Millicom International Cellular S.A. (NASDAQ: TIGO) — The Deleveraging Is Done; The Controller Isn’t

Independent fundamental-research note. The analytical body of this article carries no investment recommendation and no price target; the sole exception is the labeled “Claude’s Take” block below.

As of: 2026-07-11 · Price: ~$97 (2026-07-10 close $96.96) · Shares: ~167.7M · Market cap: ~$16.3B · Net financial debt: ~$5.36B · EV (net financial debt): ~$21.6B · FY2025 revenue: $5,819M · FY2025 Adjusted EBITDA: $2,783M (47.8% margin) · Reported FY2025 EPS: $7.86 (≈⅔ non-recurring) · Recurring EFCF: ~$864M

⚡ Claude’s Take

The author’s own independent opinion and general information only — not investment advice. The analytical body of this article below carries no position and no price target.

Verdict: HOLD / AVOID-chasing-here. A genuinely fixed, cash-generative LatAm telco — but the re-rating has already paid out the thesis, and the risk/reward at an all-time high is unattractive. I’d want the low-$60s–low-$70s (≈6x net-debt EV/EBITDA, ≈8% recurring-EFCF yield) before the reward justifies the EM-FX and governance tails. Not a short. Conviction: medium.

Millicom is a real turnaround, not a story. A distressed, over-levered, sprawling operator that nearly broke in 2022 (the SDR traded to ~$10–11) has, under Xavier Niel’s Atlas control and the “Project Everest” cost program, expanded Adjusted-EBITDA margin from ~38% to ~48%, inflected to a record ~$864M of recurring equity free cash flow, reinstated a covered $3.00 dividend, and is rolling up Telefónica’s distressed Latin-American exits (Ecuador at a bargain ~0.9x revenue; Uruguay and Colombia at fuller prices). The stock has re-rated roughly 8x, from $12.63 to $97. That is the good news, and it is fully in the price.

Here is why I won’t chase it. The multiple has done the heavy lifting: TIGO now trades at its richest-ever price/book (5.2x, 99.97th own-history percentile) and price/sales (2.5x, 99.97th percentile), ~7.8x net-debt EV/EBITDA (~9x including the ballooned leaseback leases) — the richest of the LatAm peer set (AMX ~8.7x, Liberty LatAm ~5.8x, Telefónica ~5.5x), i.e. priced like the quality leader while earning a ROIC of only ~10%, essentially equal to its cost of capital. That last fact is the whole problem: when ROIC ≈ WACC, the acquisition roll-up is value-neutral — growth doesn’t create value unless Niel’s cost-out genuinely lifts returns above WACC, which is unproven on the acquired assets. The reported P/E of ~13x is a mirage (two-thirds of 2025 GAAP EPS is the one-time tower-sale gain and a deferred-tax release); on normalized earnings it’s ~19x on cash (EFCF) and far higher on accounting earnings. The framing is a value-stock-turned-crowded-momentum-winner (+183% over the trailing year, at the highs, richest-ever on book and sales) — the kind of name that gives multiple points back on the first EBITDA disappointment.

And the controller is not the minority’s friend. Atlas/Niel has crept from ~44% to 49.5% in about three-and-a-half months, financed with bank equity-forwards, one share short of majority — and did it without paying minorities a control premium, having tendered at just $25.75 in 2024 (now ~73% below the tape). This is a de-facto controlled company run on the Iliad low-cost playbook; the eventual exit for minorities may well be a take-private timed and priced on Niel’s terms, not theirs.

What flips me bullish: hard evidence that post-synergy ROIC on the Colombia roll-up clears WACC and recurring EFCF sustains above ~$1.2B — that would validate a durable compounder and justify the AMX-like multiple. What flips me bearish (harder): a below-market take-private bid from Niel, or a 10%+ COP/BOB/PYG devaluation flowing straight through to USD EFCF. Tag: the balance sheet got fixed; the cap table didn’t.

📈 Stock Price Action — Five-Year Event Map

Millicom’s US-listed share has completed a violent round trip. From ~$38 in early 2020, through a COVID-era dividend suspension, LatAm FX stress and a dilutive 2022 rights issue, it fell to a near-distress low around $10–11 in October 2022, then re-rated roughly 8x to an all-time high of $96.96 (2026-07-10). It now trades essentially at its record, ~43% above its ~$68 200-day EMA, inside a 52-week range of roughly $37–$97 and ~0% off its high. It is a low-beta (0.34), high-alpha name whose entire recovery is a deleveraging/self-help story rather than a thematic melt-up. (Prices from a 5-year adjusted price series, adjusted for the 2022 dividend/rights and subsequent dividends, so historical prints differ modestly from reported closes. Price move = Fact; attributed driver = Interpretation. No price target, no recommendation.)

Cycle narrative. (1–2) The 2021–22 collapse was a genuine near-distress event — a suspended dividend, a $746M rights issue that diluted holders to fund the Guatemala buy-in, and EM FX/rate stress drove the stock to ~$10–11 by October 2022. (3) The bottom coincided with Xavier Niel’s Atlas opening a 6.99% position and the start of the cost-reset playbook; delivering FCF stabilization re-rated the shares ~50%. (4) In 2024, Atlas’s $24 tender (lifted to $25.75, ~$4.4B) — which the independent board rejected as too low — put both a floor and a control overhang under the stock as Atlas passed 40%. (5) Dividend reinstatement, ~$975M of tower monetization and the (later-blocked) Costa Rica JV kept shares range-bound near $21–24 into early 2025. (6) 2025 was the inflection: record ~$916M EFCF, all-time-high margins and fast Ecuador/Uruguay integration doubled the stock to ~$55. (7) Through 1H-2026 the Colombia (Coltel) and Chile deals closed and turned EFCF-positive faster than guided, management reiterated ≥$900M FY26 EFCF with a path back to ~2.5x financial leverage, and the shares ran another ~75% to a record ~$97.

1. Executive Summary

Millicom International Cellular (“Tigo”) is a leading fixed-plus-mobile operator across eleven Latin American markets, run as country-level segments and reporting in USD under IFRS. FY2025 revenue was $5,819M (+0.2% reported, +2.0% organic), service revenue $5,451M (+2.8% organic), and Group Adjusted EBITDA $2,783M at a 47.8% margin — a structural ~10-point margin gain versus 2020, driven by the Project Everest efficiency program rather than mix alone. The company is a genuine turnaround: from a near-distress balance sheet in 2022 to record equity free cash flow (~$864M recurring) and a reinstated $3.00 dividend in 2025.

