Far EasTone: Taiwan's Telecom Consolidator Bets on "Essential Services"

I. Introduction & Episode Roadmap

On the morning of August 6, 2026, Far EasTone's finance team walked analysts through a quarter that had almost no bad news in it. Consolidated revenue for the second quarter came in at NT$28.66 billion, up 13.6% year over year. EBITDA hit a quarterly record of NT$10.15 billion. Net income of NT$4.01 billion beat the company's own guidance by 18%, and earnings per share of NT$1.11 blew past the NT$0.94 the company had told the market to expect.¹

Records are not, by themselves, interesting. What made this one worth pausing on is the shape of the business underneath it. This is a mobile phone company in a market where almost every adult already has a mobile phone. Management has repeatedly told analysts the telecom market is saturated. Taiwan's population is not growing. And yet mobile service revenue has now risen year over year for twenty-one consecutive quarters.¹

That is the puzzle worth unpacking.

遠傳電信 Far EasTone Telecommunications Co., Ltd., ticker 4904.TW on the Taiwan Stock Exchange, is the island's third-largest integrated telecom operator by revenue. It closed fiscal 2025 with record revenue of NT$110.35 billion and net income of NT$13.7 billion, its highest in two decades.² In early August 2026 the stock traded around NT$101.50, giving it a market capitalization of roughly NT$366 billion.[^3]

Third place is usually a bad place to be in telecom. The economics of the industry punish subscale players brutally: you pay nearly the same for spectrum, nearly the same for base stations, and nearly the same for handset subsidies as the market leader, but you spread those fixed costs over fewer customers. For most of its history, Far EasTone lived that math.

Then something unusual happened. Between 2020 and 2023, Far EasTone bought its smallest competitor. And after the deal closed, instead of celebrating a bigger subscriber number, it deliberately threw away nearly 900,000 accounts.

That decision — to shrink the customer base on purpose in an industry religiously devoted to subscriber counts — is the analytical heart of this story. It either represents genuine capital discipline and a differentiated read on where value lives in a mature telecom market, or it represents an accounting-flattering optical trick that papers over the fact that the company lost share. The evidence, as we will see, points more toward the former than the latter, but with real caveats that management has not yet resolved.

Here is where we are going. We start with the origins: a Taiwanese textiles-and-cement conglomerate that talked AT&T into a joint venture just as the government cracked open a state telecom monopoly. We move quickly through the middle years, because the story that matters for investors today begins later. We stop hard at 2018, when Taiwan's three big carriers set fire to their own pricing in a single week. We examine the 2020 spectrum auction, which produced the most expensive 5G bandwidth in the world and a genuine test of each operator's capital discipline. We spend real time on the twenty-one-month regulatory odyssey that turned five operators into three. Then we look at what the company actually is today — a mobile business that still generates nearly all the profit, wrapped in a fast-growing enterprise IT and AI services arm that management wants you to believe is the future.

And finally we test the claim. Because "essential services" is a phrase, not a moat, and the difference between the two is what long-term holders are actually paying for.

↑ Back to Top

II. Origins: A Conglomerate's Bet on Deregulation (1996–2005)

To understand why a company that makes polyester fiber and cement ended up running a mobile network, you have to understand 遠東集團 Far Eastern Group and the family that built it.

The group traces back to 1940, when 徐有庠 Hsu Yu-Ziang started what Forbes has described as an underwear supply business in mainland China. When the Chinese civil war ended in 1949, the family moved the business to Taiwan and rebuilt it there.⁵ Over the following half-century it metastasized in the way successful Asian family conglomerates do — textiles into petrochemicals, petrochemicals into cement, cement into shipping, shipping into retail, retail into banking. By the 1990s Far Eastern Group was one of Taiwan's genuine industrial dynasties, with Asia Cement, Far Eastern New Century, U-Ming Marine Transport, and Oriental Union Chemical among its holdings.⁵

徐旭東 Douglas Hsu inherited the chairmanship from his father. Educated in the United States, with a bachelor's and a master's degree from Notre Dame — where he later became a trustee emeritus — Hsu is the archetype of the second-generation Asian conglomerate heir who was sent abroad to be trained and came back to run something considerably larger than what he was handed.⁵

In the mid-1990s, Taiwan's government did something that changes industries: it deregulated. The mobile telecommunications market, which had been the exclusive preserve of the state-linked incumbent, was opened to private licensees. For a conglomerate with capital, political relationships, and no telecom expertise whatsoever, this was a classic Far Eastern Group situation — a licensed, capital-intensive, government-adjacent business where the barrier to entry was money and access rather than technology.

The technology problem they solved by importing it. In 1996 the group set up a preparatory office for a new mobile venture in partnership with AT&T, the American telecom giant, and the company formally began commercial operations in early 1998.³ It was awarded GSM 900 and GSM 1800 licenses in January 1997, and at launch it operated what the group described as the world's first fully integrated GSM900/1800 dual-band system.³

That detail deserves a moment of translation for non-engineers. A "dual-band" network means the operator could run traffic across two different radio frequency ranges simultaneously and hand a call between them without dropping it. The lower 900 MHz band travels further and penetrates buildings better; the higher 1800 MHz band carries more capacity in dense areas. Running both as one integrated system meant better coverage in the countryside and better capacity in Taipei, from the same network. In 1998, against an incumbent still operating a single-band analog-to-digital transition, that was a genuine engineering edge.

It was also, in hindsight, the last time Far EasTone had a clear technology lead over 中華電信 Chunghwa Telecom that anyone outside the industry noticed.

The AT&T partnership did not last. In October 2003, AT&T Wireless announced it was selling its roughly 23% stake in Far EasTone for US$330 million, with Far Eastern Group affiliates buying it out through a tender offer — part of a broader retreat from overseas businesses to focus on its North American network.⁴ Far EasTone became, definitively, a Far Eastern Group company rather than a joint venture, and it has remained one ever since.

The capital markets chapter followed. Far EasTone moved onto Taiwan's over-the-counter market and then graduated to a full Taiwan Stock Exchange main-board listing in August 2005, trading as 4904 in the electronics sector.²¹ Commercial 3G WCDMA service — the generation that first made mobile data usable — arrived the following year.²¹

Two things carry forward from this founding era, and only two.

The first is control. Douglas Hsu has chaired Far EasTone since its founding and still chairs it nearly three decades later, a tenure long enough to span the entire arc from GSM to 5G standalone.³¹⁴ Far Eastern Group entities remain the company's dominant shareholder bloc.²¹ Whatever else you conclude about Far EasTone, you are not analyzing a company with a contestable ownership structure. There is no activist path here, no realistic hostile bid, no board coup. Governance runs through the family and the group.

The second is the business model's original sin: Far EasTone was built as a licensed toll-collector on a network, funded by conglomerate capital. That is a wonderful business when you are the only licensee and a punishing one when there are five. Which is precisely what Taiwan's regulators created next.

↑ Back to Top

III. Becoming a Top-3 Carrier: Scale, Spectrum, and the Cost of Staying Relevant (2006–2017)

Picture Taipei's Xinyi district around 2012. Within a four-block radius you could walk past storefronts for six different mobile operators, each with the same handsets in the window, each with a slightly different subsidy, each staffed by salespeople whose job was to move a customer from one identical network to another identical network for a NT$100 monthly difference.

That image is the whole industry structure of this decade, compressed.

Taiwan's mobile market hardened during these years into a rigid, unhappy oligopoly. Chunghwa Telecom, the privatized descendant of the state monopoly, sat at the top with roughly 40% of the market, an unmatched fixed-line and backbone infrastructure inherited from its government years, and the balance sheet to outspend anyone.¹⁶ 台灣大哥大 Taiwan Mobile, backed by the Fubon financial conglomerate, held second place. Far EasTone fought from third.

