CRUS › analyze
CRUS · Analyze
0. Knowledge check
No live note or prior report exists for CRUS (kb.py find CRUS returned nothing; Knowledge/INDEX.md
carries no CRUS-specific entry). This is a first-look report, built fresh. Two pitfalls were checked
before trusting any vendor number, per protocol:
- [[pitfall-vendor-forward-eps-is-the-wrong-fiscal-year]] — fired, but mildly. Yahoo's headline
forwardPE(12.19x onforwardEps$8.90) prices FY2028, not the current fiscal year. The correct current-year (FY2027, ending Mar-2027) read ispriceEpsCurrentYear12.87x onepsCurrentYear$8.4312 — only ~6% apart from the headline, not the 30-80% gaps this pitfall usually produces, but the correct number is still the 12.87x one and it is used throughout below. One useful secondary finding this cross-check surfaces: FY2027 EPS growth guided at only +3.1% y/y (from TTM $8.18 to $8.43) before reaccelerating to +5.6% in FY2028 ($8.90) — i.e. consensus itself is calling this year the air-pocket, then a partial recovery, which is the central question this report has to answer (§5). - ADR pitfalls — n/a, CRUS is a US domestic issuer, not an ADR.
- roic.ai MCP would not authenticate this session (
session not foundon every call) — cross-validation fell back to Yahoo's raw income-statement export (get_financial_statement) instead of pre-computed ratios. That export is granular enough to reconcile every number used below, so the gap is noted but did not block the analysis.
One data-quality catch worth flagging up front because it explains a real anomaly in the numbers: TTM net margin (21.0%) sits above TTM operating margin (18.4%) — the tripwire in [[pattern-net-margin-above-operating-margin-is-a-tripwire]]. Traced to source (FY26 income statement): net interest income of $36.8M on the growing cash pile (legitimate, recurring, grows with the cash balance) plus an effective tax rate that fell from 25.5% (FY25) to 16.6% (FY26), worth roughly +$42M of net income versus a normalized ~25% rate. Net effect: FY26's reported 32.9% 3yr net-income CAGR is flattered by a falling tax rate and a growing net-interest line, neither of which is business performance. A tax-normalized FY26 net income is closer to $370M (EPS ~$7.0) than the reported $414M ($7.85) — a ~10% haircut worth carrying into every multiple below.
1. Fundamentals
Business: Cirrus Logic is a fabless mixed-signal/audio chip designer. Two reporting segments: Audio (amplifiers, codecs, smart codecs, SoundClear DSP) and High-Performance Mixed-Signal / HPMS (camera controllers, haptics, battery/power ICs). Revenue is overwhelmingly one customer: Apple was ~91-92% of net sales in FY2026 (10-K, confirmed across Q4 FY26 and FY26 full-year disclosures). 1,651 employees, no debt of consequence, no dividend.
| Metric | FY2023 | FY2024 | FY2025 | FY2026 | 3yr CAGR |
|---|---|---|---|---|---|
| Revenue | $1.90B | $1.79B | $1.90B | $2.00B | +1.7% |
| Gross profit / margin | $957M / 50.4% | $916M / 51.2% | $996M / 52.5% | $1.05B / 52.7% | margin +2.3pp |
| Op income / margin | $345M / 18.2% | $345M / 19.3% | $410M / 21.6% | $460M / 23.0% | margin +4.8pp |
| Net income | $177M | $275M | $332M | $414M | +32.9% |
| Diluted EPS | $3.09 | $4.90 | $6.00 | $7.85 | +36.5% |
| OCF | $340M | $422M | $444M | $651M | +24.2% |
| Capex | -$37M | -$38M | -$29M | -$15M | (falling) |
| FCF | $303M | $383M | $416M | $636M | +28.0% |
| Buybacks | -$209M | -$205M | -$299M | -$318M | — |
| Diluted shares | 57.23M | 56.02M | 55.24M | 52.82M | -2.6% |
| Cash | $446M | $503M | $540M | $801M | — |
| Total debt | $141M | $155M | $144M | $134M | (~all lease/other, no bonds) |
| Debt/Assets | — | — | — | 5.4% | — |
FCF is the strongest single fact in this file. FCF/share reached $12.04 in FY26 against a $108.48 price — an 11.1% FCF yield, and EV/FCF of 7.4x against net cash of roughly $660-760M (≈13-14% of market cap). Capex is trivially small (fabless model, ~1% of revenue), so OCF and FCF track almost 1:1.
