At $69.40 (P/E 12.26 vs 10y avg 18.38, EV/FCF 8.64), SWKS trades at 0.21x base IV. The catch: reverse-DCF implies -6.83% growth in perpetuity, and ~60% of revenue runs through one customer (Apple) that is publicly designing them out.

Plain English

Skyworks makes the radio chips inside iPhones, Android phones, Wi-Fi routers, and some cars. Most of its money comes from one customer — Apple. Apple is now designing its own radio chips, the way it designed its own laptop chips. If Apple replaces Skyworks even partly, the company's profits drop sharply. The stock looks cheap, but cheap because the market thinks Apple is leaving. We do not know when, or how much. Buying this requires guessing one customer's roadmap. Munger's rule says skip businesses where the answer depends on guessing what one giant customer decides to build in-house. Skip.

Thesis

Skyworks Solutions designs analog and RF semiconductors that go into smartphones (the bulk), Wi-Fi/Bluetooth modules, automotive infotainment, and industrial connectivity. The business compounds when phones, base stations, and connected devices add more bands and more complexity per unit; Skyworks owns the IP, the GaAs/SOI process know-how, and the customer-qualified modules that are very hard to second-source mid-cycle.

The scorecard says the franchise was, historically, excellent: 10-year average ROIC of 24.39%, FCF conversion 101.15%, net debt/EBITDA -0.39x (net cash), interest coverage 17.12x, and a modest 2.15% net share-count reduction over a decade. Owner earnings TTM run $1.32B. None of those are bad-business numbers.

The scorecard also says something is broken: 5-year ROIIC has collapsed to 5.16%, and the reverse-DCF embedded in $69.40 implies perpetual revenue contraction of -6.83%. Translation: Mr. Market is pricing terminal decline, not a cyclical trough. The base-case IV is $335 (per the deterministic scorer, with the 28% CAGR clamped to 14% for sanity), and the low IV is $194 — meaning even the punitive scenario has the stock at ~2.8x today's price.

This is a classic Buffett setup distorted by genuine concentration risk. If Apple's modem/RF in-housing plays out as feared, the franchise is worth far less than the scorer's IV. If Apple keeps Skyworks as second-source plus broad-markets reaccelerates, the rerate is enormous. Price/IV of 0.2071 demands an answer to whether the core business survives, not whether it grows. Margin of safety exists only if you can underwrite that survival. Below ~$60 the math becomes truly Buffett-friendly; above $130 the bull case is mostly priced.

Moat

Skyworks has historically scored as a narrow-moat business; the question is whether what's left has been hollowed out.

Pricing power. Skyworks does not have Coca-Cola-style brand pricing power. RF front-end content is sold as multi-die modules into a small set of OEM customers (notably Apple, Samsung, Google, plus base-station and IoT customers). Pricing is set in annual platform negotiations, and the OEM dictates the curve. ROIC of 24.39% over 10 years [scorer] is evidence of some pricing power (or at least cost insulation), but the 5-year ROIIC of 5.16% [scorer] suggests the marginal dollar of capex no longer earns the historical spread. Damodaran is blunt about this kind of asymmetry: "the return on equity and capital is not the cause of their success, but the consequence of it" [2] — meaning the high ROIC is a result of past moat, not proof of future moat.

Switching costs. This is Skyworks' real edge. RF modules are designed in 18-24 months ahead of a phone launch, qualified against tight RF performance specs, and tuned to a specific PCB and antenna geometry. Once a part is socketed for a flagship platform, swapping it mid-cycle requires re-tape-out, re-qual, and re-certification. Damodaran's Microsoft analogy applies — the moat is "the cost to the end-user of switching from one product to a competitor" [1]. The catch is that switching costs reset every 2-3 platform generations, and the OEM is the user — not the consumer. When Apple decides its in-house modem and RF roadmap is good enough, there is no consumer lock-in protecting Skyworks.

Network effects. None. RF chips don't get more valuable as more users adopt them.