Below them sat a tail of smaller operators — 亞太電信 Asia Pacific Telecom, backed at various points by 鴻海 Hon Hai, and 台灣之星 Taiwan Star Telecom — that were individually too small to earn a return on network investment but collectively large enough to keep prices down for everyone. This is the worst possible market structure for incumbents: enough small players to prevent pricing discipline, but not enough scale among them to consolidate naturally.

The mechanism that made this so painful is worth explaining plainly, because it is the engine of everything that follows.

A mobile network is almost entirely fixed cost. You buy spectrum from the government at auction — a one-time payment, often in the billions, for the right to use a slice of radio frequency for fifteen or twenty years. You then build base stations to use it, and you must build enough of them to cover the whole island, because a network that only works in Taipei is not a network anyone will pay for. Once built, the marginal cost of carrying one more customer's data is approximately nothing.

Two consequences follow. First, scale is everything: the operator with the most subscribers spreads the same fixed cost over more revenue and earns a structurally higher margin. Second, once the network exists, the temptation to fill it with cheap traffic is nearly irresistible, because any revenue above zero looks accretive in the short run. This is why telecom price wars are so destructive and so recurrent. The economics that make discounting rational for any one operator on any one day make the industry collectively poorer every day.

Through the 2006–2017 stretch, Far EasTone did what a well-capitalized number three does. It paid its toll at each spectrum auction. It upgraded through 3G, then 4G LTE, then what the industry marketed as "4.5G." In 2016 it aggregated 700 MHz, 1800 MHz and 2600 MHz spectrum into a tri-band service and set up Taiwan's first 5G laboratory — a genuine, if modest, claim to technical leadership in the region.⁶

None of it produced durable differentiation. Coverage converged. Speeds converged. Handsets were identical, because they came from the same two or three vendors. What was left to compete on was price and subsidy, and both compete away.

The investor takeaway from this decade is structural rather than dramatic. Far EasTone entered the late 2010s with a competent network, a respectable third-place share, and no defensible source of pricing power. Its return on capital was hostage to how disciplined its two larger rivals felt in any given quarter, and to how desperate the two subscale players below it became.

In May 2018, everyone found out exactly how desperate.

↑ Back to Top

IV. The NT$499 Chaos: A Self-Inflicted Wound (2018)

The queues started before dawn.

In the second week of May 2018, retail branches of all three of Taiwan's major carriers were mobbed. Customers waited hours. Store systems buckled. Call centers were overwhelmed. Over a seven-day window beginning May 9, Chunghwa Telecom alone processed roughly 1.44 million applications; Taiwan Mobile and Far EasTone took in about 300,000 each.⁸ For a market of roughly 29 million mobile accounts, this was a national event.

What everyone was queuing for was a mobile plan priced at NT$499 a month with unlimited data.

The trigger was a Chunghwa Telecom decision to relaunch an unlimited flat-rate plan at that price, and to open it not just to new subscribers but to existing customers renewing contracts and to switchers.⁷ The previous year, comparable plans had been priced around NT$699. Taiwan Mobile and Far EasTone matched within days, because in a market where networks are indistinguishable, not matching means watching your base walk out the door.

Then came the number that turned a promotion into a catastrophe: roughly 80% of the applicants were existing subscribers downgrading from more expensive plans.⁸

Read that again, because it is the entire lesson. The carriers did not buy growth. They bought a discount for their own most valuable customers, and paid for it with their own margins. Every one of those 1.44 million Chunghwa applications that came from an existing NT$699 customer was a NT$200 monthly self-inflicted wound, locked in for a 24- or 30-month contract term.⁷

The damage showed up immediately in average revenue per user — ARPU, the industry's central metric, which is simply total service revenue divided by subscribers. Far EasTone's fell from NT$883 to NT$843. Taiwan Mobile's dropped from NT$861 to NT$825. Chunghwa's went from NT$590 to NT$581.⁷ All three reported year-on-year profit contractions exceeding 10% in the first quarter, and their share prices fell as investors did the arithmetic on what an entire subscriber base repricing downward would do to earnings.⁷

The regulator's response added insult. On May 23, 2018, the 國家通訊傳播委員會 National Communications Commission fined all three operators — not for the pricing itself, but for the operational chaos. The NCC found they had failed to plan adequately, underestimating the staffing needed to handle the volume and degrading service for everyone else. Chunghwa took the heaviest penalty at NT$2 million because it drew the most applicants and had the worst staffing shortfall; Taiwan Mobile and Far EasTone were fined NT$600,000 each.⁸

The fines were trivial. The precedent was not. Taiwan's regulator had established that it would intervene in retail mobile pricing conduct, and that intervention risk has hung over the sector's pricing behavior ever since.

There is a management-credibility footnote here worth preserving. In the middle of the episode, Far EasTone's then-president Yvonne Li characterized the NT$499 plans as a "short-term phenomenon" and said the three carriers had "no intention of initiating a price war in a nearly saturated market."⁷ Chunghwa's relaunch contradicted that within days. It is a useful reminder, when reading any telecom executive's assurances about competitive rationality, that pricing discipline in an oligopoly is a Nash equilibrium, not a promise. It holds until one player decides it doesn't.

So what does an investor take from 2018? Two things.

First, a hard demonstration that in a market with no product differentiation, price is the only competitive lever, and pulling it destroys value for every participant simultaneously. Nobody won the NT$499 war. All three carriers ended it poorer.

Second, the strategic question that the next eight years of this story exist to answer: if you are the number three player in a commoditized market where the leader can start a price war whenever it wants, what do you actually do?

Far EasTone's answer came in two parts. Buy technology leadership through spectrum. Then buy the market structure itself.

↑ Back to Top

V. The 5G Bet: Spectrum Arms Race and a Real Benchmarking Test (2019–2020)

The auction ran for 261 rounds.

By the time Taiwan's 5G spectrum auction concluded on January 16, 2020, the bidding had gone on so long and climbed so high that local commentary was calling the 3.5 GHz band the most expensive 5G bandwidth in the world. The first-stage quantity auction produced total bids of NT$138.08 billion — roughly US$4.6 billion — across five bidders for a country of 23 million people.⁹

To calibrate: that is a per-capita spectrum bill several times what operators paid in most European markets for comparable 5G bands. Taiwan's carriers, collectively, wrote one of the largest relative checks in the world for the right to broadcast.

The allocation at 3.5 GHz — the workhorse mid-band frequency where nearly all real 5G capacity lives — broke down as follows. Chunghwa Telecom took 90 MHz for NT$45.675 billion. Far EasTone took 80 MHz for NT$40.6 billion. Taiwan Mobile took 60 MHz for NT$30.4 billion. Taiwan Star took 40 MHz for NT$19.708 billion. In the high-band 28 GHz range, Chunghwa took 600 MHz, Far EasTone 400 MHz, Taiwan Mobile 200 MHz, and Asia Pacific Telecom 40 MHz.⁹ Including the second-stage auction in February that determined where in the band each operator's spectrum would physically sit, Far EasTone's total 5G spectrum outlay reached approximately NT$43.04 billion.¹⁰

Now, a correction to a common framing of this auction, and it matters for judging capital allocation.

It is often said that Taiwan Mobile secured better unit economics than its rivals. It did not. The 3.5 GHz auction cleared at a uniform price of NT$5.075 billion per 10 MHz — meaning roughly NT$508 million per megahertz for every winner.⁹ Chunghwa, Far EasTone, and Taiwan Mobile all paid essentially the identical price per unit of spectrum. Nobody got a discount and nobody overpaid relative to the others.