Capital allocation is simple and disciplined: FCF funds buybacks (~50% of FCF each of the last two years) with the remainder building the cash balance — no dividend, no M&A since a small FY22 deal, no debt raised or retired at scale. Buybacks have cut diluted shares by ~8% over three years, a genuine per-share tailwind layered on top of revenue growth that is otherwise close to flat.
The growth-quality flag: revenue CAGR is +1.7%/3yr while net income grew +32.9%/3yr — a 19x gap. Decomposing it: ~2.3pp of gross-margin expansion, ~4.8pp of operating-margin expansion (R&D held flat to declining in dollar terms while revenue grew, i.e. real opex leverage on a stable headcount base — not a limitless lever per [[pattern-margin-expansion-is-a-finite-growth-lever]], but a real one with a visible floor around today's ~52-53% gross margin ceiling for a mixed-signal chip supplier), ~8% from buybacks, and the balance from the falling tax rate and growing interest income flagged above. Most of the reported EPS growth is real (margin discipline + buybacks), but a meaningful slice (tax rate + interest income) is not repeatable at the same rate — normalize expectations toward the ~20-24% underlying NI growth range, not the headline 33%.
Per-share view: revenue/share $37.81, FCF/share $12.04, both compounding through the buyback even while top-line revenue barely moves — this is the mechanism by which a low-single-digit-revenue-growth company can still be a legitimate per-share compounder, as long as the buyback continues and the price paid for it is reasonable (it is, at 7.4x EV/FCF).
No dividend — payout ratio 0%, no yield. The dividend overlay (§4 of the framework) is N/A; this name is evaluated purely on FCF/buyback compounding plus eventual multiple recovery.
2. Moat & Competitive Advantage — the concentration question, stress-tested
ROIC. Reported "InvCapital" (~$2.13B, effectively = equity given negligible debt) understates true operating returns because ~$800M of it is a non-operating cash pile. Op-income-based ROIC on total invested capital: 460M × (1-16.6%) / 2.13B ≈ 18%. Stripping the excess cash (~$660-760M, since the business needs little working capital) to isolate operating capital (~$1.4-1.5B) gives an operating ROIC closer to 25-27% — genuinely strong for a component supplier, and the gross-margin trend (50.4%→52.7% over 3 years) confirms it is improving, not eroding. On the standard test, this reads as an intact and slightly strengthening moat.
But the standard test is the wrong lens for a single-customer supplier — the concentration itself is the moat question. Adversarial stress-test: if I were Apple, how easily could I eliminate this supplier?
- Apple has done exactly this before: it in-sourced graphics (killing ~half of Imagination Technologies' revenue) and has methodically brought silicon in-house across the Bionic/M-series line. Cirrus's audio codecs and amplifiers are precisely the kind of mixed-signal analog IP Apple could target next, and the concentration means a full or partial in-sourcing decision would be an existential-magnitude event for CRUS, not a segment headwind.
- Counter-evidence for durability: Cirrus has held this position for well over a decade through multiple iPhone generations, multiple points where in-sourcing speculation has recurred, and has actually expanded content per device over time (camera controllers, haptics, battery/power ICs — the HPMS build-out) rather than losing ground. Mixed-signal analog design (ADCs/DACs, power management, haptic drivers) is a different, harder discipline than the digital logic Apple has in-sourced so far (graphics, CPU cores, modems it is still trying to finish) — high switching costs and multi-year qualification cycles apply on both sides. Cirrus was also named to Apple's American Manufacturing Program in 2026 and is co-developing process technology with Apple/GlobalFoundries for the Malta, NY fab — a sign of a deepening, not weakening, relationship, at least for now.
