Te Connectivity Plc TEL

Thesis

TE Connectivity (TEL) designs and manufactures roughly 500,000 SKUs of connectors, sensors, and electrical components that sit inside cars, factories, data-center racks, medical devices, and aircraft. The economic model is attractive in theory: a single connector might cost $0.50 but live inside a $40,000 vehicle, where the cost of a failure is catastrophic and the cost of switching suppliers mid-platform is even worse. That structural setup has supported a 10-year average ROIC of 11.7% and 5-year FCF conversion of 107% — solidly above the cost of capital and consistent with a real, if narrow, moat.

The problem is that the math from here looks miserable. Two scorecard numbers are the entire investment case in compressed form. First, 5-year ROIIC is 2.09%. That means every incremental dollar TEL has reinvested over the last half-decade — into auto-content growth, factory automation, data-center connectivity, sensor M&A — has earned roughly the rate on a Treasury bill. A compounder is not what generates 11.7% on its legacy capital; a compounder is what reinvests at a rate above its cost of capital. TEL is currently failing that test.

Second, the price-to-base-intrinsic-value ratio is 8.12x. Owner earnings TTM are about $322M against a market cap above $60B; EV/FCF is 26.8x; trailing P/E is 581 (distorted by a one-time charge but still telling). Base IV is $25.54. At $207.43, you are paying roughly $182 per share for growth that the last five years of capital deployment say isn't coming.

Margin of safety appears only near $25-33 (base-to-high IV). At any price north of ~$50 this is a Pass, regardless of how good the connectors are.

Moat

TE Connectivity sits in the gray middle of the moat spectrum: real durable advantages, but ones that have been narrowing rather than widening, and that have not been wide enough to convert reinvestment into excess returns lately.

1. Switching costs — the load-bearing moat. This is the strongest leg. Connectors are designed-in. An automaker spends 3-5 years qualifying a wire-harness architecture for a vehicle platform; once the connector geometry, materials, current rating, and vibration tolerance are locked into the engineering drawings and validated for crash, EMI, and thermal cycling, swapping suppliers mid-program is operationally and legally painful. Damodaran's framing applies almost verbatim: 'the most significant barrier to entry…is the cost to the end-user of switching from one product to a competitor' [1]. A $40,000 vehicle that fails because of a $0.50 connector is a recall event; OEMs trade a few cents per part for the certainty that the part won't be the failure. The moat is widest in automotive (~58% of revenue historically), industrial automation, aerospace, and medical, narrower in consumer/data-center where life cycles are shorter.

Stress test ($10B / 5 years): A well-funded entrant could match TEL's tooling, materials science, and global plants. What it could not buy in five years is the design-in position across thousands of active vehicle platforms and the qualification history. Aptiv (formerly Delphi), Amphenol, and Yazaki are the only credible peers, and the structure is an oligopoly precisely because the switching cost protects all three.

Erosion risk: EV architectures are forcing a connector redesign cycle (high-voltage, high-current, fewer parts per vehicle as wiring harnesses simplify around centralized compute). Every redesign cycle is an opportunity for an OEM to re-shop. The Chinese auto value chain (BYD, CATL ecosystem) is increasingly sourcing connectors domestically. Both pressures are visible in the 2.09% ROIIC.

2. Cost advantages — modest, scale-driven. TEL has ~85,000 employees and global manufacturing footprint giving unit-cost advantages over sub-scale competitors, plus vertical integration in stamping, plating, and molding. Damodaran observes cost advantages 'can run the gamut from brand name…to lower cost structures (in manufacturing) to superior technology' [3] — TEL's flavor is the manufacturing kind, which is real but the most easily replicated. Not a primary moat.

3. Intangibles — patents and engineering know-how. Thousands of active patents, decades of materials and contact-physics expertise. Damodaran is appropriately skeptical: value comes 'not necessarily [from] the companies that spend the most on R&D, but those who have the most productive R&D' [2]. TEL's R&D is productive enough to defend share, not productive enough to expand returns — exactly what 2.09% ROIIC tells us.

4. Pricing power — narrow. TEL is a price-taker on commodity connector lines, a price-setter on highly-engineered application-specific parts (medical, aerospace, high-speed data). The blended result is the ability to pass through copper and resin costs with a lag, not the ability to expand margins structurally.

5. Network effects — none. Connectors don't get more valuable as more people use them; this lever doesn't apply.

Competitor stress + erosion summary. The moat is real but oligopolistic, not monopolistic. Amphenol earns higher returns on incremental capital than TEL, which is itself the strongest evidence that TEL's specific moat is narrower than the category's. The 11.7% legacy ROIC reflects past design-in wins; the 2.1% incremental return reflects an industry where every new platform redesign is contestable.