The real capital-allocation decision was not price. It was quantity.

Far EasTone chose to buy 80 MHz — 89% as much mid-band spectrum as the market leader, despite having meaningfully less than 89% of the market leader's revenue. Taiwan Mobile chose 60 MHz, two-thirds of Chunghwa's holding, and kept roughly NT$10 billion in its pocket. Those are genuinely different bets, and they reveal genuinely different philosophies.

Here is the case for Far EasTone's choice. In mobile networks, spectrum is capacity, and capacity is the physical ceiling on how many customers you can serve at a given speed. Think of it as lanes on a highway: more spectrum means more lanes, which means traffic flows faster at rush hour without you having to build more interchanges. Crucially, spectrum is a substitute for capital expenditure — an operator with more spectrum can serve the same traffic with fewer base stations, because each site carries more. Buying spectrum up front is, in part, buying the right not to spend on cell sites later.

Spectrum is also strictly finite and awarded on multi-decade licenses. If you skip a round, you do not get another chance for fifteen years, and you spend that time as the network that is measurably slower during peak hours. In a category where the product is otherwise identical, measurable slowness is the one difference customers can actually feel — and the one thing that would let a rival re-open the price war on favorable terms.

Here is the case against. NT$43 billion is an enormous sum for a company that would generate NT$13.7 billion of net income in its best year two decades later.² It is capital that cannot be returned to shareholders, cannot fund acquisitions, and earns nothing until it is built out and sold. And the honest historical record on spectrum auctions — Europe's 3G disaster in 2000 being the canonical example — is that operators consistently overpay in auctions because each is bidding against the fear of exclusion rather than against a discounted cash flow model.

Which read is right? Six years on, the evidence leans toward Far EasTone. By mid-2026 the company was reporting the highest postpaid 5G penetration among Taiwan's three carriers and 5G renewal pricing carrying roughly a 40% uplift over the 4G plans customers were leaving.¹¹⁸ That uplift is the payoff. It is the first time in this story that a technology transition translated into higher prices rather than the same prices with more cost behind them.

But note the qualifier. The spectrum bet worked in combination with what came next. Had Taiwan remained a five-operator market, that NT$43 billion would have been the price of maintaining parity in a price war, not the price of pricing power. The auction was necessary. It was not sufficient.

What made it sufficient was a phone call Far EasTone made to the smallest operator in the market, eight months later.

↑ Back to Top

VI. The Consolidation Shockwave: 5-to-3 and the Asia Pacific Telecom Merger (2020–2023)

On a Friday evening in early September 2020 — the classic slot for news a company would rather not have chewed over during trading hours — Far EasTone announced it would pay up to NT$5 billion, about US$169 million, for 500 million newly issued shares of Asia Pacific Telecom, taking an 11.58% stake in Taiwan's smallest major carrier.¹⁰

On its face, it looked odd. Why would the number three operator buy a minority position in the number five?

The stated logic was spectrum sharing. Asia Pacific Telecom had won no 3.5 GHz spectrum in the January auction — it walked away with 40 MHz at 28 GHz, a high-frequency band useful for dense hotspots and nearly useless for broad coverage.⁹ Without mid-band spectrum, APT could not offer competitive 5G at all. Far EasTone had 80 MHz of it. The arrangement gave APT twenty-year access to Far EasTone's 3.5 GHz frequencies, with APT agreeing to pay NT$9.47 billion toward network deployment costs, while Far EasTone got a partner to help fill the capacity it had just paid a fortune for.¹⁰¹¹ President 井琪 Chee Ching framed it as a way to "gain better spectral efficiency and reduce network deployment costs."¹⁰

The strategic logic underneath was less about efficiency and more about optionality. The agreement also contemplated Far EasTone lifting its stake to 23.8% by June 2022 through a subsequent share swap with Hon Hai, APT's largest shareholder.¹⁰ Far EasTone had, in effect, bought a call option on the consolidation of Taiwan's telecom market — and made itself the only logical acquirer of APT by making APT's 5G service dependent on its network.

Eighteen months later, it exercised.

On April 15, 2022, Asia Pacific Telecom's shareholders approved a full merger valued at NT$24.7 billion, about US$882 million. The structure was a stock swap: one APT common share for 0.093 Far EasTone shares, with Far EasTone issuing 356.68 million new shares including a 93.44 million-share private placement tranche.¹¹ The subscriber math was straightforward — Far EasTone's base would go from 7.05 million to 9.2 million.¹¹

Simultaneously, Taiwan Mobile moved to acquire Taiwan Star Telecom. Taiwan's five-operator market was about to become three.

Then the deal ran into the machinery of the Taiwanese state, and stayed there for a long time.

The National Communications Commission approved conditionally on January 30, 2023, attaching network build-out obligations: Far EasTone committed to installing 500 new 5G and 1,000 new 4G base stations within one to two years, with APT adding 500 4G sites.¹¹¹³ But the 公平交易委員會 Fair Trade Commission — the competition regulator, whose concern was not coverage but concentration — took considerably longer. The merger agreement's deadline was extended in May 2023. By July 1, 2023, with the FTC still deliberating, Far EasTone declined to extend again, saying only that it would "cautiously evaluate issues regarding the merger based on the spirit of the original agreement."¹²

That was a negotiating posture as much as a statement of fact. Taiwan Mobile had by then materially cut the price it was paying for Taiwan Star, and the market read Far EasTone's refusal to extend as an attempt to reopen terms.¹² The company had also told investors to expect roughly NT$3 billion in synergy benefits within a year of closing — a number worth remembering, because it is the yardstick against which the deal should be judged.¹²

The Fair Trade Commission finally cleared the transaction on July 20, 2023, and the conditions it attached are instructive about how Taiwan regulates concentration.¹³ Far EasTone had to honor all existing APT subscriber contracts through the end of 2025. Vulnerable subscribers — those 65 and over, low-income households, and people with disabilities — were guaranteed discounted plans for at least five years. General customers got access to reduced-rate plans for a minimum of one year. And the company had to report annually for five years on tariff benefits and service quality.¹³

The FTC was explicit that it worried the deal "could elevate FET's share in the mobile broadband service market," and approved it only on the strength of those consumer protections and network commitments.¹³ In other words: Taiwan's regulators permitted consolidation, but priced it in consumer concessions rather than blocking it outright. That is the pattern to expect if further structural change is ever contemplated.

The merger took legal effect on December 15, 2023.⁶¹⁴ Chairman Hsu called the process "challenging." Chee Ching said it "heralds a new phase of accelerated and sustainable development."¹⁴

The physical integration was where the value actually lived. Far EasTone set about consolidating 5,000 base stations — decommissioning redundant sites where two networks had covered the same ground twice.¹⁴ The company estimated the consolidation, combined with AI-driven network management, would cut annual electricity consumption by 86 million kilowatt-hours and carbon emissions by roughly 93,000 tons.¹⁴ Former APT customers got access to nearly double the base station coverage in the first integration phase. On the spectrum side, the combined entity held 980 MHz of total 4G and 5G bandwidth, including 120 MHz of 5G — the maximum any single operator was permitted.¹⁴

The subscriber purge

And then came the move that makes this story worth telling.

In the first quarter of 2025, Far EasTone shed 881,073 subscribers. The company attributed the decline to clearing non-contributing and duplicate accounts and to cooperation with government anti-fraud measures.¹⁵

Nearly nine hundred thousand accounts — roughly 9% of the post-merger base — deleted. In an industry where subscriber growth is the headline every quarterly release leads with, this was the opposite of a victory lap.

