Skip to content
primarysourced Photonics sector Tower
TSEM
~12 min read · 2,711 words ·updated 2026-04-29 · confidence 67%

Capital returns — dividend, buyback, share count, $353M Intel-fee deployment

As of: 2026-04-29 (data through FY 2024 20-F filed 2025-04-30 + Q4 2025 6-K release filed 2026-02-11)

Confidence legend: ✓ verified-primary 20-F / 6-K · ◐ partial / aggregator · ⚠ inferred / estimate

1. The headline — Tower has zero capital-return program

Tower Semiconductor has historically run a 100% reinvestment capital-allocation policy. The combination of:

  • No dividend (none paid since the company’s modern listing structure)
  • No active share repurchase program (no current authorization)
  • No history of opportunistic open-market repurchases

…makes Tower structurally unique among NASDAQ-listed merchant foundries. GlobalFoundries (NASDAQ: GFS) authorized its first $500M repurchase program in February 2026. TSMC runs a continuous buyback. UMC pays a substantial dividend. Vanguard International Semi (VIS) — the closest pure-play comp — pays out 70-80% of earnings as dividend (Taiwanese specialty foundry standard).

FY 2024 20-F dividend disclosure (paraphrased): Tower’s articles do not contractually limit dividend declarations beyond standard Israeli company-law solvency tests. Management has historically prioritized capacity-extension investment over capital returns. There is no current dividend policy and the Board has not signaled plans to initiate one.

FY 2024 20-F repurchase disclosure (paraphrased): No current share-repurchase authorization. Tower has not historically maintained a treasury-share program at scale. The diluted-share-count progression (FY 2023 → FY 2024 → FY 2025) shows modest annual dilution of ~1-2M shares per year from employee equity-plan vesting net of withholding — i.e., net share count is increasing modestly rather than decreasing.

2. The $353M Intel termination fee — fully reinvested, not returned

The single largest discretionary capital event in Tower’s recent history was the receipt of the $353M Intel termination fee in August 2023 (EE Times ✓; Tower release ✓) following the mutual termination of the $5.4B Intel acquisition. The fee was paid in cash to Tower at the August 16, 2023 mutual-termination event.

How was the $353M deployed? Per public record: 100% reinvested into balance-sheet liquidity / capex — zero returned to shareholders.

Tracing the $353MTreatment
August 2023: $353M cash received from IntelRecorded as “other income” in FY 2023 P&L; cash inflow to balance sheet
FY 2023 P&L impactFY 2023 net profit inflated to $519M (vs ~$166M operating-business net profit ex-fee)
FY 2023 balance-sheet impactCash + ST deposits at FY 2023 YE = $1,244M — the post-Intel-fee anchor liquidity level
FY 2024 deploymentTower deployed ~$432M capex against $449M OCF; cash + deposits stayed roughly flat at $1,218M YE
No dividend declaredNone
No special distributionNone
No share repurchase initiatedNone
No tender offerNone

Implication. Management’s choice to retain the entire $353M for capacity reinvestment is the clearest historical signal of Tower’s capital-allocation philosophy. By August 2023 the bull thesis (silicon-photonics merchant foundry + RF Infrastructure pivot) was already forming; the cash was deployed into the Tower-Intel Fab 11X $300M equipment investment (announced September 2023, just 20 days after the Intel termination), capex on the Tower-ST Agrate 300mm JV continued ramp, and the Newport Beach Fab 3 SiPh capacity preparatory investment that ultimately led to the FY 2024 $432M / FY 2025 $437M capex run-rates.

A counter-factual: if Tower had returned the $353M as a special dividend, that would have been ~$3.16/share (against the FY 2023 share count of ~111M). At the time the stock was trading ~$30/share — a 10%+ special distribution that would have rewarded shareholders for the Intel-deal collapse rather than channeling the proceeds into capacity. That would have been the “yield-oriented” allocation choice — the opposite of the path taken.

3. Why no dividend? — strategic vs governance reasons

The structural reasons Tower has not paid a dividend:

3.1 Capacity-extension cycle prioritization

Tower has been in a continuous capacity-extension mode since the 2021 Tower-ST Agrate JV announcement. Annual capex has run $251M (FY 2021) → $377M (FY 2022) → $471M (FY 2023) → $432M (FY 2024) → $437M (FY 2025) — and is now ramping toward a $700-900M FY 2026 peak for the $920M SiPh/SiGe envelope. There has been no point in the past 5 years where Tower had FCF “available” for capital returns — every dollar of OCF has been absorbed by capex + debt service.

