T-Mobile US Inc (TMUS) 2026 Q3 法說會逐字稿

Good afternoon. (Operator Instructions) I would now like to turn the conference over to Cathy Yao, Senior Vice President of Investor Relations for T-Mobile US. Please go ahead.

下午好。(接線員指示) 我現在要把電話會議交給 T-Mobile US 投資人關係資深副總裁 Cathy Yao。請開始。

Good afternoon. Welcome to T-Mobile's third quarter 2026 earnings call. Joining me on our call today are Srini Gopalan, Peter Osvaldik, Jessica Uhl, as well as other members of the leadership team.

下午好。歡迎參加 T-Mobile 2026 年第三季財報電話會議。今天與我一同出席的有 Srini Gopalan、Peter Osvaldik、Jessica Uhl，以及領導團隊的其他成員。

During this call, we will make forward-looking statements, which involve risks and uncertainties that may cause actual results to differ materially. We encourage you to review the risk factors set forth in our SEC filings. Our earnings release, Investor Factbook and other documents related to our results as well as reconciliations between GAAP and non-GAAP results discussed on this call can be found on our Investor Relations website.

在本次電話會議中，我們將發表前瞻性陳述，其中包含風險與不確定性，可能導致實際結果出現重大差異。我們鼓勵各位查閱我們向 SEC 提交文件中所列示的風險因素。我們的財報新聞稿、投資人資料手冊（Investor Factbook）以及其他與本次業績相關的文件，連同本次電話會議所討論之 GAAP 與非 GAAP 結果之調節表，均可於我們的投資人關係網站查閱。

With that, let me now turn it over to Srini.

那麼，接下來我把時間交給 Srini。

Thanks, Cathy, and good afternoon, everyone. We're here in Bellevue today, very much looking forward to discussing our results. But first, I'd like to welcome two new faces at the table. Jessica Uhl, CFO-designate, and Chris Sambar, Chief Enterprise Officer. Jessica and Chris bring tremendous talent and experience to this leadership team, and I'm thrilled to have them join us, and I'm looking forward to all of you getting to know them very well.

謝謝你，Cathy，各位下午好。我們今天在貝爾維尤（Bellevue），非常期待與各位討論我們的業績。不過首先，我想歡迎兩位新面孔加入我們的桌邊團隊。Jessica Uhl，候任財務長（CFO-designate），以及 Chris Sambar，企業事業長（Chief Enterprise Officer）。Jessica 與 Chris 為這個領導團隊帶來卓越的人才與豐富的經驗，我非常高興他們加入，也期待各位能更深入地認識他們。

Turning to Q3. This was another great quarter of executing on our strategy, and it reflects our ability to optimize both volume and value to deliver industry-leading results. In Q3, we once again led the industry in postpaid phone net additions and 5G broadband net additions, both of which helped translate to strong postpaid net account additions of 274,000.

接著談第三季。這又是一個我們成功落實策略的優異季度，展現了我們同時最佳化規模（volume）與價值（value）、並交出業界領先成果的能力。在第三季，我們再次在後付費手機淨增數以及 5G 寬頻淨增數上領先同業，並帶動後付費淨新增帳戶數達到 274,000。

Alongside this, importantly, we saw over 4% postpaid ARPA growth year-over-year. We also saw 5G broadband ARPU grow year-over-year and sequentially in Q3. These strong unit results translated to industry-leading financial growth.

同時，重要的是，我們看到後付費 ARPA 年增超過 4%。我們也看到 5G 寬頻 ARPU 在第三季不僅年增，且較前一季成長。這些強勁的單位（unit）表現轉化為業界領先的財務成長。

Postpaid service revenue grew 9%, total service revenue 6%, core adjusted EBITDA 11% and we reported a really strong adjusted free cash flow margin of 26%. Now what underlies all of this is our unique proposition of best value, best network, best experience. So let me double-click into that. From a value standpoint, this is something we continue to guard zealously.

後付費服務營收成長 9%，總服務營收成長 6%，核心調整後 EBITDA 成長 11%，而我們也繳出非常強勁的調整後自由現金流利潤率 26%。支撐這一切的，是我們獨特的主張：最佳價值、最佳網路、最佳體驗。所以讓我更深入說明一下。就價值而言，這是我們持續嚴格守護的核心。

We are maintaining our unique back book to front book pricing advantage, which is that on average, our existing customers actually pay less than our new customers. We do not, however, take our best value position for granted. Earlier this week, we launched our T-Mix portfolio, the latest example of disruption by solving customer pain points.

我們維持了獨特的「舊方案（back book）對新方案（front book）」定價優勢，也就是平均而言，我們的既有客戶實際支付的費用比新客戶更低。不過，我們並不會把「最佳價值」的地位視為理所當然。本週稍早，我們推出了 T-Mix 產品組合，這是透過解決客戶痛點來實現顛覆的最新例子。

For far too long, carriers have chosen how customers get value. 84% of surveyed customers want more control of how their wireless plans are structured. And T-Mix is smashing through the pain point by allowing customers to choose what benefits they want and allowing them to change them when their needs change.

長久以來，電信業者一直由業者決定客戶如何獲得價值。受訪客戶中有 84% 希望能更有掌控權，決定自己的無線方案如何被設計。而 T-Mix 透過讓客戶自行選擇想要的權益，並在需求改變時可隨時調整，直接打破這個痛點。

On customer experience, we're leveraging our amazing team empowered with digital and AI tools to deliver a best-in-class experience. Last quarter, we shared with you that the customer excitement around Member Month drove us to over 30 million monthly active users of T-Life. That was in June. That momentum continued throughout Q3, and T-Life now has 32 million monthly active users.

在客戶體驗方面，我們運用一支優秀的團隊，並賦能其使用數位與 AI 工具，以提供同級最佳的體驗。上個季度我們與各位分享，會員月（Member Month）帶來的客戶熱情，推動 T-Life 的月活躍用戶數超過 3,000 萬。那是在 6 月。這股動能在第三季持續延伸，T-Life 現在的月活躍用戶數已達 3,200 萬。

As a result, more and more of our transactions are being done directly in the app ecosystem. In Q3, over 80% of our upgrades and half of our phone line activations were done digitally between customers using the app on their own or being guided through it by our associates.

因此，越來越多交易直接在 App 生態系中完成。在第三季，我們超過 80% 的升級（upgrade）以及一半的門號開通（phone line activation）都是以數位方式完成：要嘛由客戶自行在 App 操作，要嘛由我們的門市/客服同仁引導客戶透過 App 完成。

Now underpinning both value and experience is, of course, our network, which positions us for outsized growth for the next decade and beyond. We have the industry's best network both through the lens of real, tangible, lived customer experiences as well as recognized through third-party awards.

而支撐價值與體驗的，當然是我們的網路；它讓我們在未來十年乃至更長時間具備超額成長的定位。無論從真實、可感知、貼近客戶日常的使用體驗來看，或是從第三方獎項的肯定來看，我們都擁有業界最佳網路。

Our overall median speeds are approximately 60% to 70% faster than our two largest competitors based on Ookla data. And the awards, they just keep coming. Among many this quarter, Opensignal awarded us the global winner in 5G coverage. In fact, T-Mobile has been named an Opensignal 5G global winner every year the program has existed. And no other US operator has ever been named a 5G global winner.

根據 Ookla 數據，我們整體的中位數速度約比兩大競爭對手快 60% 到 70%。而獎項也持續不斷。在本季眾多獎項中，Opensignal 授予我們 5G 覆蓋的全球冠軍。事實上，自該計畫設立以來，T-Mobile 每一年都被評為 Opensignal 5G 全球冠軍。而且沒有任何其他美國電信業者曾被評為 5G 全球冠軍。

Our margin of network leadership will continue to widen through our unreplicable spectrum holdings and our focus on leadership and innovation. Talking of innovation, we completed the world's first AI-RAN call here in Bellevue, not in the lab, but out there in the real world using our network. We also continue to see more proof points in the physical AI opportunity where we've progressed beyond connecting humanoid robots with eSIMs and are now testing how the network helps deliver the intelligence itself.