The result was the number that vindicated it. Far EasTone's blended mobile ARPU rose to NT$513.1, overtaking Chunghwa Telecom's NT$512.7 — the first time the perennial number three had earned more per subscriber than the former state monopoly. Taiwan Mobile, over the same period, ran at NT$448.1.¹⁵

This is the cleanest available window into what the merger was actually for. Far EasTone did not buy Asia Pacific Telecom to get 2.15 million customers. It bought APT to get spectrum, to remove a price-cutting competitor from the market, and to take a network out of service. The customers it wanted, it kept and upgraded. The ones that generated nothing — dormant SIMs, duplicate accounts, prepaid cards sitting in drawers — it deleted, because carrying them cost money in regulatory reporting, billing systems and fraud exposure while contributing nothing.

Taiwan Mobile, absorbing Taiwan Star on a similar timeline, ran the opposite playbook. It kept the volume, ending the period with 9.38 million subscribers, and extracted its value through cost synergies instead — driving telecom operating profit to a seven-year high with a 30% year-over-year improvement.¹⁵

Both approaches worked. That is worth saying plainly, because the temptation is to declare Far EasTone's the smarter one. Taiwan Mobile's route delivered a larger immediate profit step-up and left it with a bigger customer base to monetize later. Far EasTone's route delivered a cleaner base, a higher revenue per customer, and a better position from which to raise prices — but at the cost of market share, which in a fixed-cost business is a real cost.

The tell for long-term investors is which base compounds better as the market matures. A smaller, higher-value base with low churn should generate rising revenue per customer for years. A larger base with lower revenue per customer has more upgrade headroom but more exposure to price competition. Neither question is settled.

What is settled is the market structure. Taiwan today is a three-player oligopoly: Chunghwa Telecom with roughly 40% mobile market share, Taiwan Mobile second, Far EasTone third.¹⁶ Five desperate operators became three disciplined ones. And that, far more than any operating improvement, is what changed the economics of the business Far EasTone runs today.

↑ Back to Top

VII. The Core Business Now: How a #3 Telecom Actually Wins

Strip away the narrative and ask a simple question: where does Far EasTone's money actually come from?

The answer, in 2026, is still overwhelmingly this — people paying monthly bills for mobile phone service. In the second quarter of 2026, mobile service revenue was NT$15.8 billion out of NT$28.66 billion in total consolidated revenue.¹ The rest is handset and merchandise sales, which carry thin margins, fixed broadband, and the enterprise ICT business we will get to shortly.

But revenue share understates the picture badly. Mobile service revenue is high-margin and largely fixed-cost-covered; handset sales are close to pass-through; and the enterprise ICT segment runs at margins of roughly 20–25%, well below the group EBITDA margin of 35.8%.¹⁸ Which means mobile service contributes a share of profit substantially larger than its share of revenue. Everything else in this company is, financially speaking, a rounding adjustment on top of a mobile phone business.

Hold that thought. It is the discipline that keeps the rest of the analysis honest.

The operating evidence

The mobile business is, by its own metrics, performing about as well as a mobile business in a saturated market can.

Mobile service revenue grew 3% year over year in the second quarter of 2026 — the twenty-first consecutive quarter of growth.¹ Postpaid ARPU reached NT$733 per month, moving back toward the NT$758 level the company ran at before the merger diluted it with APT's lower-value base.¹⁸ Postpaid 5G penetration hit 49.3%, with management targeting a crossing of 50% during 2026.¹ The churn rate in the first half of 2026 was 0.7%, a record low.¹

Take churn seriously, because it is the single most informative number in this entire story.

Churn measures the percentage of subscribers who leave each month. In Taiwan, mobile number portability means a customer can switch carriers and keep their phone number, which removes the main friction that used to lock people in. Switching is easy, coverage is comparable across all three networks, and every operator will happily subsidize a handset to win a port-in. Under those conditions, a churn rate of 0.7% per month — implying roughly 8% of the base turning over annually against an industry where double that is unremarkable — is not a marketing claim. It is a behavioral fact about customers who could leave and don't.

And crucially, it coexists with the highest ARPU in the market.¹⁵¹⁸ Charging the most while losing the fewest is the observable signature of pricing power. It is what a moat looks like in operating data rather than in a slide deck.

The mechanism management points to is 5G upgrade pricing. When a customer's 4G contract expires and they migrate to a 5G plan, the monthly fee rises roughly 40%.¹⁸ With half the postpaid base still on 4G, there is a mechanical, contractual pipeline of price increases stretching out several years — not a hoped-for improvement, but a repricing that happens automatically as contracts roll.

Porter's five forces, honestly applied

Rivalry is the force that changed most. Three operators with converged networks would normally still compete on price, but the 2018 experience taught all three exactly what that costs, and the removal of two subscale players eliminated the participants with the least to lose. Rivalry today is intense but rational. It is not, however, structurally guaranteed to stay that way — nothing prevents Chunghwa from repeating 2018 if it decides share matters more than margin.

Supplier power is real and mostly comes in two forms. Equipment vendors — Ericsson, Nokia and their peers — hold genuine leverage in a market with few alternatives. More importantly, the state is effectively the monopoly supplier of the single most critical input, spectrum, and sets its price through auctions it designs. The NT$43 billion Far EasTone paid in 2020 is what supplier power looks like when the supplier is a government.¹⁰ Spectrum licenses expire and must be renewed or re-auctioned; that is a recurring, unavoidable claim on future capital.

Buyer power is high and getting no lower. Number portability, price transparency, and identical products mean any individual customer can walk. What Far EasTone has done is not eliminate buyer power but make walking less attractive by attaching things to the subscription that a competitor's SIM card does not come with.

Threat of new entrants is essentially nil. To enter, you would need to win spectrum at auction — where the price is set by three incumbents who can each fund the bid from operating cash flow — and then build a national network from zero. Asia Pacific Telecom and Taiwan Star, both better capitalized than any hypothetical new entrant, could not make the economics work. Nobody is entering.

Substitutes are the most underrated force. OTT messaging and calling — the WhatsApps, LINEs and FaceTimes of the world — have already destroyed voice and SMS as revenue lines across global telecom. That damage is largely done. Wi-Fi offload continues to cannibalize data usage at home and in offices. The forward substitute risk is low-earth-orbit satellite connectivity, which Far EasTone is treating as a partner opportunity rather than a threat — the company told analysts it is preparing low-orbit satellite services and fielded questions about Amazon's LEO constellation economics.¹⁸²⁰ Whether satellite ends up as a complement to terrestrial networks or a competitor for the connectivity dollar is genuinely unresolved, and telecom operators have historically been wrong about the direction of that answer.

Seven Powers: what Far EasTone actually has

Hamilton Helmer's framework asks which specific, durable mechanism prevents a competitor from replicating your returns. Applied honestly to Far EasTone, most of the seven powers are absent.

There are no network economies — a mobile network is a physical network, not a network effect; the product does not get better for you because your friends use the same carrier. There is no meaningful brand power in the pricing sense; nobody pays more for a Far EasTone SIM the way they pay more for a Hermès bag. There is no cornered resource — spectrum is licensed to all three. There is no process power worth claiming; network operations are broadly commoditized.

Two powers are genuinely present.

Scale economies are real but relative. Far EasTone is subscale against Chunghwa and always will be. What changed is the denominator: the fixed costs of running a national network are now spread across three operators instead of five, and Far EasTone's own fixed cost base fell as it decommissioned 5,000 redundant base stations.¹⁴ Its scale did not improve so much as everyone else's opportunity to undercut it disappeared.

