Texas Instruments

Phase A — Understand the business

Company Overview

TI designs and manufactures analog and embedded-processing semiconductors — the un-glamorous, high-margin "jellybean" chips that condition signals and manage power inside essentially every piece of electronic equipment. Operations began in 1930; incorporated in Delaware; HQ Dallas. 2025 revenue $17.68B across two reportable segments + Other:

• Analog — $14.01B (79% of revenue): Power (DC/DC regulators, battery management, power switches) + Signal Chain (amplifiers, data converters, interface, motor drivers, sensing).
• Embedded Processing — $2.70B (15%): microcontrollers, processors, wireless connectivity, radar.
• Other — $979M (~6%): DLP, calculators, custom ASICs.

The portfolio is >80,000 products sold to >100,000 customers. Contract structure is the opposite of a take-or-pay infra business: short standard commercial terms, point-in-time revenue recognition on shipment or consignment-pull, no resale contingency. Concentration is low by design — about half of revenue comes from outside the top 50 customers; only one end customer = 12% of revenue in both 2025 and 2024 (recognized in Analog — unnamed in the filing, but the consensus read is Apple). >80% of revenue is now direct (incl. TI.com), <20% through distributors. End markets (2025): Industrial 33%, Automotive 33%, Personal Electronics 21%, Data center 9%, Communications 3%, calculators ~1%.

The stated objective is singular and worth quoting because it governs every other lens: "the best metric for owners to measure our progress is through the growth of free cash flow per share over the long term."

Supply Chain

TI is vertically integrated — the defining structural fact. It owns and operates wafer-fab + assembly/test in North America, Asia, Japan and Europe, and in 2025 sourced the majority of both wafer fab and assembly/test internally.

Upstream → TI → end customer, named:

• Inputs: silicon wafers, gases, photoresist, leadframes/substrates, and — critically — wafer-fab equipment from the lithography/deposition/etch oligopoly (ASML, Applied Materials, Lam Research, Tokyo Electron, KLA). TI does NOT name these in the 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes. (it says "diverse set of suppliers globally") but the 300mm build-out is an equipment-intensive program. ]
• TI fabs: 300mm fabs in Richardson & Sherman, Texas, and Lehi, Utah (LFAB), ramping; legacy 150mm being closed; 200mm in South Portland (Maine), Aizu/Miho (Japan), etc. PP&E by geography: US $8.76B of $12.32B total (71%), Malaysia $1.40B, Rest of Asia $2.45B, China $674M.
• Foundry supplement: TI selectively uses outside foundries/subcontractors for overflow and advanced-logic process development it does not own.
• Channel: >80% direct (TI.com + sales force); <20% via a single worldwide distributor + regional distributors.
• End customers: the ~33/33/21/9 industrial/auto/PE/datacenter mix above — i.e. the entire electronics economy, with autos (infotainment/ADAS/EV powertrain) and industrial (factory automation, energy infra, medical) the strategic priority.

Chokepoints: (1) the single 12%-of-revenue end customer; (2) China — end-customers HQ'd in China = ~20% of revenue, but products shipped INTO China = ~50% of revenue. That ship-to figure is the real China exposure. (3) WFE equipment access under US/China export controls. TI's counter is the "geopolitically dependable capacity" pitch: own the fabs, keep them mostly in the US/allied geographies.

Competitive Advantages (moats)

TI states four moats; the durable two are real, the other two are table-stakes dressed up:

1. Manufacturing & technology cost advantage (REAL, the core moat). A chip on a 300mm wafer costs ~40% less than on 200mm. TI is the only analog player at meaningful 300mm scale and is internalizing toward >95% of wafers in-house by 2030, >80% on 300mm. This is a structural per-unit cost wedge competitors using foundries or 200mm cannot match — it is the source of the gross margin and the FCF.
2. Catalog breadth + design longevity (REAL). 80,000 products, designed into systems that ship for 10–20 years; in Embedded, customers write their own software on TI parts, raising switching costs. Low single-point dependency. The "diversity & longevity" advantage = an annuity book that decays slowly.
3. Reach of market channels — the direct/TI.com shift is a genuine data/insight advantage but is closer to a sustained operational edge than an unassailable moat.
4. Broad portfolio — overlaps with (2).