憑藉我們無可複製的頻譜資產，以及對領先與創新的專注，我們在網路領先的差距將持續擴大。談到創新，我們在貝爾維尤完成了全球第一通 AI-RAN 通話——不是在實驗室，而是在真實世界中、使用我們的網路完成。我們也在「實體 AI（physical AI）」機會上看到更多驗證：我們已不僅是用 eSIM 連接人形機器人，現在更進一步測試網路如何協助把「智慧」本身傳遞到位。

To enable all of these use cases, we're partnering with Amazon Web Services, AWS, to build the first instances of our Kinetic Edge. Now Kinetic Edge is a distributed intelligence infrastructure for physical AI. It combines our nationwide 5G Advanced network and distributed edge with AWS's cloud orchestration and AI services. This puts together deterministic connectivity, compute, and intelligence closer to the machines and gives enterprises exactly what they need to take physical AI from isolated demos to production at scale.

為了支援所有這些使用情境，我們正與 Amazon Web Services（AWS）合作，建置我們 Kinetic Edge 的首批實例。Kinetic Edge 是一套用於實體 AI 的分散式智慧基礎設施，結合我們全國性的 5G Advanced 網路與分散式邊緣能力，以及 AWS 的雲端編排與 AI 服務。這將確定性連線、運算與智慧更靠近機器端，讓企業獲得把實體 AI 從零星展示推進到可規模化量產部署所需的一切。

Let me now turn to how we're monetizing our unique capability to deliver the best network, best value and best experience. In wireless, the consistent improvement in our network perception year-over-year and our heritage of value and experience leadership allowed us to once again grow our share of postpaid households across the top 100 and in smaller markets and rural areas.

接下來談談我們如何將「最佳網路、最佳價值、最佳體驗」這項獨特能力變現。在無線業務方面，我們的網路口碑（network perception）逐年持續改善，加上我們在價值與體驗領先的傳統，使我們再次提升在前 100 大市場、以及較小市場與鄉村地區的後付費家庭市占。

T-Mobile for Business now under Chris's stewardship continues to flourish as we bring customers technology solutions powered by America's best network. Q3 marked the 17 straight quarter of positive ports, contributing to healthy double-digit revenue growth year-over-year. And in the quarter, JD Power named us the most awarded business for business wireless customer satisfaction as well as number one in both large enterprise and small business customer satisfaction. This is an incredible accomplishment and a testament to the power of our network, our brand and our value.

在 Chris 的帶領下，T-Mobile for Business 持續蓬勃發展，我們以「美國最佳網路」為基礎，為客戶提供科技解決方案。第三季是我們連續第 17 個季度攜碼轉入（ports）為正，帶動年增健康的雙位數營收成長。此外在本季，JD Power 將我們評為企業無線客戶滿意度「獲獎最多」的業者，並在大型企業與小型企業客戶滿意度兩項中皆名列第一。這是令人難以置信的成就，也證明了我們網路、品牌與價值的力量。

Governments and enterprises today are not only rigorously testing networks, but also focusing on how new network capabilities can solve their business needs before they make the buying decision. We recently won the top spot in the Department of Justice's wireless purchasing contract. As one of the largest public safety agencies in America, the DOJ is an incredibly discerning buyer.

如今，政府與企業不僅會嚴格測試網路，也會在做出採購決策前，聚焦於新網路能力如何解決其業務需求。我們近期在司法部（Department of Justice, DOJ）的無線採購合約中奪得第一名。作為美國最大的公共安全機構之一，DOJ 是極其挑剔的買方。

They've spent hours of testing hundreds of hours of testing and evaluating our network against others before they chose the best public network -- public safety network in America. This is again proof that we're growing our TFB presence with large and influential customers. Broadband, another key component of our growth portfolio. Our 5G broadband offers near fiber-like download speeds over Wi-Fi, seamless setup and an award-winning customer experience.

他們投入了數小時的測試——實際上是數百小時的測試——並在選定美國最佳公共網路——公共安全網路之前，將我們的網路與其他業者進行評估比較。這再次證明，我們正以大型且具影響力的客戶為切入點，擴大 TFB 的市場版圖。寬頻也是我們成長產品組合中的另一個關鍵構成。我們的 5G 寬頻透過 Wi‑Fi 提供近似光纖的下載速度、無縫的安裝設定，以及屢獲獎項的客戶體驗。

Ookla once again ranked us as the fastest 5G broadband provider for both upload and download speeds. It's no wonder our 5G broadband customers continue to have the highest NPS of any broadband connectivity category, and that includes fiber. We're tracking well ahead of the pace we set for you to reach 15 million customers by the end of the decade.

Ookla 再次將我們評為 5G 寬頻上傳與下載速度最快的供應商。難怪我們的 5G 寬頻客戶持續擁有所有寬頻連線類別中最高的 NPS，而這也包含光纖在內。我們的進度追蹤顯示，已明顯領先於我們先前向各位設定的步調，朝著在本十年末達到 1,500 萬客戶的目標邁進。

Importantly, in addition, our phone-like ARPUs in FWA have grown on a year-over-year and sequential basis. Our fiber business is also scaling well with approximately 20% 12-month penetration and a sequential and year-over-year acceleration in customer growth. We're well on track towards our target of 3 million to 4 million customers by 2030.

重要的是，此外，我們在 FWA 的類手機 ARPU 也在年增與季增的基礎上持續成長。我們的光纖業務同樣擴張良好，12 個月滲透率約 20%，且客戶成長在季比與年比上皆呈現加速。我們正穩步朝著 2030 年達到 300 萬至 400 萬客戶的目標前進。

To conclude, the environment we're operating in has been dynamic for quite some time. But through it all, this team right here is laser-focused on delivery, execution and controlling our controllables. We're incredibly confident in our differentiation consistently leading to outsized growth today and in the future.

最後總結，我們所處的營運環境在相當長一段時間以來一直充滿變動。但在這一切之中，眼前這個團隊始終高度聚焦於交付、執行，以及掌控我們能掌控的事項。我們對於自身差異化能在當下與未來持續帶來超額成長，抱持極高信心。

Peter, over to you.

Peter，交給你了。

All right. Thank you, Srini. First of all, I'd also love to welcome Jessica to the T-Mobile team. The depth and breadth of Jess' experience, not only as CFO but also importantly, as a strategic business leader at large complex organizations makes her a tremendous win for T-Mobile. Jessica and I will continue to transition over the next few months as part of this long-planned succession process.

好的。謝謝你，Srini。首先，我也想歡迎 Jessica 加入 T-Mobile 團隊。Jess 的經驗深度與廣度，不僅是擔任 CFO 的經驗，更重要的是她在大型、複雜組織中作為策略型業務領導者的歷練，對 T-Mobile 來說是一大收穫。作為這項長期規劃的接班流程的一部分，Jessica 和我將在接下來幾個月持續進行交接。

All right. Turning to our outlook. We delivered a fabulous Q3, which reinforces our strong guidance for the remainder of the year, so let me add some additional color. Starting with accounts, we now expect approximately 1 million account net additions for the year once again, far exceeding the industry on real value-creating share-taking.

好的。接著談談我們的展望。我們交出了一個非常出色的第三季表現，這也強化了我們對今年剩餘期間的強勁指引，因此我再補充一些細節。先從帳戶數來看，我們現在再次預期全年帳戶淨新增約 100 萬戶，遠遠超越產業，真正以創造價值的方式取得市占。

Turning to service revenues. We now expect full year service revenues of approximately $76.8 billion, representing 8% growth for the full year, driven by ongoing outperformance of the postpaid business, partially offset by declines in the wholesale and other line, including from the faster roll-off of the Boost MVNO as well as the ongoing trends in prepaid.

再看服務營收。我們目前預期全年服務營收約 768 億美元，全年成長 8%，主要由後付費業務持續優於預期所帶動；但部分被批發與其他線路的下滑所抵銷，其中包括 Boost MVNO 更快的退租/流失，以及預付費業務的持續趨勢。

As part of the postpaid outperformance, we now expect full year postpaid ARPA growth of approximately 3.5%, up from our previous expectation of 2.5% to 3%, driven by our continued success in customer adoption of premium rate plans, deepening relationships with customers as well as our rate plan modernization initiatives. This drives the all-important postpaid business to continue to excel in growth, with Q4 postpaid organic service revenue growth now expected to be approximately 7% year-over-year.

