T-Mobile bets its balance sheet on “no-trade-off” 5G – and investors lean in

A CEO debut framed by physics, not fashion

On a cold New York morning, Srini Gopalan used his first big outing as T-Mobile’s chief executive to mount an almost philosophical challenge to the US wireless industry. For decades, he argued, there was a “law of physics” that forced customers to choose: either pay up for the best network, or accept a weaker one in return for value and friendlier service. T-Mobile, he said, has broken that law.

That assertion – “no trade‑offs” between network quality, value and experience – was not just rhetoric. It underpinned a set of upgraded financial ambitions, a deliberate shift in how the company wants investors to measure it, and a conspicuously confident capital return plan.

Gopalan’s narrative rested on three pillars: widening network leadership, a structural value gap versus AT&T and Verizon, and an experience machine increasingly powered by AI and the T-Life app. Around that, finance chief Peter Osvaldik sketched a balance sheet designed to support both heavy network investment and a steady stream of cash back to shareholders.

Rewriting the yardstick: from lines to relationships

If there was a single symbolic break with industry habit, it came in how T-Mobile wants its performance to be judged. From the first quarter, it will stop reporting postpaid phone subscriber adds and phone ARPU – the metrics that have dominated US wireless commentary for years – and focus instead on postpaid accounts and ARPA.

The rationale is blunt. More than 90% of T-Mobile’s postpaid phone lines sit on a multi‑line account, and a “vast majority” of its 34‑plus million postpaid accounts already carry more than one product – from phones and tablets to home broadband. In that world, a single‑line phone metric can mislead.

Osvaldik pointed to a recent quarter where a key rival added more than 600,000 postpaid phones but only 26,000 postpaid accounts. T‑Mobile, by contrast, reported 261,000 postpaid net account additions in the fourth quarter – ten times the only peer that discloses that figure – and layered 2.7% ARPA growth on top, 3.6% on an organic basis once the lower‑revenue bases of MetroNet and US Cellular are stripped out.

In Gopalan’s framing, accounts are “relationships” – families and businesses – and account‑level metrics map more cleanly onto customer lifetime value and cash generation. Internally, he stressed, the entire field force is already incentivised on accounts, not lines.

For investors, the shift means the familiar phone-growth league tables will matter less. T‑Mobile offered a bridge for the transition: its 2026 guide for 900,000–1,000,000 postpaid net account adds implicitly assumes roughly 2.5 million postpaid phone net adds, in line with recent performance, alongside continued expansion in broadband and other devices.

Balance sheet as growth engine – and cash machine

Around that “p times q” story of accounts and ARPA, Osvaldik laid out a cash‑heavy model that depends heavily on operating leverage.

Service revenue is expected to reach roughly $77bn in 2026, up 8% year on year, with 6% organic growth once M&A contributions are excluded. By 2027, that rises to $80.5–81.5bn, a further 5% increase, with organic growth holding at 5%. At the top end, that implies more than $10bn of incremental service revenue between 2025 and 2027.

Core adjusted EBITDA is forecast at $37–37.5bn in 2026 and $40–41bn in 2027. On an organic basis, that translates into 7% and 8% growth respectively, faster than the 2025 pace. It also means, at the high end, more than $7bn of additional core EBITDA in two years – a drop‑through of roughly 70 cents of EBITDA for every new dollar of service revenue.

Those gains, in turn, are meant to sustain an unusually high free cash flow conversion. T‑Mobile generated free cash flow equivalent to 25% of service revenue in 2025 and 22% in the fourth quarter alone, while still spending heavily on its network. The company expects adjusted free cash flow to rise to $18–18.7bn in 2026 and $19.5–20.5bn in 2027, even after absorbing a lump of integration and restructuring costs.

The guidance is built on conservative plumbing. T‑Mobile assumes it will maintain leverage at 2.5 times EBITDA and fully draw the “strategic capacity envelope” that entails, even though consensus models tend to assume faster deleveraging. That prudence shows up in cash interest assumptions of $4.3bn in 2026 and $5bn in 2027 – around $500m and $1.1bn higher than street estimates, respectively – inflating the cost base in the guidance but leaving upside if the company chooses not to maximise debt.