Bargaining power: high over customers for catalog parts (no one part is critical, but the breadth + lead-time dependability make TI hard to design out wholesale); moderate over suppliers (the WFE oligopoly has its own pricing power). Net: TI is a price-maker in its catalog — evidenced by the 2025 China price hikes (below) — but that power is being tested at the low end by Chinese entrants. Ground truth: positioning.md / bottlenecks.md exist on the hardware shelf but were not separately quoted; the moat read here is filing-grounded.

Segments

All `` from the 10-K (FY) and 10-QForm 10-QThe quarterly version of the annual report. Lighter, and not audited. (Q1'26). Segment operating profit excludes unallocated items.

FY2025 vs FY2024:

Q1 2026 vs Q1 2025:

Trend & cause: Analog is accelerating (+15% FY → +22% Q1 YoY) and its margin is expanding (37.6% → 41.7%) as factory loadings rise — operating leverage on a mostly-fixed cost base. Embedded is the laggard: margin cratered to 6.2% in Q1'25 then recovered to 16.9% as LFAB (Lehi) ramps — management explicitly flags Embedded carries disproportionate manufacturing cost until LFAB loads up, after which it "disproportionately benefit[s]" Embedded OP. Geography (Q1'26): US 37%, EMEA 22% (Germany alone 10%), China 21%, Rest of Asia 12%, Japan 6%. Embedded is the structural underperformer and the reason TI is buying Silicon Labs (Lens 5/12).

---

Phase B — Measure performance

Earnings Result (latest print: Q1 2026, reported 2026-04-22)

The most recent quarter is a clean, large beat with an above-Street guide — the cyclical recovery in full swing.

Q1 2026 actuals:

• Revenue $4,825M, +9% sequential, +19% YoY (vs ~$4.5B consensus) ]
• Gross profit $2,799M = 58.0% GM (up from 56.8% Q1'25)
• Operating profit $1,808M = 37.5% OM (up from 32.5%) — the operating-leverage story
• Net income $1,545M; diluted EPS $1.68 vs $1.28 Q1'25 (+31%); vs ~$1.37 consensus — a ~23% EPS beat ]
• Effective tax rate 10% (up from 8%, OBBBA effects)

What drove it: industrial >+30% YoY and data center ~+90% YoY (>$1B annual run-rate now). Analog led, Signal Chain ahead of Power.

Guidance (from the earnings release/call, not the 10-Q): Q2'26 revenue $5.0–5.4B, EPS $1.77–$2.05 — midpoint ~$5.2B / $1.91, above Street's ~$4.86B / ~$1.58, described by management as "slightly above seasonal". Tone shifted from the defensive July-2025 stance (Lens 8) to constructive.

Balance-sheet flags: inventory $4.80B at YE'25, days of inventory 222 (down from 241) — improving; DSO 40 (vs 39) — benign; FCF inflecting hard (Lens 11). Market reaction: stock rose on the print; TXN is +63% YTD to ~$331 (June 2026). The tape says the FCF-recovery thesis is being aggressively re-rated in real time.

Earnings Calls (sentiment trend)

No transcripts/ on the shelf (folder empty) — sentiment read is `` across the last ~4 calls:

• Q2 2025 (Jul 2025) — sharp negative. Stock −13%; CEO Ilan warned the auto market is struggling with 25% import tariffs, and flagged that some Q2 strength was a tariff-driven pull-forward. Defensive, cautious.
• Q3 2025 (Oct 2025) — soft. Earnings + guidance "miss," stock fell; analysts mixed.
• Q1 2026 (Apr 2026) — decisively positive. "Best-ever Q1," industrial recovery "broad across all sectors, geographies and customer sizes — sequential growth for the first time in years," data center +90%. Multiple upgrades followed.

Recurring phrases: "free cash flow per share," "300mm," "geopolitically dependable capacity," "broad-based recovery," "disciplined capital allocation." Stopped saying: the heavy-defensiveness around tariff pull-forward that dominated mid-2025. The arc over four quarters is trough-pessimism (mid-2025) → recovery-confidence (early 2026) — which is exactly why the stock has doubled off the lows and why the risk is now that sentiment is priced for perfection.