作為後付費優於預期的一部分，我們現在預期全年後付費 ARPA 成長約 3.5%，高於先前預期的 2.5% 至 3%；這主要來自我們持續成功推動客戶採用高階資費方案、加深與客戶的關係，以及資費方案現代化（modernization）計畫。這也帶動至關重要的後付費業務持續在成長面表現出色，目前預期第四季後付費有機服務營收年增約 7%。

Core adjusted EBITDA is now expected to be between $37.35 billion and $37.45 billion for the full year, up $100 million at the midpoint from our previous guide. Our expectation for full year 2026 cash CapEx remains unchanged at approximately $10 billion. And finally, we now expect adjusted free cash flow to be between $18.7 billion and $18.8 billion, up $150 million at the midpoint from our previous guidance. So to sum it all up, the differentiation in the business continues to deliver industry-leading results, and this entire team cannot be more excited for the future.

核心調整後 EBITDA 目前預期全年介於 373.5 億至 374.5 億美元之間，中位數較先前指引上調 1 億美元。我們對 2026 年全年現金資本支出（cash CapEx）的預期維持不變，約 100 億美元。最後，我們現在預期調整後自由現金流介於 187 億至 188 億美元之間，中位數較先前指引上調 1.5 億美元。總結而言，業務的差異化持續帶來產業領先的成果，而整個團隊對未來感到無比振奮。

With that, I'll now turn it over back to Cathy to begin the Q&A. Cathy?

接下來，我把時間交回 Cathy，開始問答環節。Cathy？

Thanks, Peter. Okay, let's get to your questions. You can ask questions via phone by pressing star and then one via X by sending a post to @TMobileIR or @Srinigopalan using #TMUS. We will start with a question on the phone. Operator, first question, please.

謝謝你，Peter。好的，我們開始回答各位的問題。你可以透過電話按星號再按 1 提問；也可以透過 X 發文到 @TMobileIR 或 @Srinigopalan，並使用 #TMUS 來提問。我們先從電話提問開始。接線員，請進第一個問題。

Craig Moffett, Moffett Nathanson.

Craig Moffett，Moffett Nathanson。

I wonder if we could start with FWA. I know we're all sort of left a little bit in the dark without subscriber metrics. But we saw a pretty dramatic deceleration from Verizon, a little bit of acceleration from AT&T, and we're just trying to get a sense of what's going on with FWA for you specifically and for the industry more broadly. So I wonder if you could just comment on that a bit.

我想我們可以先從 FWA 談起。我知道在缺少用戶指標的情況下，我們多少有點摸不著頭緒。但我們看到 Verizon 出現相當明顯的放緩，AT&T 則有些加速；我們想了解你們在 FWA 上的具體情況，以及更廣泛的產業狀況。所以想請你們稍微評論一下。

And also, I may have missed your postpaid phone churn number that you typically give on the call, just as a housekeeping item, if I missed that. What was that number?

另外，我可能錯過了你們通常在電話會議上提供的後付費手機流失率（churn）數字，作為例行確認，如果我漏聽了。那個數字是多少？

Sure. Thanks for the question, Craig. Just before I begin to answer it, I read with fascination your piece on convergence and causality versus correlation. I really enjoyed it because it agreed with a lot of our views on it. To your question on FWA, look, we're seeing continued very strong momentum. So deceleration, et cetera, is not where we are. We're seeing really strong momentum, and we're tracking really well ahead of pace to our goal of hitting 15 million customers by 2030. I think I also commented on the fact that importantly, we're seeing ARPU growth both sequentially and year-on-year.

當然。謝謝你的提問，Craig。在我開始回答之前，我很有興趣地讀了你那篇談「融合」以及「因果關係 vs. 相關性」的文章。我非常喜歡，因為它與我們在這方面的許多觀點一致。回到你關於 FWA 的問題：你看，我們持續看到非常強勁的動能。所以所謂的放緩等等，並不是我們目前的狀況。我們看到的是非常強的動能，而且我們朝著 2030 年達到 1,500 萬客戶的目標，進度明顯超前。我也提到過，很重要的一點是，我們看到 ARPU 不論季比或年比都在成長。

And I think a lot of that comes back to the pure quality of the product. I know there have been comments from others in terms of not offering above a certain speed, et cetera, et cetera. The ultra capacity nature of our network just means this thing goes from strength to strength. And in terms of both speeds as well as performance. And I guess our NPS is the best way of thinking about how all of that comes together. And the fact that we have category-leading NPS in broadband including fiber sort of speaks to the fact that we're really not being impacted by any of these other issues in terms of network capacity or any of that.

而我認為其中很大一部分要回到產品本身的純粹品質。我知道其他人曾評論例如不提供高於某個速度等諸如此類。我們網路的超容量（ultra capacity）特性，意味著這項業務只會愈做愈強。不論是在速度或效能表現上都是如此。我想，我們的 NPS 是思考這一切如何整合在一起的最佳方式。而且我們在寬頻（包含光纖）擁有類別領先的 NPS，某種程度也說明我們並未受到其他問題影響，例如網路容量等任何此類因素。

Andre, I don't know if you wanted to add any more color to FWA itself.

Andre，我不知道你是否想再補充一些關於 FWA 本身的細節。

No, I just wanted to add, as you said, I think the performance of the product has been outstanding. And to your concern, Craig, around nets, although we don't divulge in net as you -- line net, as you know, but I think we can state that we had industry-leading FWA nets this quarter. So you can take that as an indication of what we're seeing in terms of our strength in the market.

沒有，我只是想補充一下，如你所說，我認為產品表現一直非常出色。至於你關心的淨新增（nets），Craig，雖然我們不揭露淨新增——如你所知，是線路淨新增——但我想我們可以說，本季我們的 FWA 淨新增是產業領先的。所以你可以把這當作我們在市場上所看到強勢表現的一個指標。

Craig, we didn't give a postpaid phone churn number. We really are focused in on postpaid account net additions and postpaid account churn. Honestly, the postpaid phone churn came in about as we expected and also led to industry-leading postpaid phone net additions.

Craig，我們沒有提供後付費手機流失率的數字。我們真正聚焦的是後付費帳戶淨新增，以及後付費帳戶流失率。坦白說，後付費手機流失率大致符合我們的預期，同時也帶來產業領先的後付費手機淨新增。

But much as we said at the start of the year, and I think what's demonstrated in our results, is when you focus on the right thing, which is account share taking and then translating that into financial leadership, that's what we're trying to keep our eyes on. That's what we're all focused on as a management team and the day-to-day. So we're trying to resist the urge to kind of walk backwards and focus on KPIs that we don't think are the most relevant for the all of value-creating picture.

但正如我們在年初所說的，我想我們的結果也證明了：當你把焦點放在正確的事情上——也就是帳戶市占的取得，並將其轉化為財務領先——這就是我們努力保持關注的方向。這也是我們管理團隊以及日常營運所共同聚焦的重點。因此，我們試著克制那種「倒退回去」轉而關注我們認為與整體創造價值圖像不那麼相關的 KPI 的衝動。

All right. Operator, let's go to the next question.

好的。接線員，我們進入下一個問題。

Kutgun Maral, Evercore ISI.

Kutgun Maral，Evercore ISI。

Great. Thanks for taking the question. I wanted to ask about physical AI as well as your AI-driven network efficiency efforts. So first, Srini, John Saw had a very interesting blog post this week on physical AI discussing, for example, the Kinetic Edge build out with AWS than to work with Figure AI and Serve Robotics.

很好。謝謝讓我提問。我想請教實體 AI（physical AI）以及你們以 AI 驅動的網路效率提升工作。首先，Srini，John Saw 本週有一篇非常有意思的部落格文章談到實體 AI，例如討論與 AWS 合作建置 Kinetic Edge，以及與 Figure AI 和 Serve Robotics 的合作。

As these initiatives move toward commercial scale, what does T-Mobile actually get paid for, is it primarily connectivity or a share of the edge compute economics as well? And who funds the incremental compute infrastructure? Is T-Mobile or your partners? And just generally, anything you could share around the model and future opportunity more broadly would be very helpful.

隨著這些計畫朝商業化規模推進，T-Mobile 實際上會因為什麼而收費？主要是連線服務，還是也會分享邊緣運算的經濟效益？另外，新增的運算基礎設施由誰出資？是 T-Mobile 還是你們的合作夥伴？以及更一般地說，若能分享一下商業模式與更廣泛的未來機會，將非常有幫助。

And then separately, you've recently reported spectral efficiency gains from the Ericsson trials. How close are those gains to being reflected in your capacity planning? And could they meaningfully expand the runway for 5G broadband or change your need for incremental spectrum? Thanks so much.