Against that, there is a near‑term drag. T‑Mobile expects about $1.2bn of merger‑related costs in 2026, largely from the accelerated two‑year integration of US Cellular, plus roughly $450m of network optimisation charges and $150m of workforce restructuring. Together, those will consume about $2.5bn of cash in 2026, with integration outlays dropping to around $1bn in 2027 as synergies are harvested.

Even so, Osvaldik sketched an ample capital allocation envelope. Since mid‑2022, the company has spent roughly $12bn on strategic investments, including US Cellular, the MetroNet and Lumos fiber joint ventures, and ad-tech plays such as Vistar and Bliss. Over the same period it has returned more than $45bn to shareholders.

From here to the end of 2027, T‑Mobile expects to have more than $52bn of capacity to deploy. It currently envisages committing up to $30bn of that to shareholder returns, including approximately $10bn annually in buybacks, balanced by dividends. In a conspicuous gesture of conviction, the board has authorised up to $5bn of repurchases in the first quarter alone, doubling the previous run rate.

Majority shareholder Deutsche Telekom added its own exclamation mark, telling investors it does not plan to sell any T‑Mobile stock in 2026 and is actively exploring ways to increase its exposure.

That still leaves more than $22bn of “flexible” capacity, which could be used for further strategic investments – additional fiber platforms, spectrum purchases under the long‑delayed “one big beautiful bill” of US spectrum reform – or additional shareholder returns, depending on opportunity.

Network as unfair advantage – and future AI fabric

Gopalan’s central claim is that T‑Mobile has moved from being a price insurgent to the clear network incumbent in 5G, and that this leadership has room to widen.

The building blocks are familiar but potent. The company holds more spectrum than any rival, including its prized 2.5GHz “Goldilocks” mid‑band frequencies, which Gopalan said cover 70% more area than C‑band. The dense grid of towers built in the LTE era to squeeze performance from high‑band spectrum has become a strength in 5G, allowing high capacity and speed.

More subtle, but arguably more important, is the head start in 5G standalone core. T‑Mobile moved its core network to standalone 5G in 2021. It estimates competitors will not match that until 2025, giving it a three‑ to four‑year window to experiment with features such as network slicing, ultra‑reliable low‑latency modes and advanced traffic management before anyone else can deploy them at scale.

The result, Gopalan argued, is an “ultra capacity” network that is not only faster on benchmarks – median download speeds roughly twice its nearest rival, and on the latest iPhone around 85% faster than one competitor and nearly 50% faster than the other – but more resilient under load. He cited data showing that while fixed wireless access (FWA) customers grew 77% and usage per customer rose 27%, average speeds still increased 50%, and nearly doubled on the newest routers.

External recognition helps. After 35 consecutive J.D. Power reports over 17 years in which a perennial incumbent led on network quality, T‑Mobile now sits in the top slot. And critically, customer perception has followed: among “network switchers” – customers who research most deeply before moving carrier – the share that view T‑Mobile as the best network has doubled from one in eight in 2020 to roughly one in four today, closing the gap on Verizon.

The company is determined not to coast on that lead. Gopalan lingered on its “customer‑driven coverage” model, where AI and machine learning use anonymised location and usage data to guide capex. Rather than chase population‑coverage vanity metrics, T‑Mobile’s teams tune the network to where people “live, work and play” – including, in one anecdote, building out around Lake Tahoe to improve the experience for Sacramento subscribers who spend their weekends there.

The same model is now being used in reverse for opex, identifying towers and small cells that contribute little to customer experience and can be decommissioned or repurposed, freeing cash to reinvest in capacity hotspots. That network optimisation is one pillar of the $1.3bn and $2.7bn in annualised AI and digital savings that T‑Mobile expects by 2026 and 2027, respectively.