Comps

Multiples are `` (June 2026) or n/a.

5-yr avg ROE for the peer set: n/a (do not fabricate). TXN's own returns are strong: 2025 ROE ≈ $5.00B / ~$16.6B avg equity ≈ ~30%. Read: TXN and ADI sit together at the top of the analog multiple range (~34–38x fwd); MCHP is the value/cyclical end (~18x). TXN is priced at/above the premium peer, and notably above the Street's own average price target (mean PT ~$275–294 vs $331 spot) — i.e. the consensus analyst no longer sees upside to the average target; the bull case lives entirely in the Street-high ($400, Seaport).

Stock-Price Catalysts (>5% moves, ~5yr)

Pattern is overwhelmingly earnings-guidance-driven, with a secular China/tariff overlay:

• Jul 22–23, 2025: −13% on weak Q3 guide + tariff/auto warning.
• Oct 22, 2025: down on Q4 guide miss.
• 2024 (ongoing): Elliott catalyst. May 28, 2024 Elliott disclosed a $2.5B stake; the stock and the capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. debate became the story for 12+ months.
• Apr 2026: up on the Q1 beat + above-Street guide; +63% YTD into June 2026, all-time-high close ~$324.89 (May 26, 2026).
• Recurring: any China analog-substitution / tariff-investigation headline (2025) pressures the stock.

What the market actually reacts to for TXN: (1) forward guidance vs seasonal (the single biggest lever), (2) capex / FCF-per-share trajectory (the Elliott axis), (3) data-center AI-power narrative (the 2026 upside lever), (4) China policy (the downside lever). It is NOT a story stock on a single product — it trades on cycle position + capital-return math.

---

Phase C — Judge people & books

Management

• Haviv Ilan — Chairman, President & CEO since April 1, 2023. Israeli EE, ex-startup founder; ~25-yr TI insider (SVP 2014 → EVP/COO 2020 → board 2021 → CEO 2023), named chairman subsequently. Internal-promote archetype, steeped in TI's manufacturing-and-FCF religion — continuity, not a turnaround agent.
• Rafael Lizardi — SVP & CFO, 5+ years as executive officer; the discipline-keeper.
• Track record / capital allocation: this is TI's signature. Under predecessor Rich Templeton, TI grew FCF/share by >6x and built "the most distinctive capital-allocation philosophy in the semiconductor industry". Over 2016–2025 TI allocated $109B, ~$24B of it to capex. 22 consecutive years of dividend increases (raised 4% to $1.42/qtr; annual ~$5.68). 2025 returned $6.48B to holders ($5.00B dividends + $1.48B buyback).
• Skin in the game / dilution: SBC only $419M (~2.4% of revenue) in 2025 — low and clean for a chipmaker. Share count is falling (913M diluted 2025 vs 919M 2024) via buyback — accretive, not dilutiveDilutionIssuing new shares, so each existing share owns a smaller slice of the same company.. our figures not on the shelf, so insider-ownership %: n/a.
• Red flags: the one genuine governance tension is that management's 2022 capacity plan was aggressive enough to draw Elliott (overshoot demand by ~54% in 2026 per Elliott's math) — i.e. capital-allocation discipline was questioned, and the activist effectively forced the 2026 capex cut to $2–3B. Not fraud — a strategy-vs-shareholder dispute that management has now substantially conceded. Founder vs professional manager: professional internal-promote at a culture that genuinely thinks like owners; implies steady-hand stewardship, low surprise risk on governance.

Forensic Red Flags

TI is, forensically, one of the cleaner large-caps — but the capital-return math is the thing to watch.