Intangibles (IP / know-how / regulatory). Real but bounded. Skyworks holds thousands of patents in GaAs PA design, BAW/SAW filters, and RF system integration. Process IP in compound semiconductors is not casually replicable; Qorvo, Broadcom (ex-Avago), and Murata are the only credible peers with comparable depth. But "competitive advantage that comes from exclusive licensing or a legal monopoly is a mixed blessing" [1] — patents in semis age faster than the cycle, and the customer can route around any single patent family by going to the duopoly competitor or building in-house.

Cost advantages. Skyworks owns its U.S. fab (Newbury Park) and operates a hybrid fab/foundry model in BAW filters. This was an advantage when GaAs and BAW capacity was tight; it is an open question now that the cycle has reversed and depreciation must be carried whether or not units ship. Net debt/EBITDA of -0.39x [scorer] means the balance sheet doesn't penalize them for empty-fab quarters, but the unit economics still suffer.

Competitor stress test ($10B + 5 years). Could a $10B-funded competitor displace Skyworks in 5 years? In broad markets — yes, partially; Chinese RF FE vendors (Maxscend, Vanchip) are climbing the stack with policy support. In Apple flagships — no $10B startup gets there, but Apple itself is the $10B+ competitor, and the threat is well-publicized. That's the moat hole.

Erosion risk. High and ongoing. The 5y ROIIC drop from a 24%-class historical level to 5.16% [scorer] is the moat erosion in real time. Buffett's Precision Castparts comment — "management focused on translating a healthier industry backdrop into margins that better reflect its long-term potential" [3] — captures the bull's case (cyclical) but does not refute the bear's case (structural).

Moat verdict: NARROW.

Management

Philip Brace took over as CEO in February 2025 — a year and change in seat at the time of this filing. He came from Inseego (executive chair) and before that Sierra Wireless (CEO 2021-2023), both small-cap IoT names. This is a turnaround/operator profile, not a founder/owner profile, and not a Buffett-style multi-decade compounder profile. New CEO + concentration crisis is a meaningful overhang on "trust the capital allocator" reasoning.

Let's grade against the five capital-allocation choices.

1. Reinvest in the business. ROIIC of 5.16% over 5 years [scorer] is a red flag. When ROIIC drops below cost of capital, additional reinvestment destroys value. Skyworks has continued to invest in BAW capacity and advanced filter R&D anyway, betting on a content-per-phone recovery. If ROIIC stays single-digit, this is value-destructive reinvestment dressed as capex discipline.

2. Acquire. Skyworks made the Silicon Labs infrastructure & automotive deal in 2021 (~$2.75B), explicitly to diversify away from mobile. Strategically correct; the diversification has not yet shown up in revenue mix or margins to a degree that has rerated the stock. Mixed grade: the right idea, lukewarm execution.

3. Debt. Conservative. Net cash position; net debt/EBITDA of -0.39x [scorer]; interest coverage 17.12x. No criticism here.

4. Buybacks (Buffett's question: P/IV at the time of repurchase?). Skyworks has reduced share count by 2.15% over 10 years [scorer] — modest. Most of the cash returned has been via dividend, not buyback. Repurchase has been mostly opportunistic during dips. Given current Px/IV ratio of 0.2071 [scorer], management should be buying back aggressively right now if they believe their own forward plan. The pace of repurchase will be a critical tell over the next 4 quarters. Slow buybacks at this price = management doesn't believe IV either.

5. Dividends. Skyworks pays a meaningful dividend (~4% yield at $69.40). For a cyclical RF semi, the dividend-first posture is unusual and signals limited high-ROIC reinvestment opportunities — which is honest, but also confirms the bear case that growth is hard.

Communication quality. The 10-K/A is by-the-book. Disclosure on customer concentration is present but minimal (the geographic mix in the 10-Q shows U.S. = $805.7M of $1,035.4M, ~78% [filing], a proxy for Apple exposure). Management has not aggressively quantified Apple risk, nor has it committed to a specific diversification milestone with timelines. Communication is competent but not Buffett-clear.

Insider ownership. Modest. No founder-class skin in the game.