Switching costs are the power management is actively trying to build, and it is the most interesting part of the story. The strategy — which the company calls "essential services" — is to attach things to the mobile subscription that create friction beyond the phone number: insurance products sold through the carrier relationship, direct carrier billing that turns the phone bill into a payment rail, IoT and smart-home devices, and content through the friDay Video streaming service. In the fourth quarter of 2025 the Mobile Circle app had 8 million downloads and over 2 million monthly active users; direct carrier billing transactions rose 14% year over year; mobile insurance EBITDA more than doubled at +107%, and travel insurance grew 266%.¹⁷

The logic is sound: a customer with a phone plan can leave tomorrow, but a customer with a phone plan, an auto-billed insurance policy, a streaming subscription and three connected devices has to do actual work to leave. Record-low churn is at least consistent with this working.

Myth versus reality

Myth: Far EasTone won the consolidation round because it got bigger. Reality: It got smaller in subscribers and lost market share.¹⁵ What it won was a better market structure and a better customer mix. Those are different achievements, and only the second is within management's control.

Myth: Above-peer ARPU proves Far EasTone has premium brand power. Reality: ARPU leadership came substantially from deleting low-value accounts and from a 5G upgrade cycle that all three carriers are running. Chunghwa reported 4G-to-5G upgrade fee increases of about 38%, Taiwan Mobile roughly 45%, and Far EasTone about 40% — the uplift is an industry phenomenon, not a Far EasTone one.¹⁶ The differentiated part is churn, not price.

Myth: 2026's 6.5% revenue guidance signals accelerating growth. Reality: Management's CFO told analysts on the fourth-quarter 2025 call that consolidating the electronic toll collection business adds roughly NT$3 billion of revenue — close to two-fifths of the guided increase — which is why guided revenue growth of 6.5% sits alongside guided net income growth of just 3.2%.¹⁷ The organic revenue growth rate underneath is closer to 4%. Investors should anchor on the smaller number.

The evidence that is still missing

Here is where the bull case runs out of proof.

Management describes the telecom market as saturated. On the fourth-quarter 2025 call, an analyst asked directly how the company intends to grow core telecom ARPU in that environment. Chee Ching's answer was that Far EasTone offers a "variety of services that are of value to our users," adding: "The telecom piece may be saturated, I think the users' demands are always there, we definitely haven't done enough in there."¹⁷

That is a directionally reasonable answer and a completely unquantified one. There is no disclosed target for how much ARPU the essential-services strategy adds, no disclosed attach rate for bundled services against the subscriber base, no disclosed ARPU differential between a customer with three attached services and one with none. The mechanism is plausible; the magnitude is unmeasured.

Until management puts numbers on it, the switching-costs thesis rests on a single output metric — churn — and output metrics can be produced by many inputs, including simple network quality and competitive restraint by rivals. It is genuine evidence. It is not conclusive evidence.

Which brings us to the part of the business management most wants to talk about, and where the numbers are growing fastest.

↑ Back to Top

VIII. The Adjacent Bets: Smart ICT, AI, and the Digital Ecosystem

If you want to know what a telecom executive believes about the future, listen to which segment they describe first when they have good news.

Through 2025 and 2026, Far EasTone's management has led with enterprise. And unlike most telecom diversification stories — which tend to be small, expensive and quietly abandoned — this one has reached a size where it genuinely moves the reported numbers.

In fiscal 2025, Smart ICT revenue grew 28% year over year, and combined with fixed services accounted for 19% of total company revenue.¹⁷ In the first half of 2026 that share reached 20%, with ICT revenue up 23%.¹ The composition of that growth is where it gets interesting: telecom-based systems integration services grew 64% in 2025, contract value tied to Microsoft Azure and Copilot deployments rose 94% year over year, and cloud-based hospital information systems were adopted by more than 100 clinical facilities.¹⁷ In the first half of 2026, cloud services grew 30%, smart city IoT solutions grew 102%, and digital transformation revenue rose 50%.¹¹⁸

The forward indicator is contract value rather than revenue, and it grew faster than revenue in both periods. First-half 2026 accumulated contract value exceeded NT$8 billion, up 103% year over year, and management noted the contracted amount for digital transformation services already exceeded the whole of 2025.¹⁸²⁰ Contract value is the backlog that converts to revenue over subsequent quarters, so this is a leading signal that ICT revenue growth should persist for at least several more quarters.

What is this business, actually?

Strip the jargon and the enterprise segment is three related things.

The first is systems integration — Far EasTone acting as the contractor that builds and runs an organization's technology infrastructure. When a hospital chain moves its patient records system to the cloud, someone has to design it, migrate it, connect it and keep it running. Far EasTone bids for that work and wins it partly because it already owns the network the data travels over.

The second is smart city infrastructure, and this is the most distinctive piece. Through 博弘雲端 Nextlink Technology, which it owns 67%, and 遠通電收 FETC, the electronic toll collection operator in which it holds 40%, Far EasTone builds and operates AI-driven traffic management, energy management, air quality surveillance, smart metering, EV charging infrastructure, smart harbor energy systems and flood monitoring.¹⁹ Contract value in green and smart city solutions grew 48% in 2025, and smart city revenue doubled year over year in the first half of 2026.¹⁷¹⁸

Both subsidiaries entered Southeast Asian markets during 2025, and in March 2026 the company publicly framed regional expansion as a growth vector, with Chee Ching arguing that AI adoption in city management and urban planning will drive demand.¹⁹ Far EasTone put smart city services at 15–20% of overall revenue.¹⁹

The third is AI enablement services. Far EasTone's framing on the second-quarter 2026 call was that it sees itself as its clients' "AI transformation partners" rather than a software vendor — meaning it sells the consulting, integration and managed-service work of getting an enterprise from an AI pilot to production.¹⁸ Management described these AI adoption services as "relatively high margin," leveraging human capital rather than capital equipment.¹⁸

The discipline signal, and the margin problem

One decision here deserves credit, because it runs against the strongest fashion in global capital markets.

Far EasTone has explicitly declined to pursue large-scale AI data center construction, judging the market still early. That is a notably restrained posture at a moment when telecom operators worldwide are announcing gigawatt ambitions, and when its own largest rival was booking AI data center contracts exceeding NT$1 billion in a single quarter.¹⁶ Choosing not to spend billions on depreciating GPU infrastructure in a market whose end-demand curve nobody can yet see is a defensible act of capital discipline — and one that will look either prescient or timid depending on how the next three years break.

The consumer side of the ecosystem is smaller but strategically coherent. friDay Video is Taiwan's leading local paid OTT platform, with paid subscribers up 18% year over year in 2025, supported by investments in LiTV and Korean drama production.¹⁷¹⁶ It is not a Netflix competitor and does not need to be. Its job is to be one more thing a customer would have to cancel.

Now the honest counterweight, and it is a real one.

ICT growth is dilutive to margins. Far EasTone disclosed on the second-quarter 2026 call that ICT segment margins run at 20–25% against a group EBITDA margin of 35.8%.¹⁸ Every incremental dollar of ICT revenue therefore lowers the blended margin. This is why revenue growth of 13.6% in the quarter translated to EBITDA growth of only 8.1%.¹ The segment is accretive to absolute profit but dilutive to profitability ratios — and investors who reward telecom operators on margin will find that uncomfortable.