• Revenue recognition: vanilla. Point-in-time on shipment/consignment-pull; no resale contingency for distributors; no long-dated percentage-of-completion games. Low risk.
• Cash vs earnings: CFO $7.15B > net income $5.00B (2025) — earnings are well cash-backed (D&A $1.92B is the bridge). No divergence flag.
• The real flag — capital returns >> FCF. 2025: returned $6.48B on FCF of $2.94B — a ~2.2x payout of free cash flow, funded by drawing down short-term investments (−$2.7B total cash) and net new debt (+$0.45B). Sustainable only because (a) the balance sheet had a big investment buffer and (b) FCF is now inflecting up hard (Lens 11). If the FCF recovery stalls, the dividend (22-yr streak, a quasi-promise) + buyback are funded from the balance sheet — watch total cash ($4.88B at YE'25, down from $7.58B) and leverage.
• Receivables/inventory vs revenue: both improving (DSO 40, DOI 222 ↓). Inventory built ahead of demand by design (low-obsolescence catalog) — a deliberate, defensible policy, not a stuffing signal.
• Goodwill/intangibles: goodwill $4.33B (12.5% of assets), and TI took a non-cash goodwill impairment on custom ASIC in 2025 (inside Restructuring/other $117M) — small, disclosed, honest. Silicon Labs will add goodwill in 2027.
• SBC flattering non-GAAP: TI barely uses non-GAAP — it reports GAAP EPS and only adds free cash flow as a non-GAAP measure (transparently reconciled, incl. the CHIPS-Act-proceeds add-back). Refreshingly low gamesmanship. One nuance: FCF is defined as CFO − capex + CHIPS Act incentives — i.e. TI adds government cash back into "free" cash flow ($335M in 2025). Defensible but worth knowing.
• Debt: $14.05B total, almost all long-dated fixed-rate low-coupon (e.g. 2063 at 5.05% $1.55B, 2048 at 4.15% $1.5B, several 1–2% notes) — cheap, terming out the capex cycle. Fair value $13.05B < carrying (rates rose). Net debt ~$9.2B vs ~$7.2B EBITDA → ~1.3x. Comfortable.

Regulatory findings (required):

• SEC: 0 Litigation Releases and 0 AAERs naming TI, 2021-06-24 → 2026-06-24, verified via EDGAR EFTS (LR + AAER).
• 10-K Item 3 (Legal Proceedings): TI's own disclosure — "involved in various inquiries and proceedings that arise in the ordinary course… we believe that the amount of our liability, if any, will not have a material adverse effect." Elects a $1M environmental-proceeding disclosure threshold. No material litigation disclosed.
• Non-SEC: the live, material item is China's regulatory/antidumping scrutiny of analog imports — China is investigating interface and gate-driver chips (CAN/RS485, isolated/multichannel gate drivers); TI has ~11.4% revenue exposure to the targeted categories (the highest among US vendors; ADI 7.8%, ON 10.2%). This is a trade/industrial-policy risk, not an enforcement action against TI's conduct. The Silicon Labs deal will face standard merger antitrust review (incl. plausibly China SAMR), closing 1H 2027.
• Net: no accounting or fraud findings across SEC EFTS, web, and 10-K Item 3 as of 2026-06-24. The only regulatory overhang is geopolitical/trade, already covered in Lenses 2/12/13.

---

Phase D — Project & stress-test

Forward Projection (FY2026 / FY2027 / FY2028 EPS)

Built bottom-up from FY2025 actuals + Q1'26 + the Q2 guide. All outputs `` with arithmetic shown; no our model logged (watchlist breadth — no committed forecast).

Anchors: FY2025 rev $17.68B, dil EPS $5.45; Q1'26 rev $4.825B / EPS $1.68; Q2'26 guide mid ~$5.2B / EPS ~$1.91; 2026 capex guided $2–3B (down from $4.55B) — the FCF lever; consensus FY2026 EPS ~$7.71 and revenue ~$21B.

• FY2026 BASE — EPS ≈ $7.70. H1'26 ≈ $1.68 + $1.91 = $3.59 already booked/guided. Assume H2 normal-to-slightly-above seasonal (TI's H2 is typically ≥ H1) → H2 ≈ $4.05–4.20 → FY ≈ $7.65–7.80. Revenue ~$20.5–21B (+16–19%), GM ~58–59%, OM ~37–38%, share count ~912M, tax ~11%. → converges with the ~$7.71 Street number.
• FY2027 BASE — EPS ≈ $8.90. Revenue +12% to ~$23.5B (continued analog/industrial/data-center recovery, no recession), GM to ~59–60% on rising factory loadings + capex rolling off depreciation pressure, OM ~39%; buyback shrinks shares ~1%; Silicon Labs closes 1H'27 (modestly accretive after synergies). → ~$8.80–9.00.
• FY2028 BASE — EPS ≈ $10.20. Revenue +10% to ~$25.8B, GM 60%+, full SiLabs synergy run-rate ($450M), shares −1–2%. → ~$10.00–10.40.