Moat verdict: NARROW.

Management & Capital Allocation

Capital allocation at TEL has been competent in the housekeeping sense — the balance sheet is in shape (net debt/EBITDA 1.45x, FCF conversion 107%) — and uninspiring in the compounding sense. We grade across the five capital-allocation choices Buffett emphasizes: reinvest, acquire, debt, buybacks, dividends, plus communication.

Reinvestment. This is the dominant story and the dominant problem. 5-year ROIIC of 2.09% is not a rounding error; it is an indictment. Either (a) management is over-reinvesting into a structurally lower-return mix (EV connectors at lower margins than legacy ICE, data-center automation at competitive prices), (b) underlying unit economics have eroded faster than reported margins suggest, or (c) the denominator is inflated by acquisitions that haven't yet earned out. All three are partly true. A Buffett-quality allocator would have already tilted reinvestment dollars toward buybacks at the first sign of incremental returns falling below cost of capital. TEL has done some of this, not enough.

Acquisitions. TEL has been an active but middle-of-the-pack acquirer (Richards Manufacturing, Schaffner, various sensor and fiber tuck-ins). Buffett's standard from the 2011 letter is the 'bolt-on' done by operators like Iscar or TTI — small, in-circle, accretive, with the existing management staying [5]. TEL's deals fit the bolt-on shape but have not visibly moved consolidated returns higher; the ROIIC drag suggests deal multiples have been full and synergies modest. Not value-destructive, not value-creative.

Debt. Conservative. Net debt/EBITDA of 1.45x and investment-grade credit. A-tier on this dimension. Management has not used debt to juice buybacks at distressed prices, which would have been the high-IQ move in 2020 and 2022 dislocations. Missed offense, not defense failure.

Buybacks. Share count down ~3% over 10 years — tepid for a company with this much FCF. The right Buffett question is the average P/IV at which buybacks were executed. With base IV at $25.54 and the 10-year average P/E around 20.7x vs current 581x trailing (distorted) and a stock that has spent most of the last five years between $100-$200, buybacks have largely been done at multiples of intrinsic value, which destroys, not creates, per-share value. This is the single biggest capital-allocation criticism. A disciplined allocator would have stopped buying when the stock crossed 2x base IV; TEL has not.

Dividends. Steady and growing. A reasonable cash-return mechanism for a business whose reinvestment math has gone soft. Neutral.

Communication quality. Investor materials are clear, segment disclosure is good, organic-vs-FX-vs-acquisition revenue bridges are honest. CEO Terrence Curtin has been in the seat since 2017 with stable tenure. No accounting controversies, no aggressive non-GAAP framing beyond industry norms. The TTM P/E of 581 reflects a one-time charge that is disclosed straightforwardly in the filings; that's a positive tell.

Synthesis. Honest, technically competent, conservative on the balance sheet, undisciplined on price-paid-for-buybacks, and most importantly running a reinvestment engine that is currently producing 2 cents on the dollar of incremental capital. A Buffett operator would already have shifted to a special-dividend or large-tender posture. TEL has not.

Capital allocator: C.

Industry Structure

Porter's Five Forces on the engineered-connector and sensor industry:

1. Rivalry — Moderate-to-High. The category is a global oligopoly at the top (TE Connectivity, Amphenol, Aptiv, Molex/Koch, Yazaki, Sumitomo, Hirose, JST) with hundreds of regional and specialty players underneath. Top-tier rivals do not generally compete on price for design-in slots — they compete on engineering hours and qualification speed. Where they DO compete on price is at platform-renewal moments and in commodity SKUs. The presence of a vertically-integrated Chinese tier (Luxshare, BYD-internal) has intensified rivalry in the auto and consumer-electronics segments over the past five years, which shows up in TEL's flat-to-down incremental returns.

2. Buyer power — High and rising. Auto OEMs (35-40% of TEL revenue) are sophisticated, multi-source by policy, and have engineering staff capable of dual-qualifying. Tier-1 harness makers (Aptiv, Yazaki, Lear) sit between TEL and the OEM and apply their own pricing pressure. Industrial and data-center buyers are increasingly hyperscalers (Microsoft, Amazon, Meta, Google) — among the most pricing-disciplined buyers on earth, who routinely commoditize their suppliers. Medical and aerospace buyers are stickier, but they are smaller mix.

3. Supplier power — Low-to-Moderate. Inputs are copper, brass, gold/tin plating, engineering plastics, and rare-earth magnets for some sensor lines. Commodity inputs with multiple sources; pass-through pricing works with a one-to-two-quarter lag. Labor in low-cost geographies is the second supplier-power vector and has been rising but remains manageable. Not a structural problem.