- The realistic risk is not full replacement but incremental in-sourcing or dual-sourcing at the margin — Apple splitting a socket with a second supplier, or bringing one product category in-house while leaving others with Cirrus. That is a slow bleed risk, not a single-quarter cliff, but it means the multiple this stock deserves should always carry a concentration discount versus a diversified peer at the same ROIC.
Revenue-stream map and the diversification nuance that matters most: Q1 FY27 segment split was Audio 54% / HPMS 46% of revenue, up from 41% HPMS a year ago — headline "diversification" language in every earnings deck. This is a segment shift, not a customer shift. Camera controllers, haptics, and battery/power ICs are overwhelmingly also Apple content (they ship inside the same iPhone/iPad); the 10-K's ~91% single-customer concentration figure is a whole-company number that the audio/HPMS mix shift barely touches. The only genuine customer-diversification lever in the current pipeline is the laptop/PC push (smart codecs for AI-PCs, projected to roughly double in FY26 off a small base) — real progress, but from a low starting share of total revenue, and itself now suffering the "delayed high-end PC product" timing headwind cited in the Q1 FY27 guide. Read every "diversification" claim in this company's materials as content diversification within Apple until proven otherwise by a customer-mix disclosure, not a segment-mix one.
Disruption forecast (5-10yr): the near-term disruption vector is customer concentration/in-sourcing, not a technology substitute — nothing in the audio/mixed-signal space (no new interface standard, no software-defined alternative) threatens the category itself. The company's own long-run answer to the concentration risk is the PC/automotive/industrial adjacency build-out; it is real but multi-year and unproven at scale.
Evergreen assessment: Cirrus is a well-run, high-ROIC, net-cash, buyback-compounding business inside a single-customer cage. It is not a "forever business" in the sense of a business independent of any one counterparty's goodwill — it is closer to a very well-defended tenant, and its evergreen rating should be capped accordingly regardless of how good the operating metrics look.
3. Valuation
No dividend, so DDM/DYT are not applicable (per framework, dividend-conditional models are skipped). Graham and FCF/EV-based approaches carry the weight.
| Model | Inputs | Output |
|---|---|---|
Graham IV √(22.5×EPS×BVPS) |
EPS $8.18 (TTM, reported), BVPS $43.40 | $89.37 — price $108.48 is ~21% above this. Using the tax-normalized EPS (~$7.0-7.4) instead, Graham IV falls to ~$80-85 — a materially more cautious floor. Graham is a poor natural fit for a zero-dividend, high-FCF, low-book-value-relative-to-earnings fabless chip designer (it structurally undervalues asset-light compounders), so this is weighted low, but it is the most conservative anchor in the set and worth respecting as a downside marker. |
| FCF / EV-based | FCF $636M, EV $4.68B → 7.4x EV/FCF; at a more normalized 10-12x EV/FCF for a net-cash, ~25% operating-ROIC semi supplier | Implied EV $6.0-7.6B + net cash ~$700-760M ≈ market cap $6.7-8.4B → ~$133-166/share |
| Current-year P/E | priceEpsCurrentYear-corrected EPS $8.43 (FY27E, current fiscal year) at a 14-18x multiple (below the stock's own 2024-2025 band, reflecting the concentration discount) |
$118-152 |
| Bogle expected return | No dividend; realistic earnings growth ~5-8%/yr once margin/buyback levers partially exhaust (see §1); possible P/E normalization from 12-13x toward a mid-teens multiple if the print confirms "timing not structural" | Wide range: roughly flat to +20%/yr over a 2-3yr hold depending almost entirely on whether the multiple re-rates; this is a "close call" band, not a point estimate |
Fair value range: $115-165. Current price $108.48 sits below the low end of every model above except the conservative, tax-normalized Graham floor (~$80-85) — i.e. the market is currently pricing this stock closer to the most bearish anchor in the set, not the central estimate. Consensus analyst targets agree directionally: mean target $165 (4 analysts), Stifel just cut to $160 from $197 post-print while maintaining Buy. Entry zone: $100-120 — the current price is already inside it, with room to add if the Q2 FY27 print (due ~early November) disappoints further on the PC/iPhone timing questions and pushes the stock toward the Graham floor.