Putting it together. A new outsider CEO inheriting a structural concentration problem, conservative balance sheet, mixed M&A track record, value-destructive marginal reinvestment by ROIIC math, and dividends-over-buybacks at a price that screams buy back. The good news is the franchise generates $1.32B owner earnings TTM [scorer] and converts 101.15% of earnings to FCF [scorer] — the cash machine still works.

Capital allocator: B-/C+. Round to C+. Solid stewardship, no obvious malpractice, but not the kind of allocator who turns a concentration crisis into a Berkshire-style rerate. Capital allocator: C.

Industry

Industry: RF front-end semiconductors for mobile, IoT, automotive, and infrastructure.

1. Threat of new entrants — moderate-high (and rising). Compound-semiconductor process expertise (GaAs PA, BAW filter) is a real barrier; you can't show up with a Series A and ship a flagship-grade module. But there is now substantial state-backed Chinese investment in RF FE (Maxscend, Vanchip, OnMicro), targeting domestic Android first and global tier-2 OEMs second. Five years ago: low threat. Today: meaningful threat in everything below the iPhone-flagship tier.

2. Bargaining power of buyers — very high. This is the dominant force in this industry. The smartphone OEM customer base is concentrated: Apple, Samsung, Xiaomi, Oppo/Vivo, Google. Apple alone is widely reported to be ~60% of Skyworks' revenue. The U.S. revenue share of 77.8% in the most recent quarter ($805.7M of $1,035.4M [filing]) is the proxy. When one customer is the majority of revenue and is also a vertically-integrating buyer (Apple's modem and RF in-housing), buyer power is structurally higher than any moat short of a true patent monopoly.

3. Bargaining power of suppliers — moderate. Wafer suppliers (TSMC for SOI, internal fabs for GaAs, partners for SiGe) and rare materials (gallium, indium phosphide) matter, but Skyworks has a hybrid fab strategy that internalizes much of it. Suppliers are not the binding constraint.

4. Threat of substitutes — moderate and structural. The biggest substitute is integration: SOC vendors (Qualcomm, MediaTek, Apple) absorbing more RF function into the modem die or AiP packages. Every generation moves a little more silicon away from discrete RF modules. This is a slow erosion, not a cliff, but it never stops.

5. Competitive rivalry — high. Three credible peers: Qorvo (now Onsemi-owned, in transition), Broadcom (ex-Avago, won the Apple BAW filter socket war), Murata (Japanese, dominant in low-band filters). All three contest every Apple/Samsung platform. Pricing each cycle is a margin grind.

Where is the value pool, and where is it heading? The value pool in RF FE has been migrating from discrete components to integrated modules to fully-integrated modem+RF systems. Skyworks captured the integrated-module wave. The next wave (modem-RF integration, in-housed by Apple) routes around Skyworks. The pool is shrinking for the merchant RF supplier even as it grows for vertically-integrated platform owners.

Industry Verdict: Average. Excellent in the historical window when integrated modules were the architectural answer; Average-trending-Poor in a world where Apple in-houses and Chinese RF closes the gap on Android. The duopoly economics of BAW filters (Skyworks/Broadcom) provide a structural floor, which is why "Average" rather than "Poor."

Inversion

I am short Skyworks at $69.40. Here is why.

1. The single event that kills this. Apple announces, in a non-flagship year (2027 or 2028), that its in-house cellular RF front-end is shipping in the lower-tier iPhones, with the high-tier following the year after. This is not speculation — Apple has been openly hiring RF engineers, has shipped the C1 modem, and the strategic logic is identical to the modem path. When the announcement lands, Skyworks loses ~$3B of annual revenue over 24 months (call it ~60% of total). At that point, owner earnings of $1.32B [scorer] become $300-500M, gross margins compress 600-900 bps because the lost revenue was high-mix flagship parts, and the broad-markets/automotive segments cannot remotely fill the hole. The company is fine — it has $222M of marketable securities [filing] and net cash — but it is a different and much smaller company.