4. Threat of new entrants — Low at the high end, Moderate at the low end. Designing a connector that survives a 15-year automotive thermal-cycle qualification is a genuine engineering moat. Designing a USB-C connector for a consumer device is not. New entrants (mostly Chinese) have been climbing the engineering ladder credibly and are now at parity for many EV applications. The barrier is being lowered slowly, not collapsed.

5. Threat of substitutes — Moderate and growing. This is the under-appreciated force. EV architectures need fewer connectors per vehicle than ICE platforms (centralized compute, zonal architectures, high-voltage backbone). Wireless replaces some board-to-board connectors. Optical replaces some copper interconnect in data centers. Each substitution removes content per platform even when TEL wins the platform.

Value pool location and trajectory. The value pool is migrating from auto-ICE (shrinking, structurally) toward EV (TEL is winning share but at lower content-per-vehicle and lower margins), data-center high-speed interconnect (growing fast, very competitive on price), industrial automation (steady, decent margins), and medical/aerospace (small, high margins, slow growth). Net trajectory is sideways-with-a-mix-shift-headwind, which is consistent with the 2.09% incremental returns observed.

Industry Verdict: Average.

Inversion (Bear Case)

I am playing a short-seller. No hedging.

1. The single event that kills this. A 2026-2027 confirmation that TEL has lost a meaningful platform — a Volkswagen Group ID-series re-source, or a Tesla high-voltage harness migration to a Chinese supplier, or a hyperscaler standardizing on Amphenol for next-gen 224G interconnect — combined with a quarter where book-to-bill drops below 0.9x. The headline risk isn't a single contract; it's the market realizing that 'design-in stickiness' has a half-life now that EV platforms redesign every 4-5 years instead of ICE's 8-10. One credible loss + one weak quarter and the multiple compresses from ~30x EV/FCF toward the high-teens that engineered-component peers like Vishay, Sensata, and Belden trade at. From $207, that is a $90-110 stock.

2. Why the moat is narrower than bulls think. Bulls cite 'design-in' and stop. They don't ask: design-in to what, for how long, and at what margin? In ICE, a wire harness designed in 2018 was still shipping in 2030. In EV, the harness is being redesigned every program, the connector count is dropping (zonal architectures use 30-40% fewer connectors per vehicle), and the dollar content per vehicle is flat-to-down despite the bull narrative of 'more electronics per car.' Meanwhile, the qualification standards that used to be a barrier are now codified in IATF 16949 and CQI standards that Chinese suppliers like Luxshare meet. The moat that produced 11.7% ROIC on legacy capital is real. The moat that will defend incremental capital is much thinner — and the 2.09% ROIIC is the company telling you that, in numbers, right now.

3. Why management is worse than it appears. Curtin and his team look fine: clean comms, fortress balance sheet, no scandals. But the actual job of a capital allocator at this point in the cycle is to recognize that incremental returns have collapsed and to redirect capital from organic reinvestment and acquisitions to large-scale, price-disciplined buybacks. They have not done that. Buybacks have continued at multiples of intrinsic value (px/IV of 8x); acquisitions have continued at full prices that depress consolidated returns; capex has not been cut. The single most important number a CFO should track — 'are my marginal dollars earning above cost of capital?' — has been answering 'no' for five years, and the response has been to keep deploying. That is not competence; that is institutional inertia dressed in industrial khakis.

4. What bulls are extrapolating that won't hold. Three extrapolations, all wrong. (i) 'EV is a content-per-vehicle tailwind.' The data says EV is a margin-and-architecture headwind for connector incumbents because Chinese OEMs vertically integrate and Western OEMs zonal-architect away connectors. (ii) 'Data center AI is a multi-year tailwind.' True, but the value capture in 224G/448G high-speed interconnect is concentrating with Amphenol and a handful of optical-module makers; TEL is a participant, not a leader, and hyperscaler buyer power is brutal. (iii) 'They'll grow into the multiple.' At 8.12x base IV, they need to triple owner earnings to grow into a fair multiple. Owner earnings TTM are $322M; tripling them in five years requires a ~25% CAGR, against a 5-year ROIIC of 2.09%. The math is not close.

5. Valuation trap (multiple compression / regime change). TEL trades at EV/FCF of 26.8x. Industrial-component peers (Sensata, Vishay, Belden, Littelfuse, Methode) trade between 9x and 18x EV/FCF. The 10-year average P/E is 20.7x. A re-rating to peer averages — not a worst case, the average case — implies multiple compression of 35-50%. Add a single guide-down quarter and you get a 50-60% drawdown without anything 'going wrong' fundamentally. The regime change is the move from 'AI-adjacent industrial compounder narrative' to 'mid-cycle engineered-component oligopolist.' The narrative is not warranted by ROIIC. Narratives that aren't warranted by economics get repriced.