3.2 Israeli geopolitical-risk + cash-cushion preservation

Tower is incorporated in Israel with significant operating exposure to Israel (Migdal Haemek Fab 1 + Fab 2). Operating in a geopolitically volatile region has historically motivated cash-cushion conservation rather than capital-returns deployment. The ~$1B net-cash position provides:

  • Buffer against operational disruption from regional conflict events
  • Working-capital flexibility during Israel-supply-chain stress
  • Credibility with US/EU customers requiring supplier-stability assurances

3.3 Israeli company-law dividend mechanics

Israeli company law imposes solvency-test conditions on dividend declarations (similar to UK/EU but not US Delaware standards) — no contractual limitation, but distributions can only be made from “profits” as defined by Israeli company law (analogous to retained-earnings tests). This is administratively manageable but adds a complexity layer not present for Cayman / Delaware peers.

3.4 Foreign Private Issuer status + dividend withholding tax

Tower is an Israeli-incorporated Foreign Private Issuer (FPI). Dividend distributions to US shareholders would trigger Israeli withholding tax at 25% (general) or 15-20% (for shareholders below 10% threshold under US-Israel tax treaty). This makes Tower dividends materially less tax-efficient for US shareholders than US-incorporated peers. Most of Tower’s institutional ownership (~65%) is US-based; a dividend would be partially “wasted” in withholding tax for the dominant shareholder cohort.

Read. The combination of (1) capex absorption, (2) geopolitical cash-cushion, (3) Israeli legal mechanics, (4) withholding-tax friction makes a meaningful dividend authorization economically unattractive for Tower’s specific ownership structure and operating profile. The “no dividend” stance is structurally reinforced rather than a discretionary choice that could easily flip.

4. Why no buyback? — the cash-deployment squeeze

The buyback case is more discretionary than the dividend case (no withholding-tax friction; no Israeli company-law mechanics) — but Tower’s cash deployment is constrained by:

  1. The $920M SiPh/SiGe envelope consumes the cash deployment runway through FY 2027-2028. With FY 2026 capex peak of $700-900M against ~$470M OCF, Tower runs a $300M+ FCF burn that draws down the cash cushion. A simultaneous $200-500M buyback would push the cash cushion below the comfort threshold for a geopolitically-exposed foundry.
  2. No deleveraging dynamic — Tower is already net cash; there’s no debt-paydown discipline that drives an inflection toward shareholder returns.
  3. The FY 2028 model is the priority narrative — management is signaling the equity should be valued on the FY 2028 model ($2.84B / 39-40% GM / 31.7% Op margin) rather than current cyclical earnings. A buyback at 13.0× EV/Revenue + 98.7× P/E + 8.04× P/B would be value-destructive optically — it implies management thinks the stock is undervalued at peak-cycle multiples, which is a difficult message to deliver.
  4. Mubadala-type controlling-shareholder pressure absent — unlike GFS where Mubadala’s selldown trajectory creates a parallel “buyback supports float” narrative, Tower has no controlling shareholder post-Intel-deal-collapse. Float is fully distributed; buybacks would not have a structural float-supportive rationale.

The path to buyback initiation. Tower’s most plausible buyback-authorization window is FY 2027 H2 / FY 2028 H1 — after the SiPh/SiGe envelope peak deployment passes, FCF inflects positively toward $500M+ annually, and the FY 2028 model is partially de-risked by mid-year run-rate visibility. A first buyback would likely be modest in scale ($300-500M, matching the GFS first-authorization template) — consistent with management’s first calibration step rather than a permanent program.

5. Comparison to specialty-foundry peer practices

CompDividend policyBuyback policyNet cashCapital-return % of FCF
Tower (TSEM)None ✓None ✓+$1.04B0%
Vanguard Intl Semi (VIS)Yes — 70-80% payout ratio ◐None ◐(~$1B)~70-80%
DBHiTek (Korea)Yes — modest dividend ◐Yes — occasional buyback ◐(~$0.5B)~30-50% ◐
GlobalFoundries (GFS)None$500M auth Feb 2026 (first ever) ✓+$2.8B~43% (auth basis)
UMCYes — substantial dividend ◐None ◐(~$3.5B)~50-60% ◐
TSMCYes — quarterly NT$ dividendYes — continuous repurchase(~$15B)~40-50%
SMICNone ◐None ◐(~$3B)0%

Tower is in the “0% capital return” cluster alongside SMIC (state-controlled Chinese foundry that retains all cash for capacity build) and pre-2026 GFS. VIS, DBHiTek, UMC, TSMC all return material capital — Tower is structurally distinct from those payout-oriented Asian specialty foundries.