另外，你們最近也報告了 Ericsson 試驗帶來的頻譜效率提升。這些提升距離反映到你們的容量規劃有多近？它們是否可能實質延長 5G 寬頻的成長跑道，或改變你們對新增頻譜的需求？非常感謝。

Thanks so much for the question, Kutgun. So on Physical AI, let me start by saying we are in the very early innings of this, of the evolution of how this plays out. But I don't think there's anyone in this large virtual room who debate what the world will be like in 2035, whether we're going to see millions of robots, drones, all sorts of other connected moving objects. I think that's kind of a given when you future cast the world. Now the question is what kind of infrastructure do you need to support that? We've all seen what's happened in GenAI, whereas GenAI has grown, there's been a huge infrastructure build-out and TAM to support that technology.

非常感謝你的提問，Kutgun。關於實體 AI（Physical AI），我先說我們仍處於非常早期的階段，這整個演進會如何發展還在起步。但我想在這個大型虛擬會議室裡，應該沒有人會爭論 2035 年的世界會是什麼樣子——我們是否會看到數以百萬計的機器人、無人機，以及各式各樣其他連網的移動物體。當你展望未來時，這幾乎是既定的趨勢。現在的問題是：你需要什麼樣的基礎設施來支撐它？我們都看到了 GenAI 發生了什麼——隨著 GenAI 成長，為了支援這項技術，出現了龐大的基礎設施建置與可服務市場（TAM）。

Now our view on physical AI, and I think it's a view that most people agree with is you'll need an infrastructure build-out to support physical AI, especially with millions of robots, moving things, et cetera, et cetera, that rely on AI, what we call physical AI loosely. I think that infrastructure build is going to look very different because that infrastructure needs to be different. It needs for starters on these 3 different things. It needs low latency connectivity. It needs inference at the edge and most likely for a lot of use cases, it needs voice.

我們對實體 AI 的看法——我想多數人也同意——是你將需要一波基礎設施建置來支援實體 AI，特別是當有數以百萬計的機器人、移動物體等等依賴 AI（我們寬鬆地稱之為實體 AI）。我認為那一波基礎設施建置會看起來非常不同，因為所需的基礎設施本來就不同。首先，它需要三件不同的事情。它需要低延遲連線。它需要在邊緣端進行推論（inference），而且在許多使用情境下，很可能也需要語音。

Now you put those three things together, and what you get is a 5G stand-alone advanced network, which we're the only provider who has nationwide presence off. So we feel like there is going to be this big build out. I'm not going to sit here and speculate whether that is '27 or '28. But for sure, it will happen by '35. And that we're in pole position to be accessing that build-out.

把這三件事放在一起，你得到的就是一個 5G 獨立組網（SA）的進階網路，而我們是唯一一家在全國範圍具備此能力的供應商。因此我們認為，將會有一個大規模的建置潮。我不會在這裡推測那會是 2027 或 2028 年。但可以確定的是，到 2035 年之前一定會發生。而我們正處於最有利的位置，可以參與並受益於這波建置。

On the economics of the model and everything that we're doing with -- on Kinetic Edge, the early use cases have really not been about selling connectivity or selling compute separately. They've been about selling solutions. And those solutions typically what they've done, and this is the important bit, they've reduced the cost of the device because suddenly, you don't need as much battery.

至於我們在 Kinetic Edge 上所做的一切，其商業模式的經濟性——早期的使用案例其實並不是分別去賣連線或分別去賣運算。而是賣解決方案。而這些解決方案通常做到的一件事——這是重點——是降低裝置成本，因為突然之間，你不需要那麼多電池。

You don't need as much AI on board because you can offload that compute. And when you reduce the cost of a device, people are paying for a solution rather than specific pieces of it. So most of our monetization to date has been through providing that solution as a whole and charging for the solution and taking a share of the reduction of the cost of the device.

你也不需要在裝置端搭載那麼多 AI，因為你可以把運算卸載出去。當你降低裝置成本時，人們付費的是一個解決方案，而不是其中某個特定組件。因此截至目前，我們的變現主要是以整體方案提供解決方案、對解決方案收費，並分享裝置成本降低所帶來的部分價值。

Now we'll see how this evolves because I think, this is still early doors on it. We love the fact that we're in pole position on this. As we roll into beyond 5G advanced to 6G, as we double down on this, we're really excited about how this thing progresses. Part of the comment I was making earlier on people like the DOJ testing, what they're testing for is not just coverage any longer, but also capabilities because these more sophisticated buyers are fully aware that the world will evolve in a different way.

接下來會如何演進我們還要觀察，因為這仍然是非常早期的階段。我們很喜歡自己在這方面處於最有利的位置。當我們從 5G Advanced 進一步走向 6G、並加大投入時，我們對這件事的進展感到非常興奮。我先前提到像 DOJ 這類機構在做測試——他們測的不再只是覆蓋而已，也包括能力，因為這些更成熟的買方非常清楚世界將以不同方式演進。

And they're looking to test all of these capabilities before they actually buy the network. And that's a big driver to why our enterprise business has seen substantial traction.

他們在真正購買網路之前，會先測試所有這些能力。而這也是我們企業業務獲得顯著動能的一個重要驅動因素。

John, do you want to pick up the piece on network optimization using AI?

John，你要不要接著談談用 AI 進行網路最佳化的部分？

And spectral efficiency. Yes. On the question, Kutgun on spectral efficiency, yes, we do bake in spectral efficiency gains when we do our planning and growth. But what -- candidly, what we are pleasantly surprised is when we actually handed the wheel to AI for doing things like radio resource planning, the gains were actually more than what we expected. And we actually, just to be sure, took this out of the lab and tested this in New York in a lot of sites, and we're seeing some significant gains that we were not expecting without the use of AI.

以及頻譜效率。好的。針對你關於頻譜效率的問題，Kutgun，是的，我們在做規劃與成長預測時，確實會把頻譜效率提升納入考量。但坦白說，讓我們驚喜的是，當我們把像無線電資源規劃這類工作交給 AI 來做時，實際的增益比我們預期的還要高。而且為了確認，我們把它從實驗室帶到紐約，在很多站點做了測試；在不使用 AI 的情況下，我們看到了原本沒預期到的顯著增益。

So the upside is not done yet, and we are optimistic that potentially there could be even better spectral efficiency coming. And that was with Ericsson and with Nokia they are already talking about spectral efficiency gains of 20% up to 100% as AI-RAN develops. So we are very optimistic. What this means is that in the next few years, our customers especially our fixed wireless customers will get even better experience for that premium fixed wireless product that we offer as well. So very optimistic. The early results look very promising. And this is actually fielded in the market, not tested in the lab.

所以這個上行空間還沒有結束，我們也樂觀認為未來可能還會有更好的頻譜效率提升。而這是在 Ericsson 的合作下；同時 Nokia 也已經在談，隨著 AI-RAN 的發展，頻譜效率可提升 20% 到最高 100%。因此我們非常樂觀。這代表在未來幾年，我們的客戶——尤其是固定無線（FWA）客戶——在我們提供的高階固定無線產品上，將會獲得更好的體驗。所以非常樂觀。早期結果看起來非常有希望。而且這是在市場上實際部署的成果，不是在實驗室測試。

Great. Thanks, Kutgun. Operator, next question, please.

很好。謝謝你，Kutgun。接線員，請下一題。

Sam McHugh, BNP.

Sam McHugh，BNP。

Good afternoon, guys. I think you made it super clear, you don't see any good reasons to give SpaceX. And then we've talked a lot about the Kinetic Edge today already. How would you characterize your willingness to work with someone like SpaceX on Edge Mobile Compute? And you do already resell Starlink in super broadband. Could you envisage reselling it in consumer as well? And do you see any benefit of trying to engage with them on this to see if it might be an alternative to them building out their own network for similar purposes?