The business is not homogeneous. Guatemala is the crown jewel — ~26% of revenue ($1,671M) but a 55.5% EBITDA margin and #1 positions in both mobile and broadband, with core spectrum secured to 2043. It is a genuine local-scale, customer-captivity moat. So, to a lesser degree, are Paraguay, Bolivia, Panama and Honduras, where Tigo is #1. Colombia, Ecuador and Uruguay are subscale #2 positions with no moat — the Telefónica roll-up is an attempt to buy the scale Tigo lacks organically. Group ROIC of ~10.7% is exactly this blend: crown-jewel returns diluted by capital-hungry subscale positions.

Three things dominate the investment debate. First, the earnings are cash-good but accounting-poor. Reported FY2025 EPS of $7.86 is roughly two-thirds non-recurring — a $741M tower-sale gain plus a ~$142M deferred-tax release — leaving recurring EPS around $2.60. The right lens is recurring equity free cash flow (~$864M, ~$5.15/share), on which the stock yields ~5.3%. Second, the deleveraging narrative is half-optical. Net financial-debt/EBITDA fell to ~1.9x, but including the tower-leaseback leases that management created, total-obligation leverage is ~2.85x; the tower “monetization” swapped owned capex for perpetual lease payments. Third — and decisively — this is a controlled company. Xavier Niel’s Atlas owns 49.5% and is creeping to majority via debt-funded equity forwards, having tendered at just $25.75 in 2024. The minority overhang is real.

On valuation, the re-rating has closed the gap to quality peers. At ~7.8x net-debt EV/EBITDA (~9x with leases) and richest-ever price/book and price/sales, TIGO is priced like América Móvil — the regional quality leader — while earning ROIC only at its cost of capital. When ROIC ≈ WACC, the acquisition-led growth is value-neutral; the thesis from here rests entirely on whether Niel’s cost discipline can lift returns above WACC and whether EFCF sustains above ~$1B. The company is materially de-risked and better-run than at any point in a decade. The price now assumes that outcome. This memo takes no position; the labeled Claude’s Take above does.

2. Business Overview

Millicom operates the “Tigo” and “Tigo Business” brands across eleven Latin American countries — Guatemala, Colombia, Bolivia, Paraguay, Panama, El Salvador, Honduras, Nicaragua, Costa Rica, and, since October 2025, Ecuador and Uruguay (with a 49% non-consolidated joint venture in Chile from February 2026). The company is Luxembourg-domiciled, reports in US dollars under IFRS, files as a foreign private issuer (Form 20-F annual, 6-K interim), and — since delisting its Swedish Depository Receipts from Nasdaq Stockholm in 2025 — trades primarily as NASDAQ: TIGO. It employs ~14,500 people (FACT: FY2025 20-F).

How it makes money. Tigo sells four bundles of service, mostly to consumers and increasingly to businesses and governments:

• Mobile (≈60% of service revenue, $3,286M in 2025, +4.0% organic). Voice, data and SMS across prepaid (~57% of mobile service revenue) and postpaid, plus mobile financial services. Blended mobile ARPU is ~$6.2/month — classic emerging-market prepaid economics — and the key value lever is migrating prepaid users to higher-ARPU postpaid (only ~29% of the base is postpaid today). Mobile customers total ~49.3M.
• Fixed / Home (≈25% of service revenue; Home service revenue $1,371M in 2025, −4.9%). Residential broadband, pay-TV (Tigo ONEtv) and fixed voice over ~200,000 km of HFC/FTTH network. ~4.6M fixed customer relationships (4.2M via HFC/FTTH) and ~8.0M revenue-generating units. Home ARPU is ~$24.7/month — four times mobile — but is under pressure from FX and competition (down from ~$27.1). Low regional broadband penetration is the structural growth runway.
• B2B / Tigo Business (~$700M-plus, growing double digits). Connectivity, managed services, cloud, cybersecurity and value-added services to SMEs, enterprises and governments, supported by twelve Tier-III data centers. Digital B2B (cyber/cloud) grew ~19–40% in recent quarters — the highest-quality organic growth line.
• Tigo Money / mobile financial services (~2.8M active users across Paraguay, Guatemala, El Salvador, Bolivia, Honduras). Payments, transfers, remittances, savings, micro-loans and micro-insurance. Notably, the Paraguay MFS operation was reclassified as held-for-sale in early 2026 as part of “portfolio optimization” — management is a seller, not a builder, of fintech.

Revenue by country (2025, pre-elimination): Guatemala $1,671M (26%) · Colombia $1,450M (22%) · Panama $725M (~11%) · Honduras $621M (JV at 100%) · Paraguay $578M (~9%) · Bolivia $356M (6%, −41.9% on the Boliviano devaluation post-IAS-21) · Other ~$1,066M. Adjusted EBITDA by country: Guatemala $928M @ 55.5% · Colombia $604M @ 41.7% · Panama $371M @ 51.2% · Honduras $320M @ 51.5% · Paraguay $297M @ 51.4% · Bolivia $174M @ 48.9% · Other $466M @ 43.7%. Six markets now run above a 50% EBITDA margin (“Club 50”) — a testament to the cost program.

Infrastructure. Tigo has deliberately shed passive infrastructure: its tower count fell from >9,000 to ~2,000 via sale-leasebacks (the “Lati” transactions), on the stated view that towers “no longer confer a competitive advantage.” It retains the active network (spectrum, RAN, ~200,000 km of fiber, twelve data centers). Revenue is overwhelmingly recurring (subscription mobile/fixed/B2B); equipment/handset sales (~$367M) are the main non-recurring line.

Verdict. A focused, cash-generative, converged LatAm telco with a genuinely superb core asset in Guatemala and a coherent “digital highways” strategy, offset by mature organic growth, heavy FX exposure, and a set of subscale second-place positions the company is now trying to fix through acquisition.

3. Industry Dynamics

Latin American telecom is a collection of national oligopolies, not one market. Each country has its own spectrum regime, regulator, currency and competitive structure, and Tigo’s economics vary enormously across them — which is why country-level analysis, not group averages, is the correct lens.