The closest functional peer to Tower’s allocation policy is GFS pre-Feb 2026 — both are reinvesting heavily through a capacity-build cycle, both have CHIPS-like or proprietary forward growth narratives, both kept cash on balance sheet pending capital-return initiation. GFS turned the corner at Feb 2026 with the $500M authorization. Tower is on the same trajectory but probably 12-24 months behind GFS — i.e., a first authorization plausible in FY 2027 H2 / FY 2028 H1 once the SiPh capex peak passes.

6. Implication for the bull / bear thesis

6.1 Bull-case implication — reinvestment compounds

If the AI-photonics ramp + FY 2028 model materialize, Tower’s reinvestment-heavy stance is compounding the upside:

  • Every dollar of OCF gets reinvested into capacity at high incremental ROIC (FY 2028 model implies ~20%+ ROE if achieved)
  • The $920M SiPh/SiGe envelope creates a step-function capacity expansion that translates directly to revenue multipliers
  • No “leakage” via dividend or buyback — full earnings reinvestment maximizes growth-CAGR optionality
  • Customer-prepayment-funded portion of capex provides asymmetric upside: customers fund growth that primarily benefits Tower equity holders

Pure-play bull math. $437M FY 2025 capex deployed at FY 2028 target operating margin of 31.7% on each capacity dollar produces ROIC of ~22-25% ⚠ on the $920M envelope — i.e., the envelope deploys at 2.5-3× the cost-of-equity hurdle. That’s the bull-thesis reinvestment math — it justifies the lack of capital returns on pure value-creation grounds.

6.2 Bear-case implication — concentration risk

If the AI-photonics ramp underperforms or specialty cyclicality returns, the reinvestment-heavy stance is concentrating the downside:

  • No cash cushion returned to shareholders means no shareholder participation in cyclical recovery margins
  • $920M envelope deployed but underutilized would create a multi-year drag from D&A on under-absorbed capacity
  • No buyback-support floor — if the stock breaks below $100, there’s no programmatic repurchase step-in to soften the move
  • Customer prepayments could face return / refund pressure if hyperscaler demand softens, creating a working-capital reversal at the worst possible time

Pure-play bear math. $920M envelope at low utilization (say 50% of capacity reserved) would produce ROIC of <10% — below cost of equity. The implied capacity-write-down scenario could produce a GFS-Q4 2024-like impairment (GFS took a $935M Malta impairment when legacy capacity proved over-provisioned). A Tower SiPh impairment of $300-500M is not unthinkable in a deep-bear scenario where 1.6T transceiver volumes prove disappointing in 2027-2028.

7. Share count trajectory — modest dilution from equity plans

DateTotal ordinary shares O/S (M)Source
2023-12-31~111.0 ⚠FY 2023 20-F basic weighted-average
2024-12-31111.76 ✓FY 2024 20-F cover page
2025-Q4 (avg basic)~112-113Q4 2025 6-K basic weighted-average
2025-Q4 (avg diluted)~114Q4 2025 6-K diluted weighted-average
2025-12-31 (estimated)~113-114 ⚠Implied + plan-vesting cadence

Annual dilution of ~1-2M shares per year (~1-2%) reflects:

  • Employee equity-plan vestings (RSUs to senior management + technical staff)
  • Modest grant-pool replenishment
  • No share issuance for M&A (Tower has been M&A-quiet since the Intel-deal-collapse cycle)
  • No PIPE / secondary issuance since the post-Intel-deal-collapse stabilization

Read on dilution. The ~1-2% annual dilution is at the lower end of the specialty-foundry peer range (GFS runs ~1.3% annual; LWLG and pre-revenue smalls run 5-10%+ via ATM facilities). Tower’s modest dilution rate reflects its profitable operating profile — equity grants are funded by retained earnings rather than dilutive cash-raises. The implication: the “no buyback” stance has a sub-2% annual dilution offset against it — i.e., shareholders are getting modestly diluted each year without offsetting buyback support. A reasonable ASR-based buyback would offset 100% of the dilution and provide modest accretion.