Beyond 5G, the company is already angling for 6G. In a pre‑recorded cameo, Nvidia chief Jensen Huang described the companies’ joint AI RAN Innovation Center and painted a future where every base station doubles as an AI computer, processing not only bits and bytes but “tokens” – the building block of generative AI – and serving as the “nervous system” for physical AI in cars, robots and cities.

John Saw, T‑Mobile’s long‑time network strategist, called AI‑powered RAN a chance to upend the classic telecoms compute architecture by embedding powerful, general‑purpose processors at the cell site. That, he suggested, could turn “fallow compute” at the network edge into a revenue‑producing asset, in much the same way T‑Mobile has used fallow RF capacity to build its FWA business. Field trials of the new architecture, he said, should begin in 2026.

That future is not in the numbers yet. Osvaldik was clear that the 2027 guidance assumes no meaningful contribution from AI RAN or edge‑AI monetisation.

Value as moat – and device subsidies in the crosshairs

If network performance is the “best” in T‑Mobile’s triad, value may be the most strategically important. Gopalan repeatedly stressed that existing customers pay 12–15% less on average than comparable AT&T and Verizon users, giving T‑Mobile a structural cushion as economic conditions ebb and flow. New customers, he claimed, often enjoy a 20–30% value gap when the bundled streaming and loyalty benefits are accounted for.

That relative price position is something he said the company will “zealously guard.” It makes T‑Mobile resilient in a downturn, offers flexibility on acquisition pricing, and underpins its “win‑win economics” mantra – deals that are good for customer wallets and for carrier CLV.

It also shapes how the company views the industry’s addiction to device promotions. Without naming rivals, Gopalan argued that US carriers have once again drifted away from win‑win logic, “over‑focused on how free the newest phone is” and “lost track” of the everyday value they deliver. Phones, as consumer chief Mike Katz put it, are a point‑in‑time purchase that happens “every couple, three years”; customers then live with the service every day.

T‑Mobile’s answer, hinted at more than explained, is a coming “Un‑carrier” move that will “change the course” on device economics and re‑centre the proposition on enduring value rather than ever‑freer hardware. The playbook, if past is guide, will likely pair some kind of simplification on subsidies with richer experiences and content baked into premium plans.

For now, the numbers already point in that direction. The company expects postpaid ARPA to grow 2.5–3% annually, driven primarily by front‑book mix shift into premium plans, life‑cycle upgrades and more products per account. Underneath that, it is budgeting for ARPU growth of 1–1.5%, reflecting both natural mix and occasional “rate plan optimisations” on legacy offers, such as the increase implemented in January.

In other words, while the company is de‑emphasising ARPU externally, it expects per‑line revenue to rise modestly even before taking into account the impact of more devices and services per relationship.

Experience as operating system – with T-Life at the centre

The third leg of the differentiation story is the customer experience, where T‑Mobile is trying to blend its cultural heritage – high‑NPS, high‑touch retail and care – with a digital, AI‑driven operating system.

The T-Life app is emerging as that OS. It has already notched more than 100 million downloads and, according to Gopalan, there were weeks last year in which it was the most downloaded app on both Apple’s and Google’s app stores – not just among carrier apps, but across all categories. Around 24 million primary account holders use it monthly, averaging four sessions per month.

That engagement is now translating into material shifts in how customers transact. In the fourth quarter of 2024, only 22% of device upgrades were conducted via T-Life, all of them with agent assistance. A year later, 73% of upgrades flowed through the app, with 39% unassisted. Over time, add‑a‑line transactions and even new‑account acquisitions are moving onto the same rails, assisted by features like Easy Switch, which allows customers to port in more seamlessly.

Osvaldik attributed a large chunk of the $3bn run‑rate savings expected by 2027 to this digitalisation, combined with AI tools such as “Intense CX,” developed in partnership with OpenAI, which arm frontline staff with real‑time prompts and personalised offers. Crucially, Gopalan and customer chief Jon Fryer both insisted the journey has been led from the customer side, not by blunt headcount targets. Store formats are being recast as “experience stores,” with more insourced, better‑trained staff, while simpler transactions migrate to self‑service.