2. Why the moat is narrower than bulls think. Bulls say switching costs protect the Apple socket. They do — for the current platform. Switching costs do not protect the next platform, because Apple controls platform design and can choose to qualify in-house silicon for the next bill of materials. The moat is reset every 2-3 years by the customer's product cycle, and the customer is the one designing it out. Bulls also say BAW filter duopoly with Broadcom is a structural protection. Broadcom won the iPhone BAW socket years ago; Skyworks is the second-source in BAW. Being the second-source in a duopoly when the buyer is verticalizing is not a moat, it is a runway.

3. Why management is worse than it appears. Brace was hired in February 2025 with a turnaround mandate; that's not bad in itself, but it confirms the board acknowledged the prior trajectory was unacceptable. The capital allocation tell: Px/IV is 0.2071 [scorer], meaning the stock trades at ~21% of base intrinsic value per the deterministic scorer. A confident management team buys back stock here, hand over fist. They are paying a dividend instead. Either (a) management does not believe the IV (rational), or (b) management is preserving cash to buy diversification (signals fear about the core). Both interpretations are bearish. Also: the 10-K/A discloses minimal customer concentration detail. Buffett-grade communication would be a slide saying "Apple is X% of revenue, here is our 5-year plan to get to Y%." That slide does not exist.

4. What bulls are extrapolating that won't hold. Bulls extrapolate the 10-year average ROIC of 24.39% [scorer] forward. That ROIC was earned in the unique 4G-to-5G content-per-phone explosion, with Apple as a willing buyer paying for complexity it didn't yet want to in-house. Both conditions are reversing: 5G content per phone is plateauing, and Apple is now willing and able to in-house. Bulls also extrapolate the FCF conversion of 101.15% [scorer] as if it implies durable cash generation. FCF conversion is high because the company is in the harvesting phase — capex is being throttled as ROIIC has collapsed to 5.16% [scorer]. High FCF conversion in a harvesting business is not a virtue; it is a signal of run-off.

5. Valuation trap (multiple compression / regime change). The 12.26x P/E vs 10y average 18.38x [scorer] looks cheap. It is not cheap — it is correctly priced for a different regime. Pure-play Apple suppliers with verticalization risk historically retrade to 8-10x P/E, not their 5G-era average. Cirrus Logic, Lumentum, and Qorvo all have similar risk and trade at similar or lower multiples. The reverse-DCF implied growth of -6.83% [scorer] is the market saying: we expect revenue to shrink ~7% per year forever. That's not crazy. If the in-housing decision is announced, revenue will shrink ~30% in two years. Then it stabilizes 40% lower. The stock at that point — call it $35-45 — is the floor.

Multiplied probability of bear case happening: ~50-60% within 5 years. The market has assigned it ~70% (per the IV ratio). The bear is mostly priced — but "mostly priced" still leaves room for downside if the regime change is faster or sharper than embedded.

If I am right, the stock could be worth $40 within 3 years.

Lollapalooza Bias Check

Biases I am running right now, ranked by intensity.

Authority bias (high). The deterministic scorer says base IV = $335 and Px/IV = 0.2071. That number is loud, and it is tempting to defer to it because it has the appearance of objective math. But the IV is the output of growth and margin assumptions; the scorer itself flagged it: "base CAGR clamped from 28.0% to 14.0%" and "Maintenance capex uncertain (>50% spread); widen IV range" [scorer notes]. Authority bias toward the number can mask the conditional nature of the inputs. I am writing this section largely because the IV is so seductive.

Anchoring (high). I am anchored on the 10-year average P/E of 18.38x and the 10-year average ROIC of 24.39% [scorer]. These are real numbers and they describe a real franchise, but they describe the historical franchise during a unique 4G/5G content-per-phone window. Anchoring on these two figures pushes me toward "cheap, mean-reverts up." The bear case requires me to abandon that anchor and ask whether the regime that produced it has ended.