The contracts are long-cycle and government-dependent. Smart city work is procured by municipalities and national agencies on multi-year tenders. That cuts both ways: revenue visibility is excellent, but the pipeline is exposed to budget cycles, political turnover, and procurement delays. Analysts on the second-quarter 2026 call pressed specifically on project delivery schedules affecting revenue recognition — a mundane-sounding question that is actually the core execution risk in this segment.¹⁸

Southeast Asia is unproven. Winning a smart traffic contract in Taipei, where Far EasTone has three decades of relationships and owns the network, tells you very little about winning one in Ho Chi Minh City, where it has neither. The expansion is optionality, not a plan with disclosed targets.

And the sizing discipline still applies. A segment at 20% of revenue and 20–25% margins contributes materially less than a fifth of profit. Smart ICT is real, fast-growing and strategically necessary. It is not yet, and will not soon be, the thing that determines whether this company earns its cost of capital.

Which makes the question of what management does with the cash the mobile business generates the most consequential question in the story.

↑ Back to Top

IX. Management, Ownership, and Capital Allocation

There is a moment on the fourth-quarter 2025 earnings call that tells you more about Far EasTone's management than any strategy slide.

An analyst, looking at a company generating NT$25.91 billion of free cash flow against a net debt position of NT$36.03 billion — a ratio of just 0.9 times EBITDA — asked the obvious question.¹⁷ Why not pay out more than 100% of earnings? The balance sheet can clearly take it.

Chee Ching did not take the easy win. "There's always a possibility. You never say never," she said, before adding: "Given the global economy and all this uncertainty, I think it's good that we leave ourselves some buffer."¹⁷

Turning down free applause from your own shareholders is not the reflex of a promotional management team. It is worth noting because most of what follows in this section supports the same read.

The ownership structure

Far EasTone is controlled, not merely influenced, by its parent group. 遠東新世紀 Far Eastern New Century Corporation and affiliated entities hold roughly a third of the shares, a bloc that has been stable for two decades and that makes any contest for control effectively impossible.²¹ Douglas Hsu, now 83, remains chairman.⁵¹⁴

Day-to-day operations run through Chee Ching as president, and the division of labor appears clean — Hsu is the group patriarch and board chairman, Ching is the operator who fronts earnings calls, negotiates deals and answers for the numbers.

For investors this structure is a genuine trade-off rather than a straightforward negative. On the positive side, it produces the long time horizons that a business with fifteen-year spectrum licenses and decade-long network cycles actually requires. Nobody at Far EasTone is managing to a quarterly print in order to survive a proxy fight. The NT$43 billion spectrum bet and the twenty-one-month regulatory slog on the APT merger both required patience that a quarterly-driven management might not have had.

On the negative side, minority shareholders have essentially no mechanism to force change if the controlling family's interests diverge from theirs.

The activist stress test

What would a skeptical investor attack here? Three things, in order of seriousness.

Related-party complexity. Far EasTone sits inside a conglomerate spanning cement, textiles, petrochemicals, shipping, retail and financial services.⁵ Conglomerate affiliates are customers, suppliers, landlords and co-investors of one another as a matter of routine. Nothing in the public record suggests impropriety at Far EasTone, and the company is a separately listed, separately audited entity subject to Taiwan Stock Exchange and Market Observation Post System disclosure. But structurally, this is a standing item: capital allocation decisions inside a family conglomerate can be made for group reasons, and problems elsewhere in Far Eastern Group — reputational, financial or regulatory — can spill into Far EasTone's cost of capital without anything going wrong at Far EasTone itself.

Diversification drift. The Smart ICT segment is genuinely fast-growing, but it is also structurally lower-margin than the core, and its expansion into Southeast Asia takes management attention far from the network. The word for this when it goes wrong is diworsification, and the honest assessment is that it has not gone wrong yet — the growth is real, the contract backlog is real, and management has resisted the most capital-destructive version of the AI opportunity. But the pattern to watch is capital and management bandwidth flowing toward a segment that earns less on it.

Disclosure granularity. For a company whose central strategic claim is that bundled essential services raise switching costs, the absence of any disclosed attach rate, cross-sell metric or ARPU differential is a meaningful gap. Investors are asked to accept the mechanism on the strength of a churn number and a qualitative assurance.

The capital allocation record

Now the evidence, which is mostly good.

For fiscal 2025, Far EasTone proposed a dividend of NT$3.81 per share — an exact 100% payout of its NT$3.81 in earnings per share, a 7% increase over the prior year, and the eighteenth consecutive year of dividend payments.²¹⁷ Consistency across eighteen years spanning a price war, a spectrum auction and a merger is the kind of track record that is difficult to fake.

The balance sheet has strengthened rather than weakened through all of it. Net debt fell from NT$36.03 billion at the end of 2025 to NT$32.8 billion by mid-2026, with net debt to EBITDA at 0.83 times and first-half free cash flow of NT$14.5 billion.¹⁷¹⁸ The company financed a NT$24.7 billion acquisition and a NT$43 billion spectrum purchase while raising the dividend every year. That is not a common combination.

The one genuine blemish is capital expenditure discipline in the other direction — underspending.

Fiscal 2025 cash capex came in at NT$7.01 billion against original guidance of NT$8.3 billion, a 16% shortfall management attributed to cost efficiencies and deferrals.²¹⁷ Guidance for 2026 was set at NT$9.6 billion, roughly 8% of sales and a 37% increase.²¹⁹ Pressed on whether 2026 should be treated as the new baseline, given underspends in both 2024 and 2025, Chee Ching pushed back: "I wouldn't say so. Quite a bit of that is gonna be realized this year. I don't think next year will be at the same level. It should come down."¹⁷

There are two readings. The charitable one is that Far EasTone genuinely found network efficiencies — plausible, given it was simultaneously decommissioning 5,000 base stations from the merger — and is now stepping up spending for 5G standalone deployment before easing off. The skeptical one is that a company chronically guiding above what it spends is either padding its guidance or deferring necessary investment to protect near-term earnings and the dividend.

The mid-2026 evidence tilts toward the skeptical read on timing, if not on intent: first-half capex was only NT$3.0 billion against NT$9.6 billion of full-year guidance, requiring an implausibly back-loaded second half.¹⁸ The board approved an additional NT$1 billion of capex focused on the second half.¹⁸ Whether the full number lands is a concrete, checkable test of guidance credibility that resolves in about six months.

Credibility, on balance

Judge management by whether its behavior matched its words over time, and the record is better than average. It bid at parity with peers rather than recklessly at auction. It walked away from a merger extension rather than accept deteriorating terms, then closed the deal. It promised roughly NT$3 billion in merger synergies and subsequently delivered a step-change in ARPU and margins.¹²¹⁵ It declined to over-distribute capital when invited to. It declined to build AI data centers when that was the fashionable thing to announce.

The weaknesses are specific and consistent: management is precise about capital returns and imprecise about growth. Ask about the dividend and you get a number. Ask about how much ARPU headroom exists, or what return the AI and ICT investments will earn, or when, and you get direction without magnitude.

That asymmetry is not unique to Far EasTone. But it is exactly what the earnings calls reveal when you read several of them side by side.

↑ Back to Top

X. Reading the Calls: What Changed, What Didn't

Earnings calls are where a company's story meets people whose job is to poke it. Read three of Far EasTone's across eight years and you can watch the strategy change in real time — and watch which questions management has never quite answered.

2018, the price war calls. The register was defensive. The framing was service quality and the assurance that carriers had no appetite for a price war in a saturated market — a characterization of the NT$499 plans as a "short-term phenomenon" that the market falsified almost immediately.⁷ The implicit strategic model was: we compete on network and service, price is an aberration, this will pass.