各位下午好。我覺得你們講得非常清楚：你們看不出有任何充分理由要讓 SpaceX 進來。而我們今天也已經談了很多 Kinetic Edge。你會如何描述你們與像 SpaceX 這樣的公司在邊緣行動運算（Edge Mobile Compute）方面合作的意願？另外，你們已經在超寬頻（super broadband）中轉售 Starlink。你們是否也能想像在消費者市場轉售它？以及你們是否認為，嘗試在這方面與他們接觸、看看是否能成為他們為類似目的自行建置網路的替代方案，會有任何好處？

Thanks for the question, Sam. So let me start with the MVNO part of it. And I think there's two parts of that question. The way we thought about MVNOs and our MVNO philosophy has fundamentally not changed, right? The way we think about it is does an MVNO give us access to incremental TAM? Does it give us access to new distribution channels? Does it give us incremental capability of any sort? And we've been very clear that in terms of satellite, we see it as a complementary product. And therefore, we don't really see the room for an MVNO for satellite or SpaceX in that context.

謝謝你的問題，Sam。我先從 MVNO 的部分開始。我想這個問題有兩個面向。我們對 MVNO 的看法以及我們的 MVNO 理念，基本上沒有改變，對吧？我們的思考方式是：MVNO 是否能讓我們接觸到新增的可服務市場（TAM）？是否能讓我們接觸到新的通路？是否能帶來任何形式的新增能力？而我們也很清楚地說過，就衛星而言，我們把它視為互補性產品。因此，在這個脈絡下，我們不太認為衛星或 SpaceX 會有 MVNO 的空間。

Now when you're thinking about Physical and Edge AI, again, we're -- from our perspective, we're steaming ahead and building our 5G SA as well as rolling into 6G, we see ourselves a lead in terms of those capabilities. And we see no space for an MVNO here because, well, what additional capabilities does it give us? Why does it change our ability to compete. If we're talking about SpaceX as a customer or us being a customer of them, that's a purely commercial agreement. But the answer to the MVNO is I see no room for an MVNO with SpaceX either on the consumer side or on the Physical and Edge AI side.

至於你提到的實體與邊緣 AI，同樣地，從我們的角度來看，我們正全速前進，建置我們的 5G SA，同時也逐步邁向 6G；在這些能力上，我們自認處於領先地位。而我們也看不出這裡有 MVNO 的空間，因為——它能帶給我們什麼額外能力？它為什麼會改變我們的競爭能力？如果談的是 SpaceX 作為我們的客戶，或我們作為他們的客戶，那就是純粹的商業協議。但就 MVNO 而言，我看不出與 SpaceX 合作 MVNO 的空間——不論是在消費者端，或是在實體與邊緣 AI 端。

Operator, we'll move on to the next question in the queue, please.

接線員，我們將進行到隊列中的下一個問題，麻煩了。

Sean Diffley, Morgan Stanley.

Sean Diffley，摩根士丹利。

Great. Thanks very much, team. I would love to hear how you describe the competitive landscape now as we move into 4Q. It seems like others are focusing more on one and two lines where you guys have dominated historically. So how are you approaching the market there?

很好。非常感謝，各位團隊。我很想聽聽你們在進入第四季時，如何描述目前的競爭態勢。看起來其他業者更聚焦在一、二條產品線，而你們過去在那裡一直占優。所以你們在那個市場的策略是什麼？

And then secondly, just weighing buyback pacing versus future spectrum acquisition, given the share price, is it possible you could step up the pace buybacks here?

第二點，考量到股價，在回購節奏與未來頻譜收購之間權衡之下，你們是否有可能在這裡加快回購步伐？

Okay. Let me pick up the first one and then, Peter, maybe you can lean in on the second. So look, in terms of the competitive environment, I know we all like to obsess about specific elements of this and recent price moves and the rest, right? The way we think about this industry as a whole is fundamentally a multi-variate problem rather than let's take price down in one place, that changed ARPU. Now we need to go back and correct something this. So we think of this as how do we manage the balance between volume and value in a thoughtful value-creating manner.

好的。我先回答第一個問題，然後Peter，第二個也許你可以補充。所以你看，就競爭環境而言，我知道我們都喜歡執著於其中某些特定面向、近期的價格動作等等，對吧？我們看待整個產業的方式，從根本上說是一個多變量問題，而不是在某個地方把價格降下來、ARPU就變了，然後我們又得回頭去修正什麼。因此我們把它視為：如何在量與價值之間取得平衡，並以審慎、能創造價值的方式來管理。

And in many ways, you could say we have the luxury of doing that because that is enabled by our fundamental differentiation across network value and experience, which is why you see us with industry-leading account volumes with 274,000 as well as a clear breakout on ARPA, where you see 4% ARPA growth.

在很多方面，你可以說我們有這樣做的餘裕，因為這是由我們在網路價值與體驗上的根本差異化所支撐；這也是為什麼你會看到我們擁有業界領先的帳戶量，達到274,000，同時在ARPA上也明顯領先，你看到ARPA成長4%。

That formula is the heart of what we manage, and that is -- that is more than kind of drop price in one segment, look to cover that up somewhere else, et cetera, et cetera, right? In terms of the broader industry environment, let me just position where the industry is. I think if you take an arc of 3, 5 years, there's no question that we have seen an increase in subsidies and that's been driven by an increase in device pricing.

這套公式是我們管理的核心；而且那——那遠不只是像在某個區隔降價、再想辦法在別處把它補回來之類的，等等等等，對吧？就更廣泛的產業環境而言，我先說明一下產業目前的位置。我認為如果拉長到3到5年的趨勢弧線，毫無疑問我們看到補貼增加，而這是由裝置價格上升所驅動的。

Now we have been very thoughtful about that, linked our subsidies to higher CLV opportunities such as higher rate plans, T-Mix which is really a response to a customer problem, which is wanting to have more control over their wireless plan. Another way of putting that is don't give me a free phone if I don't need one, allow me to choose what I want, allow me to choose what I want and change that if my circumstances or my needs change is a further move in that direction of driving even more differentiation in terms of the way we deliver value.

而我們在這方面一直非常審慎，將補貼與更高的CLV機會連結，例如更高資費方案、T-Mix（這其實是回應客戶的一個痛點，也就是希望對自己的無線方案有更多掌控）。換句話說，如果我不需要免費手機，就不要硬塞給我；讓我選我想要的，並且在我的情況或需求改變時也能調整——這是朝著以我們提供價值的方式來推動更多差異化的進一步動作。

And as we look forward to Q4, look, Q4 always has variability. And in terms of the competitive environment in Q4, I think you'll see more variability this year because of a set of circumstances, which is a staggered iPhone launch, the increase in device prices, the fact that you don't necessarily have all of the variants, especially the sell-down variance. So we'll see how that plays out, and our guide contemplates all of that.

展望第四季，你看，第四季一向具有變動性。而就第四季的競爭環境而言，我認為今年你會看到更大的變動性，原因是一系列情況：iPhone採取錯峰上市、裝置價格上升，以及你未必能拿到所有型號，特別是清庫（sell-down）的型號差異。所以我們會看看這會如何發展，而我們的指引已把這些因素都納入考量。

Peter, do you want to pick up?

Peter，你要接著回答嗎？

Yeah. On share buybacks, it's really, I think, one, just to remind everybody, through October 23, we just disclosed in the 10-Q, we've now done $10.3 billion in buybacks this year compared to $9.9 billion in all of 2025. And also importantly, increase the dividend per share by 15%, which is multi years now of being able to do that, and that's a great position to be in. But ultimately, what you're asking about is rightly, what's the capital allocation philosophy and how do we see it going forward? In terms of the approach to capital allocation, it's been remarkably consistent and continues to be so.

好。關於股票回購，我想第一點先提醒大家，截至10月23日——我們剛在10-Q裡披露——我們今年已經完成了103億美元的回購，相較之下，2025年全年是99億美元。同時也很重要的是，每股股利提高了15%，而且這已經是連續多年能做到的事，這是一個很好的位置。但你真正問的是——而且問得很對——資本配置哲學是什麼，以及我們如何看待未來？在資本配置的方法上，我們一直非常一致，而且會持續如此。

Starting with leverage, making sure that we've made the right investments in the core of the business, of course, looking at spectrum opportunities and accretive M&A. And then after we're done with that, looking at shareholder returns in a balanced approach between share buybacks and dividends.

從槓桿水準開始，確保我們已在業務核心做出正確投資；當然也會評估頻譜機會與具增值性的併購。然後在完成這些之後，再以平衡的方式在股票回購與股利之間，回饋股東。

And as I look forward, that's -- by the way, I think the application of that formula, particularly when you think about investment in the business as well as investment in spectrum. It's what's really brought us to this position and not only the accomplishments that we've had to date and how the network has driven that, but also what we see going forward, whether it's the 20 million account opportunity for network seekers, whether it's continued growth in 5G broadband.