Confirmation bias (medium-high). Once I noticed the 5y ROIIC of 5.16% [scorer] and the negative reverse-DCF growth, every subsequent piece of information started fitting a "value trap" narrative. A balanced analyst would also be looking aggressively for the bull-side disconfirmer (e.g., concrete broad-markets revenue traction, BAW share gains, Apple multi-year supply contract leak). I have not found one in the filings excerpted, but I have not searched as hard as I would for confirming evidence.

Recency bias (medium). The current quarterly ($1,035.4M revenue, modestly down YoY [filing]) feels like the current trend. One quarter is a noisy data point in a cyclical semi business. I am giving it more weight than the deterministic 10-year metrics suggest I should.

Deprival super-reaction (low-medium). I notice an emotional pull toward "the stock was $200 a few years ago, recovery is owed." This is a classic ex-shareholder bias, even though I don't own the stock. Whenever I notice myself reasoning from "it used to be worth X" rather than from "the cash flows justify Y," I should discount that reasoning.

Social proof (low). The Buffett-Munger framing of this exercise itself biases me toward "this looks cheap on Buffett metrics, therefore it is a Buffett buy." The Munger move is to fail the circle-of-competence test on businesses that require predicting OEM design wins and tech adoption curves — and Skyworks plausibly fails that test. Social proof toward the framework should not override the framework's own gates.

Net effect. Authority + anchoring + confirmation are all pulling toward "too cheap to ignore." The Munger discipline says to make the bear case as strong as I can and let the gap between bull and bear price the conviction, not the recommendation. Conviction here is medium at best, not high — the bias check is the reason.

10-Year Outlook

Same fundamental business model in 10 years? Probably not. Skyworks today is a merchant RF front-end supplier whose largest revenue line is iPhone. In 10 years, either (a) Apple has fully in-housed RF FE for iPhone and Skyworks is a smaller, automotive- and IoT-weighted analog/RF company, or (b) Skyworks has successfully repositioned into broad markets and edge connectivity and resembles a smaller Analog Devices. Path (a) is more likely than path (b). Either way the business in 2036 is not the business of 2026.

Customer base larger? Probably broader, almost certainly less concentrated, but not necessarily larger in dollars. Diversification away from Apple is a forced march, not a choice.

Profit per customer higher? Unlikely. Broad-markets and automotive RF customers pay lower ASPs and lower gross margins than mobile flagship customers. The 10-year average ROIC of 24.39% [scorer] is unlikely to be repeated; a more realistic forward ROIC for a successful broad-markets pivot is 12-15%.

Moat wider? No. The moat narrows in either path. In path (a), the flagship socket is gone. In path (b), the company plays in industries where Texas Instruments, ADI, NXP, Infineon, and Microchip already have decades of customer relationships and broader product portfolios.

Single biggest threat? Apple in-housing the RF front-end on the same trajectory it in-housed the modem (C1 already shipped). Secondary: Chinese RF FE vendors closing the gap in Android, with state subsidy.

Confidence level. The 10-year shape of the business is genuinely uncertain. Two paths, one bad and one merely OK, neither a high-quality compounder. The Munger filter bites here: "Same fundamental shape 10 years ago, 10 years forward?" The honest answer is no. That is a yellow flag bordering on red. The price/IV ratio of 0.2071 [scorer] provides margin of safety on paper, but the IV itself is sensitive to the shape question.

CONFIDENCE: low

Position Guidance

• Recommendation: Too Hard
• Conviction: medium
• Target buy price: $55 (only if circle-of-competence concerns resolve via concrete Apple long-term agreement disclosure or broad-markets revenue acceleration above 15% YoY)
• Target trim price: $130 (above this, even a partial Apple-retention scenario is mostly priced in)
• Position sizing: 0% baseline. If the circle-of-competence concern is resolved (multi-year Apple supply visibility OR diversification past 50% Apple), revisit at 1-2% starter position with adds on weakness toward $50.
• Why Too Hard despite cheap headline metrics: The thesis depends on predicting Apple's vertical-integration roadmap and the pace of Chinese RF FE share gain. Both are tech-adoption curve questions Munger explicitly excludes from the circle of competence.