2020, the spectrum calls. The register was justification. Management's task was to explain to investors why a company its size had committed NT$43 billion to radio frequency, and the framing was the 5G race — technology leadership as the thing that would eventually restore pricing power.¹⁰ The strategic model had shifted: we cannot differentiate on service, so we will differentiate on network capability, and we will pay for it.

2022–2023, the merger calls. The register was regulatory anxiety, then vindication. The dominant question through 2022 and the first half of 2023 was timeline — when would the Fair Trade Commission rule, what conditions would attach, would the deal survive its own deadlines.¹²¹³ After closing, the framing flipped to ARPU discipline and synergy delivery.¹⁵ The strategic model had shifted again, and more fundamentally: we cannot differentiate on network either, because all three networks are good enough — so we will change the market structure and then price rationally within it.

2025–2026, the essential services calls. The register is confident on the numbers and vague on the future. The fourth-quarter 2025 call was strong on hard metrics: record revenue, twenty-year-high net income, record-low churn, industry-leading 5G penetration.²¹⁷ It was noticeably softer wherever an analyst asked for a forward number.

Three exchanges are worth isolating, because the same pattern recurs in each.

On ARPU headroom, asked directly how a saturated market supports continued growth, the answer was about unmet user demand and services of value — no attach rate, no target, no measured differential.¹⁷

On AI and headcount, asked whether AI adoption would mean cost reduction or workforce downsizing, Chee Ching rejected the premise sharply: "If we now do so many headcount reductions, you can bet nobody is going to want to do this seriously."¹⁷ The point is a real one — if employees believe adopting AI tools costs them their jobs, adoption stops. What the company implemented instead was a hiring freeze achieved through productivity gains, with no layoffs disclosed.¹⁷ This is an unusually candid answer about the internal politics of AI adoption, and it is also a slower, less quantifiable path to margin than what peers are promising.

On capex credibility, the pushback on whether 2026's step-up was a new baseline drew the "it should come down" reply already discussed.¹⁷ Analysts were not fully satisfied, and by the second quarter of 2026 they were back on the same ground, asking about capex timing, 5G standalone requirements and ROI targets, and the economics of low-earth-orbit satellite partnerships.¹⁸

The second-quarter 2026 call added a new pressure point that had not appeared before: segment margin math. Analysts asked explicitly how ICT's 20–25% margins would affect group EBITDA of 35.8%, and how AI-related revenue should be tracked separately.¹⁸ This is the market beginning to price the mix shift — recognizing that a company growing its lower-margin segment fastest will report decelerating margins even while executing well.

Management's answer — that AI adoption services are "relatively high margin" because they leverage human capital rather than equipment — is directionally credible but, again, unquantified.¹⁸ There is no disclosed AI revenue line, no disclosed margin for it, no target.

What has been consistent across all four periods is genuinely notable, and it is the capital returns discipline. Through the price war, the auction, the merger and the current expansion, the dividend has been paid for eighteen consecutive years and the balance sheet has never been stretched.²¹⁸ Management's narrative about growth has been rewritten three times in eight years. Its behavior around capital has not changed at all.

For an investor, that is the most useful pattern in the transcripts: this is a management team whose promises about returning cash have been reliable and whose promises about future growth have been aspirational. Weight them accordingly.

↑ Back to Top

XI. Playbook: Durable Lessons and What to Watch

Step back from Taiwan and the transferable lessons are sharper than the company-specific ones.

The business lesson

In a mature, regulated oligopoly, the highest-return strategic move is usually not operational — it is structural.

Far EasTone spent two decades trying to win on execution: better networks, better service, competitive pricing, technology leadership. It achieved all of those at various points and none of them produced sustained excess returns, because a fifth operator with nothing to lose could always undercut whatever position it built. The single action that changed its economics was removing a competitor from the market.

That is an uncomfortable lesson, because it says operational excellence in a commoditized industry is table stakes rather than an advantage. It is also incomplete without its corollary: the merger only paid because of what management did after it. Buying 2.15 million subscribers and keeping all of them would have delivered a bigger company with the same bad economics. Deleting 881,073 of them was the decision that converted scale into value.¹⁵

The generalizable principle is that revenue quality beats revenue quantity in any business with high fixed costs and a service obligation attached to each customer. A customer who generates nothing still consumes billing systems, support capacity, regulatory reporting and fraud risk. Carrying them inflates the metric investors watch while degrading the metric that matters.

But note the precondition, because it is what makes this non-repeatable in most industries: this worked because market structure permitted rational pricing afterward. In a fragmented market, shedding low-value customers just hands them to a competitor who uses them to fund an attack on your good customers.

The investing lesson

A slow-growth, high-capex, regulated business is a capital allocation story, not a growth story, and it should be judged accordingly.

The relevant questions are not "how fast is revenue growing" but: Is the dividend covered and consistent? Does actual capex match guided capex? Are adjacent investments earning above the cost of capital or merely generating revenue? Is the balance sheet being preserved through the cycle or levered up at the top?

On the first and fourth of those, Far EasTone's record is strong. On the second, there is a live and checkable question. On the third, disclosure is insufficient to answer.

The three KPIs that matter

Everything above compresses into three numbers worth tracking. Not calculating — tracking, as they are reported.

1. Postpaid ARPU. This is the direct measure of whether the pricing discipline that emerged from consolidation is holding, and whether the essential-services strategy adds anything. The specific thing to watch is not the absolute level but the premium over Chunghwa Telecom. Far EasTone overtook the incumbent on blended ARPU in early 2025.¹⁵ If that lead widens, the differentiation thesis is working. If it converges, ARPU leadership was a one-time benefit of purging low-value accounts rather than a durable advantage.

2. Postpaid churn rate. The single clearest evidence of competitive strength, because it measures revealed customer behavior rather than management assertion, in a market where switching is nearly frictionless. At 0.7% in the first half of 2026 it is at a record low.¹ Sustained low churn while ARPU rises is the signature of real pricing power. Churn rising as ARPU rises would mean the company is simply pushing price until customers leave.

3. Smart ICT revenue and contract value growth. Track both, because they answer different questions. Revenue growth says whether the segment is converting; contract value growth says whether the pipeline is filling. First-half 2026 saw revenue up 23% and contract value up 103%.¹¹⁸ Contract value growing faster than revenue means the backlog is building. Should revenue growth persist while contract value growth stalls, the segment is drawing down its pipeline rather than replenishing it — an early warning that would appear roughly two to four quarters before revenue reflects it.

The bull case

A disciplined operator emerged from a decade of value-destructive competition with the best customer mix in its market. It holds industry-leading postpaid ARPU and record-low churn simultaneously, which is the observable signature of pricing power rather than a claim about it.¹¹⁵ With postpaid 5G penetration at 49.3% and 5G renewals carrying roughly 40% price uplifts, there is a mechanical, contracted pipeline of price increases across the remaining half of the base.¹¹⁸ The enterprise segment is growing at 23% with contract value doubling, giving genuine exposure to enterprise AI and cloud adoption without the capital risk of building data centers.¹⁸ The balance sheet carries under one turn of net debt, free cash flow covers a full-payout dividend that has risen for eighteen consecutive years, and a three-player market structure has ended the price-war era.²¹⁷¹⁸

The bear case

Mobile is structurally saturated, and once 5G penetration crosses 80–90% the upgrade-driven ARPU engine simply runs out — with no obvious successor technology cycle before 6G, which is years of capex away. The essential-services thesis, which is supposed to provide the next leg, remains entirely unquantified: no attach rates, no ARPU differential, no targets.¹⁷ The fastest-growing segment is the lowest-margin one, structurally diluting group profitability as it scales.¹⁸ Capital expenditure guidance has been missed to the downside in consecutive years and the 2026 number requires an extreme second-half ramp to be met.¹⁷¹⁸ The company operates inside a family conglomerate where minority holders have no path to influence and related-party complexity is permanent. And Southeast Asian smart city expansion — the most-touted optionality — depends on winning long-cycle government contracts in markets where Far EasTone has neither incumbency nor relationships.¹⁹

Risks that matter mechanically

Regulatory and political. The NCC has demonstrated willingness to intervene in retail pricing conduct.⁸ The Fair Trade Commission attached five-year consumer pricing obligations to the merger, including guaranteed discounted plans for elderly, low-income and disabled subscribers.¹³ Spectrum licenses expire and must be renewed or re-auctioned, and the state sets that price. This is a business whose largest single cost input and whose pricing latitude both sit partly in government hands.