Structure and competition. The dominant regional force is América Móvil (Claro), the Carlos Slim–controlled elephant that is #1 or a strong #2 in most of the region and sets pricing in several of Tigo’s markets (notably Colombia). The second historical major, Telefónica (Movistar), is in structural retreat from Latin America — and is precisely the seller of the Ecuador, Uruguay and Colombia assets Tigo is now buying. Liberty Latin America, regional players such as WOM (Colombia), and state-linked incumbents (Antel in Uruguay, ETB in Colombia, CNT in Ecuador) round out the field. Markets range from constructive 2–3-player oligopolies (Guatemala, Paraguay, Bolivia, Panama) to a genuinely tough four-player fight (Colombia, where WOM’s entry and asymmetric regulation weigh on returns).

Economics. These are prepaid-heavy, high-churn markets (dual-SIM, number portability, eSIM), spectrum-intensive (recurring license renewals and auctions are a real capital cost), and — the defining EM feature — FX-exposed: revenue is earned in local currency (GTQ, COP, BOB, PYG, HNL) against a debt stack that, while ~52% localized at the operating companies, still carries ~$2.4B of USD holdco bonds. Bolivia in 2025 is the cautionary tale: the adoption of IAS-21 hyperinflation accounting alongside a Boliviano devaluation cut reported Bolivia revenue ~42% and EBITDA ~35% in a single year. Low fixed-broadband penetration is the offsetting structural tailwind — the region is under-penetrated versus developed markets, giving converged operators a multi-year fixed-line runway.

Capital-cycle read (Marathon lens, ). The supply side is inflecting favorably in Tigo’s core. Telefónica’s regional exit withdraws a competitor and consolidates several markets toward 2–3 players; industry tower capacity is being taken out of the competitive equation via sale-leasebacks; and disciplined new ownership has repaired balance sheets and lifted margins across the group. This is the classic Marathon setup — capital leaving a beaten-up industry, a disciplined consolidator buying assets cheaply from a distressed seller. The caution is that the demand side (organic service revenue) is only GDP-like, and the FX overlay can erase local-currency gains in USD terms.

Verdict: mixed-favorable and improving. Structurally good in the concentrated small markets — Guatemala, Paraguay, Bolivia, Panama, Honduras — where oligopoly discipline and scale produce 50%-plus EBITDA margins. Structurally tougher in Colombia, a subscale four-player market where Tigo sits behind Claro and faces regulatory-asymmetry risk as its share rises through the Coltel consolidation. The industry is better than it was, but it is not a structurally great business the way, say, US wireless is; it is a decent business made attractive by a distressed-consolidation moment and a self-help operator.

4. Competitive Position

The honest answer on Tigo’s moat is: real where it is #1, absent where it is #2. In Bruce Greenwald’s taxonomy, Tigo’s advantage — where it exists — is a combination of local economies of scale, customer captivity (prepaid habit, convergence bundling, distribution density) and spectrum. These are genuine barriers to entry only in the markets where Tigo has the leading share and can spread network and spectrum cost over the largest local subscriber base.

Where the moat is real:

• Guatemala (the crown jewel). #1 in mobile (~11.7M subs) and broadband (~698K), a stable dual-play leader in a small, concentrated market, spectrum secured to 2043, and a 55.5% EBITDA margin — the clearest case of durable local scale plus customer captivity in the portfolio. A small market with a dominant incumbent is exactly where scale economies bite hardest; a challenger cannot replicate the network economics.
• Paraguay (#1 mobile 4.4M, #1 broadband 502K, 51.4% margin), Bolivia (#1 broadband 680K, 48.9% margin), Panama (#1 mobile 2.9M, 51.2% margin), Honduras (#1 mobile 5.0M JV, 51.5% margin). Each is a leading position in a concentrated market producing 49–51% margins — real, if smaller, versions of the same moat.

Where there is no moat:

• Colombia (#2 mobile ~13.2M behind Claro; #2 broadband; 41.7% margin), Ecuador (#2 mobile 5.1M), Uruguay (#2 mobile 1.1M). These are subscale challenger positions with structurally lower margins and no defensible scale edge. The Coltel/UNE consolidation is an explicit attempt to buy the scale Tigo cannot generate organically — merging two subscale Colombian operations to approach the scale needed to earn its cost of capital.

Market-share stability is the Greenwald test for a moat, and it separates the two groups cleanly: Tigo’s leading positions (Guatemala especially) have been durable for years, while its challenger positions have not converged toward leadership on their own. Direct comparison versus Claro (América Móvil) is instructive — where Tigo is #1 (Guatemala, Paraguay) it out-earns Claro locally; where Claro is #1 (Colombia) Tigo earns challenger economics. The towers-are-not-a-moat divestiture is consistent with a management team that understands its real advantage is active-network scale and customer relationships, not passive steel.

Verdict: a genuine local-scale moat in Guatemala, Paraguay, Bolivia, Panama and Honduras; none in Colombia, Ecuador or Uruguay. The ~10.7% group ROIC is the arithmetic of that blend — crown-jewel returns (Guatemala standalone almost certainly earns well above WACC) diluted by capital-hungry subscale positions and a debt-funded acquisition spree. The moat is real but partial, and the acquisitions dilute rather than deepen it until synergies prove out.

5. Growth History and Forward Opportunities

History: a step-up masking organic stagnation. Reported revenue went $4.26B (2021) → $5.62B (2022) → $5.66B (2023) → $5.80B (2024) → $5.82B (2025). The one large jump — 2021 to 2022 — was not organic: it was the move to 100% full consolidation of Guatemala (from a 55%-owned JV) in November 2021. Strip that, and the top line has been essentially flat for four years. Organic service-revenue growth has run just +1.9% (2024) and +2.8% (2025) — GDP-like in USD, somewhat higher in local currency, and repeatedly masked by Bolivia’s FX collapse. The genuine organic bright spot is Colombia (+5.1%) and digital B2B (+19–40%); the drag is Home service revenue (−4.9%) on ARPU pressure.