而展望未來，那——順帶一提，我認為這套公式的落地，特別是當你考慮對業務的投資以及對頻譜的投資時——正是把我們帶到今天這個位置的原因；不僅是我們迄今的成果以及網路如何推動這些成果，也包括我們往前看到的機會：不論是針對「網路追求者」的2,000萬帳戶機會，或是5G寬頻的持續成長。

And I think you heard a lot of our excitement around Physical AI and what we see happening there. In 2027 and 2028 are going to bring a couple of great opportunities to bolster that approach with the spectrum auctions that are upcoming. And so I see, while, of course, we'll continue to be tremendously diligent in our approach to how we think about spectrum, you've seen that play out, many things we've stayed out of, many things that we've purchased, the value creation that's come from it.

我想你也聽到我們對Physical AI的高度興奮，以及我們看到那裡正在發生的事情。2027年與2028年將帶來幾個很好的機會，透過即將到來的頻譜拍賣來強化這個做法。因此我認為，當然我們會持續以極度嚴謹的方式來思考頻譜——你也看到這如何展現：很多我們選擇不參與的項目、很多我們買下的項目，以及由此帶來的價值創造。

So for '27 and '28, I see with those upcoming opportunities, probably more and more of our capital allocation envelope being devoted to spectrum. What ultimately happens in those auctions, of course, again, we'll apply the diligent approach to valuation that we do there. But we'll probably reserve more of that capital envelope for that opportunity just because of what that means from a long-term value creation for the business.

所以對於2027年與2028年，隨著這些即將到來的機會，我認為我們的資本配置額度中，可能會有越來越多投入到頻譜上。至於這些拍賣最終會如何發展，當然我們仍會套用一貫在那裡採取的嚴謹估值方法。但我們可能會為這個機會預留更多資本額度，因為從長期來看，這對企業的價值創造意義重大。

Thank you, Sean. Operator, next question, please.

謝謝你，Sean。接線員，下一個問題，麻煩了。

John Hodulik, UBS.

John Hodulik，瑞銀。

Great. Thank you. Maybe for Peter, the guidance implied for the year here in '26 sort of implies a slowdown in EBITDA growth and account growth, which is a little bit odd, just given the fourth quarter is typically a strong seasonal quarter for account growth. And I guess really just dovetailing off Srini's comments on the variability in the fourth quarter. Is there -- is there anything sort of to call out as we look into the fourth quarter? Or are you guys just trying to set a low bar for Jessica here. That's number one.

很好。謝謝。也許問Peter：你們對2026年全年的指引隱含EBITDA成長與帳戶成長放緩，這有點奇怪，因為第四季通常是帳戶成長季節性很強的一季。我想這也呼應Srini剛才提到的第四季變動性。當我們看向第四季時，有沒有什麼特別需要點出的因素？還是你們只是想替Jessica把門檻設低一點。這是第一點。

And then number two is just on churn. You guys -- you said that the higher churn of the 12 bps higher churn is sort of what you guys expected, I guess, really driven by the price increase you had. Just how is that trending now as we look into the fourth quarter? Do you think that goes sort of -- you sort of get that back over time and that goes back into a more normalized area?

第二點是關於流失率（churn）。你們說流失率提高12個基點是你們預期之內，我想主要是由你們先前的漲價所驅動。那麼當我們看向第四季時，這個趨勢現在如何？你認為這會隨時間逐步回收，回到較正常的區間嗎？

Yes, sure. Well, let me take that. There was a lot in that question. So Q4, one on EBITDA, I just want to remind everybody, of course, that when you're thinking about Q4, it's sequentially when you're thinking about it, of course, Q4 is when we have the holiday season generally tends to be the most promotionally intensive season. And so you tend to see a Q3 to Q4 EBITDA investment profile as a result of that. The other thing that you have happening is M&A.

好的，當然。我來回答。這個問題涵蓋很多內容。所以第四季，先談EBITDA，我想提醒大家，當你在思考第四季、以季對季來看時，第四季是我們有假期季的時候，一般而言也是促銷最密集的季節。因此你通常會看到從第三季到第四季，EBITDA呈現一個投資（投入）輪廓。另外一件正在發生的事情是併購（M&A）。

Because for the balance of the year, what's been happening is really odd timing with respect to which quarter included what M&A last year, including Q3, if you recall, the timing of the close of US Cellular as well as some of the fiber acquisitions. So what you have happening in Q4, and this is why I mentioned the postpaid service revenue of 7% on a year-over-year basis is you're going to get a great picture of organic growth.

因為就今年剩餘期間的表現來看，去年各季納入哪些併購案（M&A）的時間點其實非常不尋常，包括第三季（Q3），如果你還記得，美國 US Cellular 的交割時間，以及部分光纖收購案的時間點。因此你會看到第四季（Q4）發生的情況——這也是我提到後付費服務營收年增 7% 的原因——你將會看到一幅非常清楚的有機成長圖像。

And so that's how I think about EBITDA, 7% postpaid service revenue growth for Q4. On an accounts perspective, yes, to your point, look, it's a 1 million postpaid net account additions guide that we're giving you, which is if you think about the heydays of the Un-carrier that we sometimes get asked about in 2014, 2013, we're delivering more postpaid net account additions in 2026 than during those days. And obviously, you're seeing that translate into just the magnitude of the absolute dollar growth in the financials as well.

所以我看 EBITDA 的方式是：第四季後付費服務營收成長 7%。從帳戶（accounts）的角度來看，沒錯，如你所說，我們給出的指引是後付費淨新增帳戶 100 萬戶；如果你回想一下大家有時會問到的 2014、2013 年 Un-carrier 的黃金時期，我們在 2026 年交付的後付費淨新增帳戶數，甚至比那段時期還多。而且很明顯，你也看到這轉化為財務報表上絕對金額成長的規模。

And so Q4, yes, it has variability for all the reasons that Srini just mentioned, there's a difference in terms of the iPhone cycle. There's no base model phone. We have to see how customers react to the OEM price increases. And so -- and there's frankly also been some Q3 inventory constraints as well that have shifted some things from Q3 into Q4 including from an EBITDA perspective because we don't recognize the cost until we actually ship it and fulfill it. So there's a lot of the variability there. But at the end of the day, it's 1 million postpaid net account additions and of course, $100 million up at the midpoint of the range on core adjusted EBITDA.

至於第四季，是的，基於 Srini 剛提到的各種原因，確實存在波動性；其中一個差異在於 iPhone 週期。這次沒有入門款（base model）手機。我們必須觀察客戶對 OEM 漲價的反應。另外，坦白說，第三季也出現了一些庫存限制，導致部分項目從第三季移轉到第四季——從 EBITDA 的角度也是如此，因為我們要等到實際出貨並完成履約時才會認列成本。所以波動性來源很多。但歸根結底，指引就是後付費淨新增帳戶 100 萬戶，當然核心調整後 EBITDA 在區間中點也上調 1 億美元。

On churn, there's a couple of dynamics at play, especially as you're thinking about it into Q4. One, again, seasonally for everybody, Q4 tends to be a higher churn quarter than Q3 because of the holiday season. There's a lot of excitement around it. I'd expect that seasonality to continue. So churn implicitly would go up, account churn as well.

關於流失率（churn），有幾個動態在發揮作用，特別是當你把它放到第四季來思考時。第一，對所有業者而言，季節性上第四季通常會比第三季有更高的流失率，因為有假期季。市場氛圍很熱絡。我預期這種季節性會延續。因此流失率會隱含上升，帳戶流失率也會上升。

And the other thing that you have happening in accounts, again is there's more 5G broadband and fiber only accounts. However, we had US Cellular as well come on board. And so you're still going to see the dynamics of yes, accounts include blended accounts, but they also include more and more 5G broadband-only or fiber-only accounts, which offer obviously a great opportunity in the future for cross-sell and deepening the relationship, but just as a result of, well, broadband churn is higher than postpaid phone, it's a pure mathematical exercise.