Trim as a multiple: set at 20x FY27E current-year EPS ($8.43 → ~$169 today), which
recomputes as EPS climbs and sits modestly below the stock's own 2024-2025 trading band (it touched
$180 on much stronger growth optimism) — appropriately discounted for the concentration risk that
caps how much re-rating this name should get even in a good print.
4. Sentiment — why the price is where it is
CRUS traded near its 52-week high of $180.42 as recently as several months ago and now sits at $108.48, essentially at the 52-week low ($108.12) — a ~40% drawdown. The proximate cause is squarely dated: Q1 FY2027 results (reported Aug 5, 2026) beat on the quarter (revenue $459.7M, +13% y/y; non-GAAP EPS $1.84) but the stock fell >10% on the Q2 FY27 guide — $510-570M (midpoint $540M, +17.5% sequential) versus Stifel's prior $575M estimate. Management's stated causes: "tighter seasonality in fiscal 2027" (i.e., iPhone-cycle content/unit timing pushed later than the Street modeled) and supply-side pressure/delays in the high-end PC segment that slow the HPMS/PC diversification ramp specifically. Both are framed by management as timing, not structural share loss — no news of a design loss, a competitor win, or an in-sourcing announcement accompanied this guide-down; it reads as a genuine cyclical air-pocket around iPhone unit/content timing and PC delays, not a change in Cirrus's position at its customer.
Sell-side response corroborates the "cyclical, not structural" read: Stifel cut its target ($197→$160) but maintained a Buy rating, and the broader analyst set (mean target $165, high $200, low $130) still implies 20-85% upside from here — nobody in coverage has called this a broken thesis. Short interest is elevated but not extreme (8.6% of float, short ratio 4.8 days) — meaningful skepticism priced in, not a crowded short squeeze setup either direction.
Insider activity: dense but entirely routine — every transaction on file is a scheduled option-exercise-and-sell or an RSU vest, consistent with 10b5-1 plans (per [[pitfall-yahoo-insider-purchases-counts-rsu-grants]], these RSU/option-exercise rows are not a buy/sell signal and should not be read as one). No open-market insider buying at all in the trailing 12 months, including through the recent drop — a neutral-to-mildly-negative tell (insiders are not stepping up to buy the dip), though routine 10b5-1 selling after vests is not itself bearish.
Cyclical vs. structural — the direct answer: the evidence points to cyclical. Revenue and gross margin both continued to expand through FY26; the Q2 guide, while below the Street, is still +17.5% sequential and +13% y/y-consistent; the stated causes (iPhone content/unit timing, PC product delays) are dated, specific, and reversible on their own terms rather than open-ended. The structural risk that should keep conviction capped is the standing one — 90%+ single-customer concentration and Apple's demonstrated appetite for in-sourcing — which this quarter's guide did not change in either direction, but which never goes away and is the reason the multiple should stay below what an equivalent-quality diversified semi supplier would command.
