TMUS · T-Mobile US, Inc.

conf 6/10 partial

enthusiasm:27.0 · trend:0 · quantifies:12 · impact:0 · under_radar:0 · credibility:5 · business_impact:8 · disruption:0 · commitment:0 · confirmation:3

Enthusiasm latest 9 / prev 10 (flat)

T-Mobile’s AI thesis is that AI is both an operating-efficiency engine and a network-product differentiator, with management tying it to customer care automation, smarter network capital deployment, live translation, AI-RAN, and edge inference for physical AI. Credibility is strongest on cost savings and chatbot containment, where management gives figures; revenue upside from physical AI and edge inference remains unquantified and explicitly early.

Grounded on actual base — revenue $88.3B · net income $11.0B · net margin 12.4% · diluted EPS 9.72

These are next-fiscal-year annual uplift estimates, not next-quarter numbers.

Aggregate next-FY est. rev uplift: 0.0% · next-FY EPS uplift: 9.34% · vs analysts: ahead · priced in: high (model's call-read: low; verdict above is the hard-data one used for ranking) · confidence: 6/10

Assumptions: Tax rate 21% on pretax opex savings; savings flow ~100% to bottom line, topline ~0 (not capacity-constrained). Next fiscal year = FY2026 (FY25 closed 2025-12-31). Headline aggregate uses ONLY the 2026 figure ($1.3B) to avoid double-counting the 2027/run-rate restatements ($2.7B/~$3B) of the same program. No quantified AI revenue claim provided; current net margin 12.45% noted but unused (no incremental revenue). Shares ~1,131.076M, so NI-% ≈ EPS-%.

Top line: Negligible from quantified AI claims. Every dollar-anchored item is a cost saving (bottom-line); the only topline AI features (80-language live translation, chatbot, edge compute) carry no disclosed take-rate or revenue figure, so they are soft/null. Supplier-side AI ambitions (edge inferencing for 'physical AI', 'fallow compute' monetization, compute in the network core) are real but entirely unquantified — supplier_rev_uplift_pct = null. Net: ~0% revenue uplift attributable to quantified adopter-side AI.

Bottom line: This is where the AI story lands. The hard, management-quantified next-FY (2026) saving is $1.3B pretax → $1.027B after-tax at 21% → +9.34% of FY25 net income of $10.992B (≈ +$0.91 on $9.72 diluted EPS). The program scales to $2.7B in 2027 (+19.4% after-tax) / ~$3B run-rate by end-2027 (+21.6%), but those are restatements of the same initiative in a later year and are excluded from the next-FY aggregate to avoid double-counting. Management frames this as efficiency/digitalization, not slash-and-burn cuts.

On next-FY math the savings point above consensus: consensus 2026 NI of $11.316B is only +2.95% over FY25 actual $10.992B, while the $1.3B 2026 saving alone is +$1.027B after-tax (+9.34%), larger than the entire modeled NI increase — so if truly incremental it is not fully credited. Consensus 2026 EPS $10.441 (+7.4%) leans on buybacks rather than NI. The gap closes by 2027: consensus NI $14.149B (+28.7%) and EPS $13.571 (+39.6%) comfortably embed the ~$2.7B run-rate, so the back half is priced in. The likeliest reconciliation is that 2026 savings are partly offset by competitive/reinvestment pressure or only partially baked — leaving a modest, not large, edge on the 2026 step.

• We use AI and huge amounts of customer data to deploy capital in our network based on what's right for customers rather than chasing a vanity stack like POPS.
• And using AI, using scale machine learning, all of our site deployment.
• All of our site deployment is surgically planned to improve our customer experience.
• We've already started introducing large amounts of AI into our network.
• And as we move closer towards AI RAN, in fact, even during things like Winter Storm Fern, you saw AI in our network being a big reason why things like antenna tilt being done automatically, things like optimizing our network, a self-healing network in many ways is not kind of science fiction.
• It's reality. It's the way our network runs every day.