第二個在帳戶端發生的事情是：純 5G 寬頻與純光纖帳戶變多了。不過，我們也把 US Cellular 納入了。因此你仍會看到這樣的動態：沒錯，帳戶包含混合型帳戶，但也愈來愈多包含僅 5G 寬頻或僅光纖的帳戶；這些帳戶未來在交叉銷售與深化關係上顯然提供很好的機會，但由於寬頻的流失率高於後付費手機，這純粹是數學上的結果。

So those 2 things will still happen in Q4. All of that is contemplated again in the postpaid net account addition guide that we gave as well as the core EBITDA range. So I'm hoping some of that helps with all the myriad of questions you threw at me. And I can't give you the spreadsheet as always.

所以這兩件事在第四季仍會發生。上述所有因素也都已納入我們給出的後付費淨新增帳戶指引，以及核心 EBITDA 區間之中。希望這些能回答你丟給我的一大堆問題。而且我一如既往不能把試算表給你。

Operator, next question please.

接線員，請下一題。

Kannan Venkateshwar, Barclays.

Barclays 的 Kannan Venkateshwar。

So maybe on ARPA, when we look at the mix here in terms of subscribers, there is a push towards family accounts that you guys are heading towards that as your peers are obviously focused on single line and double-line accounts. And so as we extrapolate some of these mix differences, it feels like some of the ARPA strength could be structural and could basically heading into next year could also help your growth. So any color around that would be helpful in terms of the trajectory of ARPA growth from here?

那我可能先問 ARPA。當我們看這裡的用戶組合時，你們似乎在推動家庭帳戶（family accounts），而同業顯然更聚焦在單門號與雙門號帳戶。因此當我們外推這些組合差異時，感覺 ARPA 的強勢可能是結構性的，並且進入明年也可能有助於你們的成長。所以能否請你們就此提供一些說明，特別是從現在起 ARPA 成長軌跡會如何？

And then on the satellite side, Srini, I mean, we get your pushback on the MVNOs broadly. But there is also an advantage if you were to work with satellite in some ways in the sense that you could open up some of your spectrum capacity if you can lean on satellite broadband to some extent. And of course, you get the coverage breadth your fiber scale is a lot smaller than your peers. So to that extent, you get an advantage of using that footprint on satellite. So is this more a SpaceX-specific pushback? Or is this just philosophically something that you're not keen on with respect to any satellite provider. Thank you.

另外在衛星方面，Srini，我的意思是，我們理解你對 MVNO 整體的反對意見。但如果以某些方式與衛星合作也有優勢，因為若你能在一定程度上依靠衛星寬頻，就可能釋放部分頻譜容量。而且你們的光纖規模比同業小很多，衛星也能帶來更廣的覆蓋。所以在這個層面上，利用衛星的覆蓋版圖會有優勢。那這比較像是針對 SpaceX 的特定反對？還是說在理念上，你對任何衛星供應商都不太感興趣？謝謝。

All right. Let me pick up the satellite question first and then go to you, Peter. So Kannan, thanks for that. I saw your note as well in terms of the shape shifting nature of SpaceX strategy. So it's hard to refer to them now as a satellite provider or a satellite provider that is going to build terrestrial. So hard to think about whether this is SpaceX alone or all of satellites.

好。我先回答衛星的問題，然後再交給你，Peter。Kannan，謝謝你的提問。我也看到了你提到 SpaceX 策略不斷變形的那份說明。所以現在很難把他們單純稱為衛星供應商，或是一家會建設地面網路（terrestrial）的衛星供應商。因此也很難判斷這究竟是 SpaceX 個案，還是所有衛星業者的共通情況。

So let me answer that question in its broadest terms, right? Here's the way we think about broadband. We've got a great product in FWA and really, in many ways, except for some deep rural parts satellite is an inferior version of FWA with even more constraints in terms of things like what is the size of the beam, right? So I don't see the synergy there between FWA and satellite. In many ways, two thirds of our customers come from dense urban areas. We see huge growth opportunities in those areas still all under the fallow capacity model. So we don't really see a synergistic opportunity with satellite providers, and that's a broader piece. We see it as a complementary product very much in the nature of T-Satellite.

所以我用最廣義的方式來回答這個問題。我們看待寬頻的方式是這樣：我們在 FWA（固定無線接取）上有很好的產品，而且在很多方面——除了某些深度偏遠地區——衛星其實是比 FWA 更劣化的版本，並且在例如波束大小（beam size）等方面限制更多。所以我看不出 FWA 與衛星之間有什麼協同效應。在很多情況下，我們三分之二的客戶來自人口稠密的都會區。在這些區域、在閒置容量（fallow capacity）模型之下，我們仍看到巨大的成長機會。因此我們不太認為與衛星供應商存在協同機會，這是更宏觀的看法。我們把它視為一種互補產品，性質上很像 T-Satellite。

But if you're thinking of that in broader MVNO terms, I think it comes back to the philosophy with which we operate MVNOs, which is what is the incremental TAM that something opens out. And am I willing to get into a deal for the incremental TAM? If it's a complementary product I'm very happy to look at a resale arrangement, which is exactly what the satellite was, which gets built into it. But an MVNO is a wholly different thing. So that's -- I hope that clarifies where our head's at on MVNOs and more broadly, satellite.

但如果你是從更廣義的 MVNO 角度來思考，我認為就回到我們經營 MVNO 的理念：某個合作能開啟多少增量 TAM（可服務總市場）？而我是否願意為了這個增量 TAM 去做一筆交易？如果它是互補產品，我很樂意評估轉售（resale）安排——衛星正是這種情況，也已經內建在其中。但 MVNO 是完全不同的一件事。所以——希望這能釐清我們對 MVNO 以及更廣泛的衛星議題的想法。

And on the ARPA question, this isn't really -- I think your inference there was, we've given up on one and two lines, and we're getting multi-lines and that's driving all of the ARPA growth. And that couldn't be further from the truth. I think what you're seeing is, again, the differentiation between the value, the network, the experience, there is plenty of customer pain points left, whether you're talking about a single-line plan or a multiline plan.

至於 ARPA 的問題，這其實不是——我想你的推論是：我們放棄了一門與兩門方案，轉而追求多門號，並由此推動所有 ARPA 成長。但事實完全不是如此。我認為你看到的是：在價值、網路與體驗上的差異化，仍然存在大量客戶痛點，不管你談的是單門號方案還是多門號方案。

You just saw T-Mix and that launch starting to address yet another carrier customer pain point that we're here to break. I think what's really driving customers and what's really driving ARPA and we've highlighted some of this in the fact book as well that you can see are a few things, introduction of new rate plans and the ongoing adoption of premium rate plans by our customers.

你剛看到 T-Mix，以及那次上市開始解決我們要打破的另一個電信業客戶痛點。我認為真正驅動客戶、也真正驅動 ARPA 的因素——我們在 fact book 裡也重點提到、你也看得到——有幾點：新資費方案的推出，以及客戶持續採用高階資費方案（premium rate plans）。

We continue to see for new accounts coming to T-Mobile, 60% of those lines are on our premium rate plans. That's one big factor. Obviously, you have deepening of account relationships, inclusive of ongoing sell in a 5G broadband and now with fiber as well, other connected devices. You're continuing to see port-in ARPUs that are 20% higher than port out ARPA.

我們持續看到，新加入 T-Mobile 的新帳戶中，有 60% 的門號選擇我們的高階資費方案。這是一個很大的因素。顯然，帳戶關係也在加深，包括持續銷售 5G 寬頻，現在也包含光纖，以及其他連網裝置。你也持續看到，攜入的 ARPU 比攜出的 ARPA 高出 20%。

So it talks about that proposition, particularly the network and the network seekers is allowing us to capture higher value customers. But in terms of your ask of what's 2027 like? Well, we're going to update 2027 guidance when we get to the Q4 call. We're heads down here in terms of how do we execute this year and continue the growth that we see. But those are the main factors driving ARPA. And I can tell you, this team is heads down on whether you're on a single line or a multiline account family, we want you here because this is a place to be.

所以這談到的是該價值主張，特別是網路以及尋求網路品質的客群，讓我們能夠吸引更高價值的客戶。但就你問的 2027 年會是什麼樣？我們會在第四季電話會議時更新 2027 年的指引。我們目前正專注於如何把今年執行到位，並延續我們看到的成長。但以上就是推動 ARPA 的主要因素。我也可以告訴你，這個團隊正全力以赴：不論你是單門號或多門號家庭帳戶，我們都希望你在這裡，因為這裡就是該來的地方。

And to your point of whether there is something structural that advantages us, I'll be boring, but go back to my standard thing of the structural thing is having the best network, the best value and the best experience. That's what drives a lot of the ARPA. And then we choose to compete in different segments based on value, but we also choose to compete based on where we get most customer traction.