If I am right, the stock could be worth $80-110 within 2-3 years.

Lollapalooza Bias Check

Active biases in the analyst (me) right now:

Authority and social proof. TEL is in the S&P 500, owned by every quality-industrial fund and most ESG screens, covered by every major sell-side desk, and consistently described as a 'quality compounder.' That collective endorsement creates an unconscious gravitational pull toward 'find a way to like it.' I am noticing this and pushing against it. The corrective: ignore the membership cards and look only at the numbers — ROIIC 2.09%, px/IV 8.12x. Those numbers don't care about S&P 500 inclusion.

Anchoring on legacy ROIC. The 11.7% 10-year average ROIC is a beautiful number and it anchors my impression of the business. But anchoring is a cognitive trap when the relevant number for forward returns is incremental, not historical. The mind wants to use the easier-to-recall number; the discipline is to use the relevant one. ROIIC is the number that matters, and it is one-fifth of legacy ROIC.

Confirmation bias toward 'industrial compounders are good.' Buffett owns Precision Castparts, Iscar, TTI [5][6]. The pattern-match 'engineered industrial component maker = Buffett-quality compounder' is seductive. But the test is not pattern-match; the test is whether THIS specific business at THIS specific price clears the four-question Munger filter. Iscar and TTI were bought private at owner-earnings multiples in the high single digits; TEL trades at 26.8x EV/FCF.

Recency bias on AI-data-center narratives. Every connector and sensor maker has been buoyed by 'AI infrastructure' commentary in 2024-2025. The recency of those headlines makes me overweight them in forecasting. Concrete corrective: TEL's data-center exposure is real but minority of revenue, and is the most price-competitive segment they operate in.

Commitment / consistency on the bear case. Once I wrote the inversion section, there is now a pull toward defending it (commitment-and-consistency tendency). The corrective is to ask: what would I need to see to flip bullish? Answer: a $40 stock price, a ROIIC trending toward 8%+, or a confirmed mega-buyback at distressed prices. None of those conditions are present.

Deprival super-reaction (FOMO). TEL has compounded at a respectable rate over a decade and the stock has outperformed in 2024-2025. Skipping it feels like missing something. The corrective: the question is not 'have I missed it' but 'is the present forward-return distribution attractive at this price.' At 8.12x base IV, the distribution is asymmetric to the downside.

Net effect: the biases push me toward 'Hold' or 'mild Buy.' Stripping them, the answer is Pass / Avoid at this price.

10-Year Outlook

Same fundamental business model in 10 years? Mostly yes — TE Connectivity will still be making engineered connectors, sensors, and electrical components for cars, factories, data centers, medical devices, and aircraft. The product taxonomy will look similar; the platform mix will have shifted further from ICE to EV, from copper to optical in some segments, from human-driven to automated factories.

Customer base larger? Roughly the same. The end markets — auto, industrial, comms, medical, aerospace — are not in secular decline, but they are not high-growth either. Auto unit volumes globally have been flat for a decade and may be lower in 10 years if shared mobility scales. Industrial and aerospace are GDP-plus. Data center is the one genuinely growing customer base; TEL participates but does not lead.

Profit per customer higher? This is the load-bearing question, and the honest answer is uncertain leaning negative. The structural pressures are: zonal EV architectures reduce connector count per vehicle, hyperscaler buyer concentration compresses pricing, Chinese tier achieves qualification parity, and engineering plastics/copper costs trend up. The structural tailwinds are: more sensors per vehicle and per factory, AI data-center high-speed interconnect, electrification content premium in industrial. Net: profit per customer flat to modestly down in real terms is the central case.

Moat wider? No. The moat that produced 11.7% legacy ROIC was built in an era of long ICE platform cycles and a duopolistic Western-Japanese supplier base. The moat that defends incremental capital today is narrower because EV platforms refresh faster and the supplier base now includes credible Chinese players. The 2.09% ROIIC is the moat, measured. The moat will likely be the same width or narrower in 10 years, not wider.

Single biggest threat? Loss of design-in advantage on EV platforms to Chinese vertically-integrated suppliers, combined with simultaneous compression in data-center pricing power. Either alone is manageable; both together compresses ROIC toward the cost of capital and removes the entire premium currently embedded in the stock.

The business will be recognizable in 10 years. The forward returns at this price almost certainly will not match the past returns at past prices.

CONFIDENCE: medium