So the real “growth” story is not organic — it is margin, cash flow, and consolidation:

1. The Telefónica roll-up. Ecuador (closed Oct-2025), Uruguay (closed Oct-2025) and Colombia/Coltel (control Feb-2026, 100% by ~April 2026) add ~$2.2B of acquired revenue bought cheaply from a distressed seller. Q1-2026 shows it landing: revenue $1,985M (+45.1% reported, +4.2% organic), service revenue +4.9% organic, Adjusted EBITDA $857M, mobile ARPU +13.4%. Pro-forma group run-rate revenue is now ~$7.5–8B.
2. Fixed-broadband / FTTH penetration in under-penetrated markets — the multi-year structural runway.
3. Prepaid-to-postpaid migration and ARPU — with only ~29% of the mobile base on postpaid, the mix-up lever is substantial and high-incremental-margin.
4. B2B / VAS / data centers — the fastest organic line.
5. Further tower / infrastructure monetization and portfolio pruning (Tigo Money divestiture).

Forward drivers, and the catch. The acquired assets are expected to converge toward the group’s ~15% EFCF/revenue run-rate, layering meaningful EFCF on top of the base as margins on the Telefónica operations move from ~30% to 40%-plus (management says Ecuador/Uruguay are already there). Coltel, initially modeled EFCF-neutral-to-negative, is now guided as a net EFCF contributor in 2026, with >$100M of year-one OpEx savings identified and the combined Colombia growing top-line ~8%. That is the bull mechanism. The catch is that acquisition-led growth at a business earning ROIC ≈ WACC creates value only if the cost-out genuinely lifts returns above the cost of capital — otherwise it is growth for growth’s sake, financed with debt.

Verdict: low-to-medium-quality growth. Headline growth is M&A; organic growth is mature-EM-telecom modest and FX-fragile. The quality of the forward story depends entirely on integration execution and on whether Niel’s cost playbook can drag the subscale acquired positions to scale economics — not on any organic secular tailwind.

6. Financial Quality

Revenue and margins. ~$5.82B revenue, flat organically but with a genuinely impressive, cost-driven margin transformation: Adjusted-EBITDA margin rose from ~38% (2020) to 47.8% (2025), and EBITDA in absolute terms nearly doubled from $1.46B to $2.78B. Crucially, this is structural, not mix — the Project Everest program (“we review every purchase order from $1,” in management’s phrasing) cut cost across procurement, headcount and network, lifting six markets above a 50% margin. This is the single most credible piece of the bull case and the clearest evidence that the Iliad/Atlas low-cost operating culture is real.

The quality-of-earnings problem. Reported FY2025 net income to owners was $1,316M and basic EPS $7.86 — but pretax profit ($1,665M) exceeds operating profit ($1,639M) despite $674M of net interest, which is only possible because of large non-operating gains. Normalizing:

Only about one-third of the $7.86 GAAP EPS is recurring. The cash-flow statement corroborates: the gains are explicitly reversed in operating cash flow (the −$1,088M “other non-cash adjustment”). Consequently the reported ~13x P/E dramatically understates the accounting-earnings multiple (~37x on ~$2.60), and the honest earnings lens for a telco of this kind is cash, not GAAP EPS.

Cash flow and dividend coverage. The right metric is management’s Equity Free Cash Flow (EFCF), a record $916M in 2025 ($864M excluding divestiture proceeds). The bridge: OCF $1,734M − cash capex $814M − lease principal $209M − net interest $558M + asset-sale proceeds $84M + JV repatriation $123M ≈ $916M. Two cautions: (i) cash capex is only ~14% of sales versus accrual capital additions of $1,226M (21%) — a gap that flatters near-term EFCF; and (ii) the tower monetization converted owned capex into perpetual lease payments (lease liabilities jumped from $954M to $2,587M), which will pressure forward EFCF. On coverage: 2025 declared shareholder remuneration was $6.25/share ($0.75 interim + $3.00 ordinary + a $2.50 special explicitly funded by tower proceeds). The recurring $3.00 base (~$503M) is comfortably covered by recurring EFCF-ex-divestitures ($864M) at ~58% payout; the special is a one-time return of monetization capital, not a run-rate.

Returns. ROIC is ~10.0–10.7% (clean EBIT ~$1,521M, NOPAT ~$1,141M on ~$11.6B invested capital) — at or just below an EM-telecom WACC of ~10–12%. Levered ROE is optically high (40.6%) but reflects the one-time gains and a thin, intangible-heavy equity base. Tangible book is ~−$4.2B (equity $3,640M less goodwill+intangibles $7,798M), so there is no asset-value floor. SBC is trivial (~$14M), so dilution is not a concern; share count is stable-to-shrinking (~167.7M).

Verdict: cash-good, accounting-poor; economics improve modestly with scale but returns remain marginal. This is a real cash generator with genuine, structural cost-driven margin expansion — but GAAP earnings are two-thirds one-time, ROIC only matches the cost of capital, tangible equity is deeply negative, debt costs 6.67%, and LatAm FX is a recurring drag. Quality of earnings is low on GAAP, moderate on cash.

7. Capital Allocation

Capital allocation under CEO Marcelo Benitez and Atlas/Niel control is mixed-improving — sharper and more disciplined than the prior regime, but with financial-engineering flourishes and a governance overhang that dominate the picture.

The good. (1) Financial deleveraging — net financial-debt/EBITDA fell from ~2.8x (2023) toward ~1.9x, restoring the balance sheet from near-distress. (2) A covered, reinstated base dividend — the $3.00 ordinary is ~58% of recurring EFCF, appropriately conservative, with management stating a cut is “not on the radar at all” and AGM authority obtained for incremental dividends and up to a 10% buyback once leverage is sustainably below 2.5x. (3) A genuinely cheap acquisition — Ecuador (Otecel) at EV $380M on ~$404M of revenue (~0.9x revenue, ~zero goodwill), a bargain from a distressed seller. (4) Real cost discipline flowing to margin (Project Everest).

The questionable. (1) The tower “monetization” is financial engineering — ~$975M of cash now in exchange for perpetual leaseback obligations (leases up ~$1.6B), which improves the headline financial-leverage optic while raising true total-obligation leverage to ~2.85x including leases and increasing forward lease-cash cost. (2) The special dividend returns one-time capital dressed as remuneration. (3) The acquisition prices are mixed — Uruguay at ~2.6x revenue (full) and Colombia funded entirely with debt (driving the leverage-up) — and, at ROIC ≈ WACC, the roll-up is value-neutral until synergies prove out. (4) A legacy of value destruction (the African exit, prior write-downs, and the Guatemala corruption matter that produced the 2025 DOJ deferred-prosecution agreement: a $60M fine plus $58.2M forfeiture).