Bull path: data center compounds >50%/yr off >$1B base + industrial broadens → FY2028 revenue ~$28B, GM 62%, EPS ~$11.50+. Bear path: China share loss + auto stays weak + a 2027 cyclical roll-over → FY2027 revenue flat/down, GM back to ~55%, EPS ~$7.00 (below 2026).

FCF/share — the actual thesis. TTM FCF (to Q1'26) already $4.4B (23.6% of revenue) vs $1.5B (9.6%) in FY2024. With capex falling to $2–3B on ~$21B revenue, management says FCF/share could exceed $8 in 2026; Elliott's demand was $9+. At ~910M shares, $8/sh = ~$7.3B FCF. The entire investment case is this number going from ~$1.50 (2023 trough) to $8+ (2026) — and that re-rating has largely happened in the stock already.

Bull vs Bear

Bull case. TI owns the only 300mm-scale analog cost moat (~40% per-chip advantage), an 80,000-SKU annuity book with 10–20yr design lifecycles, ~30% ROE, and — the timing hook — it is exiting a 6-year elevated-capex cycle exactly as the analog cycle recovers, so capex collapses ($4.55B→$2–3B) while revenue and loadings rise. That is a textbook FCF inflection: FCF margin 9.6%→23.6% in four quarters, FCF/share $1.50→$8+. Add a genuine secular leg in data-center AI power (+90% YoY, >$1B run-rate) layered onto the cyclical recovery, a 22-yr dividend-grower's capital-return machine, and CHIPS Act ITC (35%) + ~$1.6B direct funding subsidizing the US fabs. Quality + inflection + a secular kicker.

Bear case — three things that could permanently or durably impair:

1. China designs TI out at the low/mid end. ~50% of products ship into China; ~11.4% of revenue is in the exact categories China is targeting for domestic substitution; SG Micro / 3Peak / Chipsea are climbing the catalog. TI's largest-ever price hike on 60,000+ China parts (+10% to >30%) is a defensive margin-for-share trade that can lose both if Chinese buyers localize. This is a slow structural erosion of the catalog moat in TI's biggest ship-to market — the most under-priced risk.
2. The cycle is the catalyst. TXN trades on forward guidance vs seasonal. After +63% YTD and above the Street's average PT, any guide that is merely in-line (not above) re-rates it down — see the −13% July-2025 reaction as the template. Auto end-market (33% of revenue) is still tariff/EV-pressured.
3. Capital returns out-running FCF if the recovery stalls. 2025 returned 2.2x FCF; the 22-yr dividend streak is a quasi-obligation; a stalled FCF recovery forces balance-sheet-funded returns.

Pre-mortem (18 months out, thesis broke): It's late 2027. The data-center pull was partly an inventory build that air-pocketed; China substitution took 300–400bps of analog share; auto stayed soft; FCF/share landed ~$6.5 not $9. The stock de-rated from ~43x to ~28x forward on a now-lower EPS — a double compression. The franchise is fine; the 2026 price was the mistake.

Are multiples too high? Yes, on a cyclical-mid basis. ~38x NTM / ~43x on consensus EPS, above the Street's own mean target, for a high-teens-grower whose best end-market just printed +90% (i.e. tough comps coming). You are paying a secular-grower multiple for a cyclical-recovery franchise at the optimism peak of its cycle.

Contrarian view (what the market is refusing to see): Consensus treats the data-center +90% as the new secular growth identity. The market is under-weighting that (a) most of TI's body is still auto + industrial + personal electronics — late-cycle, China-exposed, tariff-sensitive — and (b) the FCF inflection is now consensus, not a discovery (FCF/share $8+ is in every model, the stock has doubled, Elliott already collected the capex win). The asymmetry has flipped: the easy money in "TI's FCF will recover" has been made; what's left is paying up and hoping the cycle doesn't roll.

Devil's Advocate (short-seller)

Dismantling the bull case.