8. Form 4 insider activity — minimal, post-HFIAA effective 2026-03-18

Per the Onboarding context, Tower as an Israeli FPI was historically exempt from Section 16 Form 4 reporting for officer / director transactions. The HFIAA (Holding Foreign Issuers Accountable Act), signed 2025-12-18 and effective 2026-03-18, repealed the FPI Section 16(a) carve-out — Tower officers/directors are now subject to Form 4 reporting on transactions occurring on or after 2026-03-18.

Status as of 2026-04-29: First-12-months Form 4 archive for Tower will materialize through 2027 Q1. The current companies/tsem/data/form4_history.json archive is empty / pre-effectiveness. No actionable insider-trading signal yet from Form 4 channel.

Pre-HFIAA insider activity inference (via aggregator + Israeli company law disclosures):

  • CEO Russell Ellwanger has held the role since 2005; long-tenure CEO with substantial vested + unvested equity holdings
  • CFO Oren Shirazi has been in role since 2008; similarly long-tenured
  • No publicly disclosed insider sales / purchases at meaningful scale post-Intel-deal-collapse — signal absent rather than negative

Forward expectation. Once the post-2026-03-18 Form 4 archive accumulates 6-12 months of data, the insider-trading channel becomes a useful signal. Watch for first management Form 4 filings in the Q3 2026 / Q4 2026 timeframe as a sentiment indicator. Pre-HFIAA, Tower had effectively no insider-transaction signal — a structural data gap that the HFIAA effectiveness now closes.

9. Open items / backfill queue

  1. FY 2024 20-F dividend / repurchase disclosure direct extraction — primary 20-F document (zk2533083.htm) is HTTP-403-blocked from automated extraction. Need authenticated session or manual download for verbatim dividend / repurchase / treasury-share text. TODO ⚠.
  2. FY 2025 20-F filing (expected April-May 2026) — first full-year audited disclosure of post-$920M-envelope balance-sheet treatment of customer prepayments, deferred-revenue line, and any updated capital-return-related language.
  3. Israeli company-law dividend mechanics specific to Tower’s articles — verify retained-earnings-test thresholds + Board authority on distributions.
  4. Sell-side coverage on capital-return outlook — analyst notes (Citi, Needham, Stifel, JPMorgan, Susquehanna, Wells Fargo) likely contain forward modeling on dividend / buyback initiation; primary-source extraction queued.
  5. Form 4 archive build-up — first Form 4 filings post-2026-03-18 effectiveness; tracking via companies/tsem/data/form4_history.json once archive populates.
  6. Tax-treaty withholding rates by holder type — US-Israel tax treaty rates apply tiered withholding on dividend distributions; precise rate by holder cohort affects buy-side analysis of any future dividend authorization.

Sources

  • FY 2024 20-F (acc. 0001178913-25-001537, filed 2025-04-30) — SEC EDGAR ✓. Cover-page share count 111.76M; full dividend / repurchase text pending direct extraction.
  • FY 2023 20-F (acc. 0001178913-24-002347, filed 2024-04-22) — SEC EDGAR ✓. Pre-FY 2024 baseline.
  • Q4 2025 6-K release (filed 2026-02-11) — Tower release ✓ and GlobeNewswire ✓. FY 2025 cash + capital deployment.
  • Q4 2024 + FY 2024 6-K release (filed 2025-02-10) — GlobeNewswire ✓. FY 2024 cash baseline.
  • Intel termination announcement (2023-08-16) — Tower release ✓ and EE Times ✓. $353M termination-fee receipt detail.
  • Tower-Intel Fab 11X capacity-services arrangement (2023-09-05) — Tower release ✓. $300M Tower equipment investment — the primary deployment of the Intel termination fee proceeds.
  • HFIAA (Holding Foreign Issuers Accountable Act) — signed 2025-12-18, effective 2026-03-18; repealed FPI Section 16(a) Form 4 carve-out. Per the Tower KB index.
  • GFS comparable — GFS capital_returns ✓. Feb 2026 $500M authorization; structural template for Tower’s eventual first authorization.
  • VIS / DBHiTek / UMC / TSMC dividend practices — aggregator data, primary-source disclosure pending.
  • Aggregator share-count + insider data — companies/tsem/data/STOCK_PRICE_DATA.json. Insider 0.5%, institutional 65%.

Cross-references