The governance overhang. The defining capital-markets fact is the controlling shareholder. Atlas Investissement (Xavier Niel / Iliad Holding / Maya) has crept from ~44.1% (Mar-2026) to 49.5% (Jun-2026) — one share short of majority — financed with BNP Paribas prepaid-forward/equity-swap agreements (6.0M shares to settle by 29-Sep-2026), i.e. buying to control in the open market without paying minorities a control premium. Four of eight directors are Niel-affiliated; Marcelo Claure is Chairman; the 20-F states Niel can control a merger, an asset sale, or amendments to the Articles. Atlas’s 2024 mandatory tender was priced at $25.75, now ~73% below the tape — evidence of an acquirer that consolidates cheaply. A Luxembourg statutory squeeze-out needs 95% (far off), so a forced squeeze-out is not near-term, but once past 50% Niel can run the company for the controller’s benefit and time an eventual take-private on his terms.

Insider activity outside Atlas is neutral: FY2025–26 Form 4s are routine director grants and tax-withholding dispositions (~$73.92/share, Jul-2026); the 144s are small planned sales. There is no management open-market conviction buying — the only buyer that matters is the controller.

Verdict: mixed-improving on operations and the balance sheet, but governance risk for minorities is the defining overhang. Management has allocated capital more intelligently than its predecessors — deleveraging, a covered dividend, a cheap Ecuador deal, real cost-out — but the tower engineering, the debt-funded roll-up at marginal returns, and above all a controlling shareholder creeping to majority without a minority premium temper the verdict.

8. Changes and Headwinds — Last Two Years

The environment has changed favorably and materially — this is a fundamentally different, de-risked company than the 2022 distressed name. The key developments:

• Aug 2024 — Atlas tender at $24 → $25.75/share (~$4.4B). The independent committee called it undervaluing; Atlas passed 40%. Established both a floor and the control overhang.
• Aug 2024 — Costa Rica JV with Liberty Latin America announced (later blocked by the regulator Sutel in Oct-2025; Millicom appealed).
• Late 2024 — dividend reinstated after the 2020–21 suspension; $1.00 interim (paid Jan-2025), then the $3.00 policy; SDRs delisted from Nasdaq Stockholm, US listing primary.
• Oct-2024 → Q3-2025 — ~$975M “Lati” tower monetization (El Salvador/Honduras sale-leasebacks).
• Q3-2025 (Nov) — DOJ deferred-prosecution agreement on the legacy Guatemala (Comcel) corruption matter: $118M provision (~$180M total cash in Q4); Telefónica/Costa Rica 2020 litigation settled; Ecuador and Uruguay acquisitions closed (footprint to 11 countries).
• Feb 5–10, 2026 — Coltel (Colombia) control acquired from Telefónica; Chile JV with NJJ (Niel) closed (footprint to 12); 100% of Colombia by ~late April after the UNE/La Nación stake closed.
• Feb 26, 2026 — FY25 results beat: record $916M EFCF, all-time-high margins; FY26 guidance set at ≥$900M EFCF and ~2.5x leverage.
• May–Jun 2026 — Q1-26 record Q1 EFCF ($225M); Atlas/Niel crossed to 46.7% then 49.5%.
• Jul 8, 2026 — $0.75/share quarterly dividend paid.

Headwinds and open frictions: Bolivia’s FX collapse (revenue −42%); the Costa Rica JV block; the DOJ legal overhang (now resolved via DPA but a reputational/compliance marker); Colombia integration and antitrust risk as Tigo’s share rises; a rising, if laddered, debt-maturity profile; and the ever-present controlling-shareholder question. On balance, the sentiment skew across the last 18 months is decisively positive — EFCF beats, faster-than-guided integrations, legal-overhang removals — which is exactly why the stock is at an all-time high. Verdict: the changes strengthen the operating thesis and de-risk the balance sheet, but they also mean the easy repair is done and the price now reflects it.

9. Risk Analysis

The dominant risks for a USD-based holder are currency/country and governance, not operations — the ITUB/MELI EM precedent (you can be right on the business and lose money in USD if the local currency slides) applies squarely here.

Catastrophic-loss / total-loss assessment. A total loss is remote — the company is FCF-positive, deleveraged on financial debt, and has hard assets and durable market positions (Guatemala above all). The realistic tail is not bankruptcy but permanent capital impairment for a USD holder via (a) a sharp regional FX devaluation compounding the already-full multiple, and/or (b) a controller-driven take-private below the trading price. Both are plausible, not remote.

10. Valuation

Embedded-expectations and scenario analysis only. No price target, no recommendation.

Where the multiple sits. At ~$97: market cap ~$16.3B (167.7M shares), net financial debt ~$5.36B → EV ~$21.6B, i.e. 7.8x FY25 EBITDA ($2.78B), ~7.2x TTM ($3.0B) — but ~9.0x including IFRS-16 leases (gross debt ~$11.7B, net ~$10.5B → EV ~$26.8B). The 2022 trough was 3.8x; the multiple has roughly doubled, which — together with financial deleveraging — drove the ~8x equity move from $12.63 to $97. On own-history percentiles, P/B 5.2x and P/S 2.5x both sit at the 99.97th percentile (richest ever); the reported P/E of ~13x looks moderate only because GAAP EPS is one-time-inflated (on normalized EPS ~$2.60 the accounting P/E is ~37x; on recurring EFCF/share ~$5.15 the cash multiple is ~19x, a ~5.3% yield). Tangible book is negative, so there is no book-value floor.

Peer context (TTM EV/EBITDA): América Móvil ~8.65x · TIGO ~7.2–9.0x · Liberty Latin America ~5.80x · Telefónica ~5.47x. Tigo now trades richest or tied-richest of the LatAm group — a large premium to Liberty and to Telefónica (the seller of the assets Tigo is rolling up), roughly in line with América Móvil, the regional quality leader. Against US deleveraging-telco benchmarks the picture is similar: TIGO’s ~5.3% recurring-EFCF yield is tighter than VZ (~10%), T (~10%) and dramatically tighter than CHTR (>25%) — the re-rating has taken TIGO to a premium FCF valuation versus levered developed-market peers, despite carrying EM-FX risk they do not.