Execution in ICT. Scaling from pilot contracts to production deployments, across borders, in a business with 20–25% margins and lumpy revenue recognition tied to project delivery schedules.¹⁸

Governance. Not an acute risk, but a permanent structural one, as described above.

What is not distinctly elevated relative to telecom peers is worth stating too, so the risk picture stays proportionate. Refinancing risk is minimal at 0.83 times net debt to EBITDA.¹⁸ There is no disclosed material cybersecurity incident or accounting restatement. Supply chain exposure to network equipment is real but shared industry-wide. Taiwan's geopolitical situation is a genuine tail risk for any Taiwanese asset, but it is not a Far EasTone-specific one and it is not something management action can address.

The risks that could actually break this investment case are mundane: ARPU converging back toward peers, ICT margin dilution outrunning ICT growth, and capex normalizing higher than guided.

↑ Back to Top

XII. Epilogue

In mid-2026, Far EasTone is a company doing almost everything an investor could reasonably ask a mature telecom operator to do.

Five consecutive quarters of record financial results. Twenty-three consecutive quarters of year-over-year revenue growth and fifteen consecutive quarters of rising net income.¹ A subscriber base that churns less than any competitor's while paying more. A balance sheet carrying less than one turn of leverage. A dividend that has risen every year for the better part of two decades. A market structure that its own dealmaking helped create, and that has stopped the bleeding of the price-war years.

The strategy going into the second half of 2026 is what the company summarizes as three amplifications — synergies, services, and AI — alongside 5G standalone network deployment and preparation for low-earth-orbit satellite services.²⁰ The 5G standalone build is the technical foundation for enterprise applications like network slicing, where a carrier can carve out a dedicated, guaranteed-performance slice of its network for a single customer — a factory, a hospital, a port. Management described it to analysts as "a technology evolution and a must-do on the way to be ready for 6G."¹⁸ That framing is honest in a way telecom capex justifications often are not: it is not primarily a revenue opportunity, it is the cost of remaining current.

Which leaves the question that actually determines the outcome for a long-term holder.

Far EasTone's operating edge is narrow and real: it charges the most and loses the fewest customers in its market. That combination is not rhetoric. It shows up in churn data and ARPU comparisons that anyone can check quarter by quarter. What is unproven is whether that edge compounds.

The optimistic path is that essential services — the insurance, the payments, the content, the connected devices — accumulate into switching costs deep enough that Far EasTone can keep raising prices faster than a saturated market should allow, while the enterprise ICT arm grows into a second profit engine large enough to matter. That is the case management makes, and the case its behavior over the past six years makes more credible than it would be from a less disciplined team.

The skeptical path is that a saturated three-player market caps the upside for everyone in it regardless of execution quality. On this view, Far EasTone's ARPU premium is the one-time yield of a cleanup, the 5G upgrade cycle is a wave that will crest and recede, and the ICT business — however fast it grows — is a lower-margin activity that dilutes the very economics it is supposed to save.

Both paths are consistent with everything currently disclosed. The evidence that will separate them is not a strategy announcement or a management assurance. It is three numbers, reported quarterly, over the next two to four years: whether ARPU keeps climbing, whether churn stays at the floor, and whether the enterprise backlog keeps filling faster than it empties.

Far EasTone spent twenty years learning that in a commodity business, you cannot talk your way to an advantage. Having finally built one, it now has to prove the advantage lasts.

↑ Back to Top

XIII. Outro & Links

Primary sources and further reading for anyone wanting to track this story quarter by quarter:

• FarEasTone Investor Relations — Annual Reports — the company's own annual reports, in English from 2001 onward, and the authoritative source for shareholder structure, board composition and segment disclosure.
• Market Observation Post System (MOPS) — Taiwan's mandatory filing portal, where material announcements, monthly revenue and financial statements are published before they reach the press.
• Taiwan Stock Exchange — listing data, disclosure rules and market statistics.
• FarEasTone Corporate — Company Information — the company's own milestone timeline.
• Far EasTone Q2 2026 earnings call transcript and Q4 2025 earnings call transcript — the Q&A sections are where the unresolved questions live.
• Taipei Times business archives — the most consistent English-language record of Taiwan's telecom regulatory history, from the 2018 fines through the 2023 merger approvals.

References

1. Far EasTone Q2 2026 slides: record results across all metrics — Investing.com, 2026 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

2. Far EasTone Q4 2025 Slides: Record Revenue, 100% Dividend Payout — Investing.com ↩↩↩↩↩↩↩↩

3. 遠傳電信股份有限公司 — 遠東集團 Far Eastern Group ↩↩↩

4. AT&T to help Far EasTone — Taipei Times, 2003-10-11 ↩

5. Douglas Hsu — Forbes Profile ↩↩↩↩↩

6. Company Information — FarEasTone Corporate ↩↩

7. New NT$499 data plan fuels price war — Taipei Times, 2018-05-10 ↩↩↩↩↩↩

8. Three telecoms fined for low-cost plan sale chaos — Taipei Times, 2018-05-24 ↩↩↩↩

9. Taiwanese carriers secure 5G spectrum following competitive auction — RCR Wireless, 2020-01-17 ↩↩↩↩

10. Far EasTone targets 11% stake in APT in unexpected deal — Taipei Times, 2020-09-06 ↩↩↩↩↩↩↩

11. Asia Pacific Telecom approves merger with Far EasTone — RCR Wireless, 2022-04-19 ↩↩↩↩

12. Far EasTone stalls on merger deal — Taipei Times, 2023-07-01 ↩↩↩↩↩

13. Fair Trade Commission declares approval of telecom merger — Taipei Times, 2023-07-21 ↩↩↩↩↩↩

14. FET and APT's merger takes effect tomorrow — Far Eastern Group news, 2023-12 ↩↩↩↩↩↩↩↩

15. Taiwan Mobile, Far EasTone see stellar results after respective mergers — DIGITIMES, 2025-07-02 ↩↩↩↩↩↩↩↩↩↩

16. 3大電信財報出爐：5G滲透率、ICT和OTT關鍵指標比一比 — 數位時代 BusinessNext ↩↩↩↩↩

17. Earnings call transcript: Far EasTone Q4 2025 reveals competitive market dynamics — Investing.com ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

18. Earnings call transcript: Far EasTone posts record Q2 2026 results, shares rise — Investing.com, 2026 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

19. Far EasTone eyes expansion of smart city services to Southeast Asian market — Taipei Times, 2026-03-18 ↩↩↩↩↩

20. 《通網股》遠傳Q2獲利創三高 下半年放大AI拚成長 — 中時新聞網, 2026-08-06 ↩↩↩

21. FarEasTone Investor Relations — Annual Reports ↩↩↩↩