The company is also starting to embed AI directly into the network layer for customer‑facing services. Gopalan unveiled “Live Translate,” which he described as the first at‑scale use of AI translation built into a carrier’s core network. The feature allows real‑time translation on voice calls with no separate app, so long as at least one caller is on T‑Mobile’s network.

Beyond its human interest appeal – six billion international calls traverse T‑Mobile’s systems each year, and more than 40% of its base travels abroad – Live Translate is a proof‑point for a new platform: a way to build multiple AI services native to the core network, available on any phone without app downloads. In an industry where most AI efforts to date have lived in chatbots and call centres, that is a quietly radical move.

All of this is reflected in NPS. In 2023, T‑Mobile sat roughly level with its two big rivals on net promoter scores. By 2025, the magenta bar had pulled away meaningfully in both top‑100 cities and small markets and rural areas. That widening NPS gap, Gopalan argued, is the “money slide” – more predictive of share gains, ARPA growth and free cash flow than any single promotion.

Broadband: fallow capacity to 19 million customers

If there is one market where all three elements – network, value, experience – intersect most obviously, it is broadband. From a standing start in 2021, T‑Mobile has built an FWA business with close to eight million customers and has consistently led the industry in broadband net adds, running at roughly two million a year.

The company’s ambitions are now larger. It raised its 2028 FWA target last year to 12 million customers; this week, it set a 2030 goal of 15 million FWA subscribers, plus three to four million fiber customers via its T‑Fiber joint ventures with MetroNet and Lumos. That would give T‑Mobile 18–19 million broadband subscribers within seven years – most of them incremental, given it does not own copper or legacy cable to cannibalise.

Underneath the headline lies a careful engineering model. T‑Mobile manages FWA using what Gopalan calls “fallow capacity.” At the level of 30 million tiny “hex bins” across the US, the company forecasts peak‑hour mobile usage for its core wireless customers and reserves enough RF capacity to support growth. Only what is left is made available for FWA. The 15 million target, he emphasised, assumes this conservative discipline is maintained; it does not rely on additional spectrum purchases, nor on yet‑to‑emerge 6G spectral efficiencies.

What it does factor in are the effects of better routers, 5G Advanced features such as L4S (which improve latency and throughput on the existing spectrum), and smarter product mix. In particular, selling more FWA into businesses – which typically consume far less between 7pm and 9pm, the peak household usage window – creates more genuinely “fallow” overnight capacity.

The economics look attractive. T‑Mobile says its FWA NPS is higher than that for fiber, a function of simple installation, transparent pricing and the performance of its 5G network. It is also increasingly confident in its ability to expand beyond urban and suburban cores into smaller markets and rural areas, where it has already lifted mobile market share from 13% in 2020 to 24%, yet still sees itself as under‑indexed among “network seekers.”

Fiber, meanwhile, remains a carefully hedged bet. All of the 12–15 million homes passed and three to four million fiber customers projected for 2030 come from existing joint ventures with MetroNet and Lumos. Gopalan made clear that T‑Mobile is open to additional fiber platforms, but only “at the right price.” He explicitly rejected the idea of being forced into overpaying for passings simply to hit an arbitrary homes‑passed milestone.

For investors, the important point is that the broadband story is deeply intertwined with wireless. FWA and fiber are both significant contributors to ARPA growth; they are also vectors into new households where T‑Mobile can cross‑sell mobile lines, leveraging the same brand, distribution and T‑Life app.

New businesses: optionality, not dependence

Beyond connectivity, T‑Mobile is pragmatic about diversification. The company evaluates new ventures through three tests: is the addressable market large enough to matter; can its existing strengths in network, data, distribution and customer relationships genuinely differentiate; and can it disrupt, rather than simply replicate, incumbents.

On that basis, three areas stand out: advertising, financial services and physical/edge AI.