Embedded expectations. At ~7.7x net-debt EV/EBITDA and a ~5.3% recurring-EFCF yield, the market is pricing Tigo as an América Móvil-quality durable compounder — deleveraging complete, FCF inflection real, Niel cost discipline credible, mid-single-digit EFCF growth — not as a distressed EM telco. The difficulty: the “easy” money (multiple normalization plus debt paydown) is spent. Further upside requires EBITDA growth at a multiple that no longer expands — or a re-rate above América Móvil — for a business earning ROIC ≈ WACC with materially higher FX risk. The Marathon capital-cycle read cuts both ways: Telefónica’s exit is a bullish supply-side signal (a disciplined consolidator buying cheap), but because ROIC ≈ WACC the roll-up is value-neutral unless cost-out lifts returns above the cost of capital — the single most important thing left to prove.

Scenarios (illustrative; explicit assumptions; NOT targets):

The base case says the stock is roughly fairly valued for what is now known; the asymmetry is unattractive because the bear (FX + derate) is a live, high-likelihood path while the bull requires proving ROIC > WACC on the acquisitions — not yet in evidence. What the market is underwriting correctly: the deleveraging, the FCF inflection, and the credibility of the cost program. What it may be underwriting incorrectly: that a ROIC ≈ WACC business deserves a quality-leader multiple; that EM-FX is benign; and that the controlling shareholder’s interests align with minorities’.

11. Variant Perception

Consensus. A deleveraged, FCF-compounding Latin American telco under a proven low-cost operator (Niel/Iliad), re-rated fairly from distress — a “quality compounder emerging from a special situation.”

The strongest bull case. EFCF inflects toward $1.5B-plus as the Telefónica roll-up (bought cheaply from a distressed seller) converges to group cash-conversion; Guatemala remains a 55%-margin cash engine; the dividend grows and buybacks begin as leverage clears 2.5x; and Niel’s cost playbook — already proven in the 38%→48% margin move — lifts the acquired assets’ returns above WACC, driving a further re-rate toward or above América Móvil. In this view, the value name that “started working” keeps working, and the momentum is fundamentally earned.

The strongest bear case. Richest-ever price/book and price/sales leave zero valuation cushion; ROIC ≈ WACC, so the consolidation is value-neutral; FX/EM tail risk is asymmetric to a USD holder (Bolivia already showed a 42% revenue hit); integration risk on the subscale acquired positions is real; and — decisively — this is a controlled company creeping to majority with a demonstrated $25.75 low-ball instinct, so the minority’s eventual exit may be dictated, not negotiated. The deleveraging/re-rating money is already made; what remains is EM-telco operating risk at a developed-market FCF multiple.

Factor read. A quantitative factor model shows Tigo carrying both Value (β 0.42) and Momentum (β 0.31) loadings, beta 0.34, +183% trailing-year return — a value name that has become a crowded momentum winner, clustering with international dividend-value ETFs. That crowding is itself a derating risk: momentum names surrender multiple points fast on the first EBITDA miss, and the −75% five-year max drawdown in this very name is a reminder that the “low-vol/low-beta” label is regime-dependent and would break in an EM-FX shock. The factor tape supports the bull today but does not neutralize the tail.

The 3–5 assumptions that matter, and their falsification tests:

1. Deleveraging converts to durable FCF. Falsified if recurring EFCF fails to hold ≥$1.0–1.2B as lease and interest cash costs and FX bite.
2. The roll-up creates value. Falsified if post-synergy ROIC on the Colombia/Ecuador/Uruguay assets stays ≤ WACC (~10–12%).
3. FX is manageable. Falsified by a 10%+ COP/BOB/PYG depreciation flowing straight through to USD EBITDA/EFCF.
4. Minorities are treated fairly. Falsified by a below-market take-private or related-party value leakage as Atlas crosses 50%.
5. The multiple holds. Falsified by a derate toward Liberty LatAm/Telefónica (5.5–6x) on any EBITDA miss.

The variant-perception tension the whole memo resolves to: done-deleveraging plus a credible operator versus richest-ever book/sales plus ROIC ≈ WACC plus a controlling shareholder whose interests may not be the minority’s.

12. Fact vs. Interpretation

13. Open Questions

1. How much of the $118M DOJ charge hit 2025 P&L versus being pre-provisioned in 2024? This swings normalized EPS by ~$0.70 (2024 “other non-operating” carried $119M of adverse-legal provisions).
2. What is the precise USD-versus-local-currency split of debt, and the hedge book? ~52% opco debt is local-currency and ~$2.4B holdco debt is USD, but the residual unhedged USD exposure against local revenue is the key FX sensitivity — not fully disclosed.
3. Does Atlas crossing 50% at the September 2026 derivative settlement re-trigger any Luxembourg mandatory-offer/creeper obligation — and at what reference price?
4. What is Guatemala’s standalone ROIC and implied SOTP value? Not disclosed; the crown-jewel economics versus the ~10.7% group blend are central to any sum-of-the-parts.
5. What are the actual post-synergy return economics of the Colombia (Coltel/UNE) consolidation — will combined Colombia clear WACC, or remain a subscale value-neutral position?
6. What is the scope and consolidation treatment of the new Chile “associate” that appeared in Q1-2026 disclosures?
7. Is Tigo Money’s held-for-sale designation a value-crystallization or a write-down? Terms and buyer unknown.

14. What Must Be True

For the bull case to be right (compounder thesis):

• Recurring EFCF sustains and grows above ~$1.0–1.2B despite rising lease/interest cash costs and FX. Falsification test: two consecutive years of recurring EFCF below $1.0B.
• The Telefónica roll-up earns post-synergy ROIC above WACC — i.e., cost-out on the subscale acquired positions is real and durable. Falsification test: combined-Colombia ROIC remains ≤ WACC through 2027 despite claimed synergies.
• FX is a net-neutral-to-tailwind over the cycle, not a persistent drag. Falsification test: a >10% adverse move in a blended local-currency basket that cuts USD EBITDA guidance.
• The controller returns capital to all shareholders (dividends, buybacks) rather than extracting value or timing a cheap take-private. Falsification test: a take-private proposal below the prevailing trading price, or a related-party transaction on off-market terms.