• What structurally breaks the money machine? The 300mm cost moat is only a moat if TI keeps the volume that fills 300mm fabs. The fixed-cost base that gives 58% GM at high loadings gives sub-50% GM at low loadings (management says so explicitly). A China-driven volume/share loss is the one thing that turns the moat into a margin trap — fixed costs spread over shrinking output.
• Revenue concentration & shift. ~50% ships into China; one customer = 12%. If China policy accelerates localization in TI's targeted categories (gate drivers, interface), that's a multi-year, hard-to-reverse share leak in the biggest ship-to geography — and unlike a cyclical dip, design wins lost to a local champion don't come back at the next upturn.
• Most dangerous competitor bulls underestimate: not ADI — it's the Chinese long tail (SG Micro, 3Peak, Chipsea) that bulls dismiss as low-end, exactly as incumbents always dismiss disruptors before the catalog climbs.
• Worst capital-allocation move: the 2022 over-build that drew Elliott — TI committed to capacity Elliott calculated would overshoot 2026 demand by ~54%, depressing FCF/share >75% from the 2022 peak. Management was right about the moat, wrong (per the activist) about the timing/size; they've since conceded the 2026 cut.
• Assumptions that must hold for ~$331: (1) FY2026 EPS ~$7.7 AND FY2027 ~$9 — i.e. an uninterrupted up-cycle through 2027; (2) GM holds ~58–60% (no China price war contagion); (3) data center stays a secular +50% leg, not an inventory pull; (4) the ~40x multiple doesn't compress. If growth disappoints 20–30% (FY2027 EPS ~$6.3–7.0 instead of ~$9) and the multiple normalizes to ~28x, that's ~$185 — roughly −44% from spot. The downside is a valuation unwind, not a solvency event.
• Single permanent-impairment scenario & plausibility: China achieves credible domestic substitution across interface/power/gate-driver analog (state-backed, already underway) → TI structurally loses its biggest ship-to market's mid-tier and the catalog moat narrows to the developed-world/auto-qualified high end. Plausibility: moderate and rising — it is policy-driven, multi-year, and TI's own defensive price hikes are evidence the pressure is real. Not a 2026 event; a 2027–2030 erosion.

Management Questions (ordered by information value)

1. China is ~50% of products shipped and ~11% of revenue is in the exact categories being targeted for domestic substitution — quantify the design-win win/loss rate vs Chinese analog vendors over the last 4 quarters in interface, gate-driver and power. Are you holding share or buying time with price?
2. The 60,000-part China price increase: how much was margin-for-share vs share-for-margin — i.e. what happened to China unit volumes after the hikes?
3. FCF/share — bridge me from $8+ (2026) to a steady-state number. How much is cyclical loadings vs the structural capex roll-off, and what's the through-cycle floor?
4. Capacity utilization by fab today, and the revenue level at which GM rounds-trips back below 55% if the cycle rolls.
5. Data center +90% — what share is genuine sell-through into deployed AI racks vs distributor/customer inventory build? What's the sustainable run-rate growth?
6. Silicon Labs — beyond the $450M synergies, what does SiLabs do for Embedded's structural margin problem, and what's your confidence on 1H'27 close given antitrust (incl. China SAMR)?
7. Capital returns ran 2.2x FCF in 2025. What's the policy if the FCF recovery stalls — protect the buyback or the dividend growth streak?
8. LFAB/Embedded — at what loading does Embedded reach Analog-like incrementals, and when?
9. Post-2026, capex is "demand-dependent." What revenue CAGR triggers a return to elevated capex, and would you risk repeating the 2022 over-build?
10. What's your honest read on the auto end-market (33% of revenue) through 2027 given tariffs and the EVEnterprise valueWhat it would cost to buy the whole company: its market value plus its debt, minus the cash you would get with it. Often very different from market cap. slowdown?
11. CHIPS Act: ~$2.9B of incentives still receivable + 35% ITC. What's the cash-timing and the clawback exposure under a changed political environment?
12. Where are you losing on the 80,000-SKU catalog today — which product lines are ceding share, and to whom?
13. The single 12%-of-revenue customer — how concentrated is the design exposure, and what's the trend?
14. Pricing power: outside China, are catalog ASPs holding, rising, or rolling as competitors add 300mm-class capacity?
15. If you could change one element of the 2016–2025 capital-allocation record, what would it be?

---