third consecutive year of gross-margin expansion, an asset-light 2.8%-of-revenue capex model, no M&A, and 94% of free cash flow returned to owners. The print the prior file was waiting on cleared: revenue +15%, an EPS beat, margins expanding, and FY2027 consensus revised up 7.7% to $5.43. The 10.8% drop the next day was about tone, not numbers.

But at $185.04 the stock trades at 34.1x those revised estimates — not the 27.6x the vendor field shows, which is built on FY2028 — and on a 1.54% free cash flow yield. That is the fair-value midpoint, not a discount. Layer on three things the prior file did not have: memory is 38% of process control systems revenue rather than 15–25%, FCF conversion has fallen 110% → 92% → 78% over three years, and the CEO sold 72,019 shares near the September low. Great company, fair price. Hold it, don't feed it.

The position is also the wrong size relative to its sibling. At 5.28% KLAC is the third-largest holding while AVGO — ACCUMULATE [8.3], in its entry zone, ~20% below fair value, with rising targets — sits at 2.21%. The account owns more of the fully-valued one. Don't fix that by selling KLAC at fair value; fix it by sending the next dollar to AVGO and capping the pair at 8.5%.

Thesis break triggers

FY2027 consensus non-GAAP EPS falls below $5.00 (estimate reversal — the thing the prior file correctly identified as the fragile part).
FCF / net income below 75% for a second straight period, or FY2027 FCF failing to grow.
Process control revenue growth <10% YoY for two consecutive quarters.
A memory customer publicly cutting 2027 capex, or memory pricing pressure widening beyond pricing into unit volumes.
A new China export-control action reaching process control tools (30% of revenue).
Combined AVGO + KLAC above 8.5% of the account — a size break, independent of price.

10. Data gaps and flags — stated, not papered over

roic.ai returned only FY2025–FY2026 for this account, not the 20-year history the framework wants. The 5–8yr CAGRs here were rebuilt from KLA's own FY2021 and FY2026 8-K/press releases plus stockanalysis.com. FY2022 diluted share count (~1.43B post-split) is interpolated, not sourced — it affects only the FY2022 FCF/share cell, not any CAGR endpoint.
Yahoo's forward P/E (27.6x) is built on the wrong fiscal year. Every multiple in §4 uses the reconstructed FY2027E of $5.43. KLAC should be added to Knowledge/Playbook/pitfall-vendor-forward-eps-is-the-wrong-fiscal-year.
Memory at 38% is of semiconductor process control systems revenue, not of total company revenue. The total-company figure was not separately disclosed in the sources reached. The direction of the correction versus the prior file's 15–25% is certain; the exact company-wide percentage is not.
Korea geographic exposure (15–20% in the prior file) was not re-verified. Carried as UNTESTED.
The FQ4 shareholder letter PDF would not parse through the fetch tool; segment and geographic detail came from the press release, the Q3 FY2026 slide coverage and the FY2026 10-K summary instead.
Per scope lock, this pass wrote no other file, ran no kb.py index, and made no watchlist, knowledge-base or portfolio edit. Three knowledge-note candidates are surfaced for the orchestrator rather than written here: (a) add KLAC to the wrong-fiscal-year pitfall; (b) a new pattern on FCF-conversion decay running ahead of an earnings-quality problem; (c) a new pitfall on pre-event reports written without frontmatter being invisible to both kb.py stale and /thesis-drift — this file sat unseen for 56 days.

Sources: KLA FQ4 FY2026 press release (PRNewswire, 2026-07-28) · KLA FY2021 8-K (SEC) · KLA FY2026 10-K · stockanalysis.com KLAC financials · roic.ai profitability ratios · Yahoo Finance via .mcp/fin.py and MCP · Zacks / Simply Wall St consensus estimates · SEMI equipment forecast (Dec 2025) · Investing.com Q4 FY2026 call coverage · Knowledge/Themes/semi-equipment-and-test.md · Output/Stocks/Semiconductors/AVGO/analyze-from-before-2026-09-10.md