For the bear case to be right (full-price, governed-for-the-controller thesis):

• The multiple derates from richest-ever levels toward the LatAm peer average (5.5–6x EV/EBITDA) on any EBITDA disappointment or FX shock. Falsification test: the stock sustains >7.5x net-debt EV/EBITDA through an EBITDA miss.
• ROIC stays at/below WACC, confirming the acquisition-led growth destroys or merely preserves value. Falsification test: group ROIC durably exceeds ~12%.
• The controlling shareholder’s interests diverge from minorities’ at the point of exit. Falsification test: a full, premium cash bid to minorities, or a binding minority-protection framework.

The two cases are separated by three testable variables over the next 4–8 quarters: recurring EFCF trajectory, post-synergy Colombia ROIC, and controller behavior. Everything else is secondary.

15. Source Appendix

Primary sources: Millicom FY2025 Form 20-F (filed 2026-03-24, for FY ended 2025-12-31) and prior-year 20-Fs (2021–2024); FY2025 and Q1-2026 earnings releases and results 6-Ks (2026-02-26, 2026-05-14); SEC SCHEDULE 13D/A filings by Atlas Investissement (Amendments 26–29, Mar–Jun 2026); Form 3/4/144 insider filings (2025–2026); aggregated fundamentals data (income statement, balance sheet, cash flow, ratios, enterprise value, valuation multiples, per-share data); own-history valuation percentiles; a quantitative factor model (loadings, leaderboard, risk-adjusted track record, factor-similar peers); a 5-year adjusted price series; and public earnings-call transcripts (Q3-2025, Q4-2025, Q1-2026). Peer context drawn from public filings of América Móvil, Verizon, AT&T, Charter and Liberty Latin America. Full per-claim source detail is in the accompanying Source Appendix (Appendix B). Management commentary is treated as hypothesis and validated against filings, financials and external data throughout.

This article contains no investment recommendation and no price target. The only position expressed is in the clearly-labeled “Claude’s Take” block at the top, which is the author’s own independent view and general information, not investment advice.

APPENDIX A — Standard Diligence Questionnaire

Millicom International Cellular S.A. (NASDAQ: TIGO) — as of 2026-07-11

Supplemental to the article. Fact/Interpretation/Assumption labels applied where material.

General

What thoughtful questions have other investors asked about this company? The central debates: (1) Is the ~48% EBITDA margin sustainable, or a cost-cut peak? (2) How much of 2025 GAAP EPS is real (answer: ~⅓)? (3) Will Xavier Niel take the company private cheaply, and where does that leave minorities? (4) Can the Telefónica roll-up (Colombia especially) earn above its cost of capital? (5) How exposed is USD EFCF to LatAm currency devaluation? (6) Is the “1.9x leverage” real, or optics that ignore the tower-leaseback leases (true ~2.85x)?

Cyclicality & Earnings Nature

Cyclical high or low? Interpretation: Operating earnings are near a self-help high — margins at all-time records after Project Everest — but not a demand-cycle peak (organic revenue is only GDP-like). GAAP net income is at an artificial high on one-time tower gains. Driven by external or internal actions? Overwhelmingly internal (cost program, deleveraging, portfolio consolidation) plus an FX tailwind in 2025–26. Revenue stability? High — recurring subscription mobile/fixed/B2B; ~57% prepaid mobile is stickier than it looks (habit/distribution). Market outlook? Growing (LatAm broadband under-penetration, prepaid-to-postpaid migration) but FX-fragile in USD; largely domestic to each Latin American market.

Business Quality & Competitive Moat

Industry more or less competitive? Less, in the core — Telefónica’s regional exit consolidates markets toward 2–3 players; more, in Colombia (four-player). Profitability (ROIC/ROE)? ROIC ~10% (≈ WACC); ROE 40.6% but flattered by one-time gains and a thin equity base. Industry profitability / barriers? Concentrated small markets earn 50%+ EBITDA margins behind spectrum, scale and distribution barriers; subscale positions do not. Easily understood? Yes — a converged telco. Undermined by foreign low-cost labor? No (local infrastructure/regulated). Do brands matter? Moderately — “Tigo” has regional recognition, but telecom is largely a price/coverage business. Switching costs? Low in prepaid mobile (portability, dual-SIM, eSIM); higher in fixed/convergence bundles. Moat verdict: genuine local-scale moat in Guatemala/Paraguay/Bolivia/Panama/Honduras; none in Colombia/Ecuador/Uruguay.

Financial Condition & Balance Sheet

Assets not on the balance sheet? Guatemala’s crown-jewel franchise value is not separately marked; spectrum is carried at cost. Off-balance-sheet liabilities? The material item is now on balance sheet — IFRS-16 lease liabilities ballooned to ~$2.6B via tower leasebacks; note also non-recourse opco debt structure. Accounting conservatism? Interpretation: Mixed — IFRS with heavy use of “Adjusted” metrics; GAAP earnings inflated by one-time gains; DTA recognition is a judgment call. CapEx-hungry? Yes — cash capex ~14% of sales, accrual additions ~21%; spectrum renewals recurring.

Capital Allocation & Management

FCF generation and use? Record recurring EFCF ~$864M, used for a covered $3.00 dividend (~58% payout), financial deleveraging, and debt-funded acquisitions; a $2.50 special returned tower proceeds. Recent acquisitions? Yes — Ecuador (~0.9x revenue, cheap), Uruguay (~2.6x, full), Colombia/Coltel (debt-funded), Chile JV (49%, non-consolidated). Buybacks? Authorized up to 10% once leverage <2.5x; not yet material. Issuing shares to insiders? No — SBC trivial (~$14M); share count stable/shrinking. Compensation / director motivations? Interpretation: Controlled by Atlas/Niel (49.5%), 4/8 directors Niel-affiliated, Claure chairman — governance is oriented to the controller. Management motivation: the Iliad low-cost-operator playbook; aligned with minorities on dividends/delevering, potentially misaligned on an eventual take-private price.