The advertising arm, bolstered by the acquisitions of Vistar and Bliss, is tracking the trajectory laid out at last year’s capital markets day. It sits at the intersection of first‑party data from tens of millions of subscribers, location insights from the network, and a rapidly growing T‑Life user base – a rich, if sensitive, set of assets in a world where third‑party cookies are fading.

Financial services is newer but already tangible. In November, T‑Mobile launched its first co‑branded credit card with Capital One. Early results, Osvaldik said, have been “tremendously successful.” The underlying thesis is that the company’s customer base and credit decisioning capabilities – honed by a decade of serving sub‑prime and near‑prime wireless customers while still delivering lower bad‑debt ratios than AT&T or Verizon – give it an edge.

Just as importantly, the cost of customer acquisition in financial services is often prohibitive. With 24 million monthly active T‑Life users and deep NPS‑driven trust, T‑Mobile believes it can take promising ideas and bring them to market with a fraction of the CAC others must pay. That same logic applies to other products that may be layered into T‑Life over time.

None of this is in the 2027 guide in any meaningful way. That gives investors a clean view of the core business while leaving upside if these options sharpen.

Competition, churn and the macro shadow

All of this is happening in a market that has rediscovered its appetite for promotions. The final quarter of 2025 was intensely competitive, with aggressive iPhone offers and richer trade‑ins across the board. Industry churn, suppressed for years by 36‑month device contracts and pandemic lethargy, is normalising.

T‑Mobile’s framing is that not all churn is equal. Account‑level churn – when an entire family or business leaves – destroys value. The loss of a dormant, single‑line device on an otherwise healthy account does not. That is why it will publish account churn, not line churn, from here.

On that measure, Gopalan argued, the company is holding up better than peers. In full‑year 2025, its churn increase was seven basis points, the lowest of the three carriers. Even in a volatile December, he said, the average value of incoming ports was about 15% higher than those leaving, reinforcing the idea that the company is trading up the quality of its base.

The macro economy lurks in the background, particularly at the lower end of the “K‑shaped” income distribution. Both Gopalan and Osvaldik stressed that connectivity is a resilient category – “never the canary in a coal mine” – but nonetheless lean heavily on T‑Mobile’s structural price advantage for comfort. If the environment turns more value‑sensitive, they argued, the carrier that already offers the best network and the lowest bills is best placed to capture switching.

That said, investors should watch the interplay between device financing, promotions and bad debt. T‑Mobile’s guidance assumes contract assets – the accounting embodiment of handset subsidies financed over time – remain broadly flat, in contrast to some rivals whose balance sheets are shouldering more of the discount. That discipline aligns with the company’s rhetoric on win‑win economics; it could also cushion credit risk if the cycle turns.

A widening gap – and a higher bar

Strip away the magenta‑hued stagecraft and the Seahawks jokes, and the core message of T‑Mobile’s New York gathering was simple: the gap is widening. In network benchmarks, in NPS, in service revenue and EBITDA growth, in free cash flow margins and – perhaps most visibly – in capital return capacity, the company believes it has pulled clear of AT&T and Verizon.

The targets it has set for 2026 and 2027 are designed to prove that this is not a one‑off. Service revenue growing 5–6% organically in a market with flat to low‑single‑digit line growth; EBITDA expanding faster than revenue; free cash flow compounding from an already high base; and tens of billions of surplus cash deployed with what management insists will be “disciplined, consistent” capital allocation.

Investors now have a clearer set of levers to watch: postpaid account growth, ARPA expansion, broadband uptake and the translation of T‑Life engagement into both savings and cross‑sell. They also have a different bar to hold the industry to. If T‑Mobile succeeds in making accounts and ARPA the new currency of wireless reporting, the carrier that wins whole‑of‑household relationships and deepens them over time will be the one that sets the tone.

For now, the company is betting that its self‑styled “no trade‑off” position in network, value and experience – amplified by AI and backed by a balance sheet geared to invest and return in equal measure – will keep the tide running in its favour. The next downturn, the next spectrum auction and the next round of device innovation will all test that thesis. But in New York this week, at least, the physics looked to be bending T‑Mobile’s way.