• And with IntentCX which is AI that we've developed, working really closely with OpenAI, where the objective is simple, it is to personalize the experience.
• We're now taking the best technology digital, AI, putting that in the hands of our incredible frontline to drive an even sharper and even more differentiated experience.
• We're excited to be rolling out live translation on beta soon, our first network native AI application that we demoed for you at our February event.
• As a step towards this, we're delighted to share today that we're connecting our 5G advanced network to Figure AI's F03 humanoid robots, enabling seamless and reliable connectivity from the moment they power on.
• And one, for example, is just the use of the chatbot, an AI-powered chatbot that is actually capturing a lot of customer questions and addressing them in a great Un-carrier fashion that you'd expect and actually containing about 60% of those already.

• Importantly, this is just the initial step in us building AI capabilities directly into our network core.
• Longer term, we see a world where our network becomes the connective tissue for physical AI and accommodates inferencing at the edge.
• This partnership, amongst others, will allow us to explore how the T-Mobile 5G advanced network and its capabilities, including assets like the network edge can support the broader evolution of physical AI.
• Now as we do more and more AI in our network and as we build for more and more AI in our network, we will be building compute into our network.
• As we build more AI into our network, we will generate a bunch of fallow compute, especially at the edge.
• And now that we have a 5G advanced network we can take on the extra capabilities that is needed to support edge inferencing for physical AI better than anybody else.
• And we believe that we have a multiyear advantage over the competition for this.
• But across our AI and digital initiatives, we expect close to $3 billion in savings by the end of '27 in our '27 run rate, which is incredible because this has not been a slash and burn, let's take out X100 people.
• We now expect between 2026 and '27 -- relative to 2025, these initiatives are going to deliver $1.3 billion of incremental savings in 2026.
• And $2.7 billion in 2027.

65/100 track record   delivers  5 calls reviewed

T-Mobile's quantified AI bets are mostly intelligent-automation CX and AI-driven network-planning goals, and the judgeable ones track well: digital-upgrade adoption beat its own in-call targets and AI-planned site deployment hit plan, while call-deflection (~50% toward a 75% goal) and multiyear targets are still in-flight. A credible, delivery-leaning record, though several headline targets originate from a Capital Markets Day outside this transcript set.

Est. revisions rising  ·  Fwd P/E 19.1  ·  EV/Sales 3.5x

AI claim maps to Branded Postpaid Revenue, Branded Prepaid Revenue, Wholesale Service Revenue

AI enthusiasm across 5 calls — trend ↗ rising

AI moved from customer-care automation to concrete network planning, native translation, edge inference, and physical AI connectivity ambitions.

• 2026-05-26 · T-Mobile US, Inc. to Present at the Evercore TMT Global Conference
• 2026-05-05 · Every Second Counts: T-Mobile Advances a Stronger, Smarter, More Resilient Network for Americans
• 2026-04-29 · Oppenheimer Upgrades T-Mobile to Outperform With a $260 Price Target: Can AI Pricing Power Drive the Next Leg?
• 2026-04-28 · The Big 3: TMUS, CRDO, MRVL
• 2026-04-24 · 3 Telecom Stocks Likely to Surpass Q1 Earnings Estimates

8/10 qualitative impact   material  near-term · mixed evidence

Where AI matters: customer care automation and network optimization

AI is already tied to meaningful, quantified cost savings: $1.3B incremental savings in 2026 and about $2.7B in 2027, with chatbot containment and AI-driven network planning as concrete operating levers. The revenue story around live translation, AI-native network features, edge inference, and physical AI connectivity is strategically interesting but still unpriced and unsized.

Caveats: Revenue upside from edge inference and physical AI is early and explicitly too early to size; Savings may be partly offset by reinvestment, competitive pricing, or customer-experience spend; AI-native services like translation may prove feature parity rather than monetizable differentiation; Execution risk in embedding AI into network core and AI-RAN without excess capex

AI DISRUPTION / CANNIBALIZATION RISK  tailwind · 1/10

AI does not plausibly automate away the need for mobile connectivity, spectrum, towers, or low-latency network infrastructure; it mostly raises the value of resilient networks and lowers service costs. Any risk is competitive execution or cloud/edge economics, not cannibalization of the core subscription model.

option liquidity: good

proxy inputs — dollar-ADV $1.1B · beta 0.32 · px $188.83

source: proxy (no options chain on FMP)
FMP /stable/ exposes no options-chain endpoint on this key, so ATM IV, bid-ask spread and open interest are unavailable. Liquidity below is a PROXY from dollar-ADV, beta and price level (a stand-in for option depth), not measured option-market data.