Valuation & Market Data

ADR/MLP/K-1? Common shares of a Luxembourg company listed on NASDAQ (foreign private issuer, files 20-F). Not an MLP; no K-1. Dividend policy? $3.00/share ordinary (4× $0.75 quarterly), ~⅔ EFCF target payout; incremental dividends/buyback once leverage sustainably <2.5x. Profitability? Cash-good (EFCF), accounting-poor (GAAP EPS ~⅔ one-time). Net income vs cash from operations diverging? Yes — 2025 GAAP net income is inflated by ~$0.9B of non-cash gains that are reversed out of OCF; cash is the truer signal.

Risks & Downside

What would cause the stock to decline? LatAm FX devaluation (COP/BOB/PYG) flowing to USD EFCF; an EBITDA miss triggering a derate from richest-ever multiples; a below-market take-private; Colombia integration/antitrust problems; a leverage/refinancing scare. Catastrophic-loss risk? Low — FCF-positive, deleveraged on financial debt, hard assets and durable Guatemala franchise. Total-loss risk? Remote. The realistic tail is permanent USD capital impairment via FX plus multiple compression, or a controller-driven cheap exit — not insolvency.

Recent News & Events

Has the environment changed recently? Yes, materially and favorably — Ecuador/Uruguay/Colombia/Chile deals closed; record EFCF; dividend reinstated and growing; DOJ and Telefónica legal overhangs resolved; Atlas/Niel crept to 49.5%. Significant acquisitions? The Telefónica LatAm roll-up (2025–2026). Accounting-policy changes? IAS-21 hyperinflation adoption for Bolivia (Jan-2025) materially distorted Bolivia and group trends. Other recent changes? Two new markets (Ecuador, Uruguay) plus a Chile JV; Tigo Money reclassified held-for-sale; Nasdaq Stockholm SDR delisting.

APPENDIX B — Source Appendix

Millicom International Cellular S.A. (NASDAQ: TIGO) — as of 2026-07-11

Primary sources prioritized over secondary; management commentary treated as hypothesis and validated against filings/financials/external data. Facts distinguished from interpretation throughout the article.

Primary — SEC / Regulatory Filings (EDGAR CIK 0000912958)

• Form 20-F, FY2025 (filed 2026-03-24, for FY ended 2025-12-31) — Item 4 (Business), Item 5 (Operating & financial review), Item 6 (Directors/management/comp), Item 7 (Major shareholders / related-party), audited IFRS financial statements (F-pages), notes on one-time items (Lati tower sale A.1.3), tax/DTA (B.6), leases, debt maturities, segment/country data. Mirrored locally at output/TIGO/sources/20-F/2026-03-24_tigo-20251231.htm.
• Prior Form 20-Fs FY2021–FY2024 (filed 2022-03-01, 2023-02-28, 2024-03-12, 2025-04-08) — historical trend, Guatemala full-consolidation (Nov-2021), Africa-exit legacy.
• Results 6-Ks — FY2025 results (2026-02-26); Q1-2026 results (2026-05-14); interim press releases through 2026-05-21. 124 6-Ks mirrored at output/TIGO/sources/6-K/.
• SCHEDULE 13D/A — Atlas Investissement SAS — Amendment 26 (2026-03-13, 44.1%), Amd 27 (2026-03-27, 46.0%), Amd 28 (2026-06-23, 46.7%), Amd 29 (2026-06-29, 49.5% incl. 6.0M via BNP prepaid-forward/equity-swap, FBF Master Agreement 2025-11-12, EX-99.64). Ownership chain Atlas ← Iliad Holding ← Maya ← Niel family.
• Form 3/4/144 insider filings (2025–2026) — routine director grants + tax-withholding dispositions (~$73.92/sh, Jul-2026); small planned 144 sales. No management open-market conviction buying.

Primary — Earnings-Call Transcripts

• Q1-2026 (2026-05-12), Q4-2025 (2026-02-26), Q3-2025 (2025-11-06) — FY2026 guidance (EFCF ≥$900M; net leverage ~2.5x by YE26; capex ~$1B); Coltel/Chile integration; dividend/buyback policy; Project Everest cost program; Guatemala performance; Tigo Money held-for-sale. Treated as management hypothesis.

Quantitative Data

• Aggregated fundamentals data — income statement, balance sheet, cash flow, profitability/credit ratios, enterprise value, valuation multiples, per-share data (FY2020–FY2025). Enterprise value ~$17.2B at year-end $55; recomputed to ~$21.6B at $97.
• Own-history valuation percentiles (2026-07-10) — own-history percentiles: P/E 13.2x (39.7th), P/B 5.18x (99.97th), P/S 2.53x (99.97th), composite 79.9th; latest price $96.96, TTM EPS $7.34, BVPS $18.73.
• 5-year adjusted price series — adjusted OHLCV, 21/50/200-EMA (~$68 200-EMA), beta ~0.34, alpha ~0.87, dividend/split columns; $0.75 dividend 2026-07-08.
• Quantitative factor model — stock-loadings (Value β 0.42, Momentum β 0.31, Base+Sector, R²~0.20), leaderboard (y1 +183% Sharpe 4.84; y3 +95%/yr; y5 +23%/yr, −75% max drawdown; m6 +273% ann.), stock-info (beta 0.34), related-stocks (IDV, DTH, EWP, SCHY — international dividend-value cluster).

Secondary / Corroborating

• Company website and IR (millicom.com); GlobeNewswire/StockTitan press releases (2022 rights-offering completion; 13D/A 49.5%; shareholder-remuneration announcements); telecoms.com (Atlas $4B/$25.75 bid coverage). Used only for dates/corroboration; primary filings govern.

Peer Context

• Public financials and disclosures of América Móvil (AMX), Liberty Latin America (LILA/LILAK), Telefónica (TEF), and developed-market deleveraging telecoms (Verizon, AT&T, Charter) for EV/EBITDA, FCF-yield and leverage benchmarks.

Notes on Data Limitations

• Foreign private issuer: no 10-Q/8-K; interim data from 6-Ks/transcripts.
• Real-time news coverage is sparse for a foreign private issuer — the recent-events timeline was built from transcripts, company/SEC releases and trade press.
• GAAP EPS is ~⅔ non-recurring (tower gain + DTA); recurring EPS/EFCF derived by normalization (Interpretation), not a reported figure.