5. Synthesis — weighted verdict
Is this a good business? Yes, on every quantitative test: ~25%+ operating ROIC on capital employed, expanding gross and operating margins, near-zero debt, a genuine net-cash balance sheet, disciplined 100%-of-FCF capital return via buybacks with no value-destroying M&A, and FCF compounding at a rate that has cut per-share cost of capital consistently. Has the market already priced that in? No — at 7.4x EV/FCF and a corrected 12.9x current-year P/E, sitting at the 52-week low after a guide-down that management, sell-side coverage, and the underlying data all characterize as a timing/cyclical air-pocket rather than a change in competitive position, this stock is priced closer to its most bearish valuation anchor than its central one.
Per the framework's four-quadrant test: Great business, cheap price → Value, with one qualifier that keeps this from being higher conviction: the business's "greatness" is inseparable from a single customer relationship that could deteriorate with no warning and no real hedge — the HPMS "diversification" story is mostly content diversification within Apple, not customer diversification, and only the small, early PC push is the real exception. This is not a value trap (nothing here is deteriorating — margins, ROIC, and cash generation are all improving), but it is also not a clean "great and cheap" case the way a diversified compounder at the same multiple would be — it is a cyclical air-pocket in a good but structurally concentrated business, priced to compensate for exactly that risk.
Weighting: Fundamentals + Valuation carry the most weight here (strong, unambiguous FCF/ROIC/ balance-sheet data, and a price below every fair-value anchor but the most conservative one); Moat analysis is the counterweight that caps conviction rather than a reason to avoid — the concentration risk is real, priced-in to a degree, and unchanged by this quarter's news, not a fresh red flag. Sentiment supports the "cyclical" read directly from the earnings-call language and analyst reaction. No named agent disagreement rises to a debate-round-worthy tension; the one real judgment call is how much of a discount the concentration risk deserves against an otherwise cheap, high-quality print — addressed by setting the trim at 20x (below the stock's own historical high-growth-era multiple) rather than assuming a full re-rating to the 2024-2025 band.
Verdict: ACCUMULATE. Conviction 6.5/10 — a genuine value case (good company priced for a cyclical air-pocket) tempered by a standing structural risk (90%+ single-customer concentration, real in-sourcing precedent at this same customer) that is not new information this quarter and does not change the "cyclical, not structural" read of the current print, but should permanently cap how much multiple this name is ever awarded relative to a diversified peer at the same ROIC. Size accordingly — this is a "good business, real single-point-of-failure" profile, not a core evergreen holding.
Key risks (named explicitly, not buried): 1. Customer concentration / in-sourcing — ~91% of revenue from Apple; Apple has in-sourced analogous silicon from suppliers before (Imagination Technologies precedent); no hedge exists against a partial or full in-sourcing decision, and it would not necessarily arrive with warning. 2. "Diversification" is mostly optical — HPMS growth is largely additional Apple content, not new customers; only the small PC/laptop push is genuine customer diversification, and it is itself currently delayed. 3. Guide-down could compound — if the Q2 FY27 print (~Nov 2026) shows further deterioration beyond "timing," particularly any signal of unit share loss rather than unit timing, the structural read would need to be revisited and the Graham floor (~$80-85 tax-normalized) becomes the relevant downside case rather than a hypothetical one. 4. No insider buying through the drawdown — a mild, not disqualifying, absence of a golden flag. 5. Earnings-quality footnote — recent net-income growth is flattered by a falling effective tax rate and rising interest income on the cash pile; both are legitimate but not fully repeatable at the same rate, so headline EPS-growth figures should be read at a discount to the reported number.
Data gaps / notes for the record
- roic.ai MCP session failed to authenticate this run; cross-validation used Yahoo's raw
get_financial_statementexport reconciled line-by-line instead of roic.ai's pre-computed ratios. No numbers here are unverified, but a future run should re-attempt roic.ai for an independent ROIC cross-check. - Only 4 fiscal years of clean statement data were available (FY22 balance/income data mostly null in the vendor feed), short of the framework's preferred 5-8yr CAGR window — the 3yr CAGRs above are what the data supports; treat them as directionally sound but not a full cycle.