至於你提到是否有什麼結構性因素讓我們占優勢，我可能會講得有點無聊，但我會回到我一貫的說法：結構性的關鍵就是擁有最好的網路、最好的價值，以及最好的體驗。這些驅動了很大一部分的 ARPA。然後我們會基於價值在不同細分市場競爭，但我們也會選擇在最能獲得客戶動能的地方競爭。

Great. Thank you, Kannan. I'm going to take one from the social board from Chetan. What assumptions about 2030 traffic profile are driving T-Mobile's network investment decisions today? Where do you expect the first meaningful capacity constraints to appear on the radio uplink, on edge compute, transport interconnection with AI infrastructure?

很好。謝謝你，Kannan。我想從社群看板上挑一題，來自 Chetan。關於 2030 年的流量樣貌，你們有哪些假設在驅動 T-Mobile 今天的網路投資決策？你們預期第一個有意義的容量限制會出現在哪裡：無線上行、邊緣運算、或是與 AI 基礎設施的傳輸互連？

I'll take that. This is John. Chetan, thank you for your question. For 2030, we plan a traffic profile that's still downlink heavy, I think, but increasingly uplink rich. driven by the rise in AI usage, AI, wearables, AI agents and also the explosive growth by then of physical AI devices, like humanoid, robots, delivery drones, vehicles and factory machines, et cetera.

我來回答。我是 John。Chetan，謝謝你的提問。針對 2030 年，我們規劃的流量樣貌我認為仍會以下行為主，但上行也會愈來愈豐富。這主要由 AI 使用量上升、AI、穿戴式裝置、AI 代理（agents），以及到那時實體 AI 裝置（例如人形機器人、配送無人機、車輛與工廠機械等）的爆發式成長所驅動。

2030 will also see 6G networks being built and T-Mobile will be the first with 6G in this country with new applications like integrated sensing and communications ISA. I think 6G will be the first G, where the base stations and radios, behave as radars, sensors unlike the previous Gs.

2030 年也會看到 6G 網路開始建置，而 T-Mobile 會成為這個國家第一個推出 6G 的業者，並帶來像是整合式感測與通訊（ISA）等新應用。

So that opens up a lot of new use cases with remote sensing, et cetera, that I think is going to drive a lot of usage as well. One thing I learned over the years of building all the Gs is that when you give customers a more capable network, they will find ways to use it and usage always goes up. In fact, in the last five years, starting from when we started rolling out 5G, we see an explosive 5x increase in tonnage that we carried until today.

因此，這會開啟許多新的使用情境，例如遠端感測等，我認為也會帶動大量使用。我多年來建置各代行動網路學到的一件事是：當你給客戶更有能力的網路，他們就會找到使用它的方法，而使用量總是會上升。事實上，在過去五年，從我們開始部署 5G 起，我們看到截至今日所承載的流量出現爆發性的 5 倍成長。

And our speeds have nearly doubled.

而我們的速度幾乎翻倍。

And our speeds have more than doubled as well. So we expect nothing less from 6G, if anything, even more. So our capacity plan already assumes network payloads increasing like this. And so we are more than ready for the growth in 2030. Now on your question on constraints, my job and my network team's job is to make sure that we have none.

而我們的速度也已經超過翻倍。所以我們對 6G 的期待不會更少，甚至可能更多。因此，我們的容量規劃已經假設網路負載會以這樣的方式增加。所以我們對 2030 年的成長已經做好充分準備。至於你問到的限制點，我和我的網路團隊的工作就是確保不會有任何限制。

But if there's one thing that we need to keep an eye on is the radio uplink. Especially in dense and BZ sectors and especially with the growth in physical AI and a lot of AI agents. This is why 5G Advanced matters. This is why we roll out the country's first and only 5G Advanced network where uplink enhancements like uplink carrier aggregation, uplink MIMO, transmit switching, diversity and higher power uplinks, right? And they are also backed by our strong TDD spectrum depth that we can actually use to support a bigger uplink requirement if there's a need to.

但如果有一件事需要密切關注，那就是無線上行。尤其是在高密度與 BZ 扇區，並且隨著實體 AI 與大量 AI 代理的成長。這也是為什麼 5G Advanced 很重要。這也是為什麼我們推出全國第一且目前唯一的 5G Advanced 網路，具備上行增強能力，例如上行載波聚合、上行 MIMO、發射切換、分集，以及更高功率的上行等，對吧？而且這些也有我們強大的 TDD 頻譜深度作為後盾；若有需要，我們實際上可以用它來支援更高的上行需求。

And this is why with 6G, I'm so happy that the FCC is now putting onto the auction pipeline, the C-band 2.0 and the 2.7 gigahertz bands all of which are TDD. It gives operators like us the option to allocate more capacity and more spectrum for the uplink.

也因此，展望 6G，我很高興 FCC 現在把 C-band 2.0 與 2.7GHz 頻段納入拍賣管線，而這些全都是 TDD。

Thanks, Dr. Saw. Operator, let's get back to the queue with the next question, please.

謝謝你，Saw 博士。接線員，請回到排隊序列，進行下一個問題。

Michael Rollins, Citigroup Inc.

Michael Rollins，花旗集團（Citigroup Inc.）。

Thanks, and good afternoon. I wanted to follow up on Dr. Saw's comments regarding AI agents. And from a customer perspective, curious if you're seeing any impact, any influence from the early use of agents? And how you look at that in terms of interacting with customers and what that can mean for customers trying to extract more efficiency out of the rate plans and hence, the potential risk to revenue for wireless and broadband for that matter?

謝謝，下午好。我想追問 Saw 博士關於 AI 代理的評論。從客戶角度來看，我很好奇你們是否看到任何影響、任何來自早期使用代理的作用？以及你們如何看待這在與客戶互動上的意義，並且這對客戶試圖從資費方案中榨取更多效率、進而對無線與寬頻營收可能帶來的風險，你們怎麼看？

And then just one other question, if I could. So, you made the decision earlier this year to move to a 3-year equipment cycle. I'm curious what you're seeing from that, what you're seeing from device replacement rates? And given what you're seeing, is there a case to push that cycle even out longer for customers? Thanks.

另外如果可以，我還有一個問題。你們今年稍早做出決定，把設備週期改為 3 年。我想了解你們從中看到了什麼、在裝置汰換率方面看到了什麼？以及基於你們目前看到的情況，是否有理由把這個週期對客戶再拉得更長？謝謝。

So let me pick up the replacement cycle, and John can lead into the agents on how we're seeing that. So on devices as a whole, look, I think the operating principle for us is really to give customers choice on when they wanted to replace their device. Part of the whole beauty of the T-Mix structure is that. Right now, whether we'll go to 48 months or 60 months, whatever is all going to be for us led by the customer and what we're seeing in terms of replacement cycles.

我先接著談汰換週期，然後 John 可以接著談代理以及我們目前看到的情況。就整體裝置而言，我認為我們的運作原則其實是讓客戶可以選擇他們想在何時更換裝置。T-Mix 架構的美妙之處有一部分就在這裡。目前來看，未來是否會走到 48 個月或 60 個月之類，對我們而言都會由客戶主導，並取決於我們在汰換週期上看到的趨勢。

At this point, we think 36 is the right duration. We offer both 24, 36. And 36 is the right duration to go to. But the important piece for me is not just to restrict our thinking on choice to device cycle length. That's a very restricted view of how you give customers flexibility. We want to give them flexibility on whether they want the device at all or not. We want to give them flexibility on -- they may not need it right now. They may wait for the new version in six months. That's okay.

就目前而言，我們認為 36 個月是合適的期間。我們同時提供 24 個月與 36 個月。而 36 個月是合適的選擇。但對我來說，重要的不只是把「選擇」的思考限制在裝置週期長度上。那是對如何給客戶彈性的一種非常狹隘的看法。我們希望給他們彈性：他們可以選擇到底要不要裝置。我們也希望給他們彈性——他們可能現在不需要。他們可能會等六個月後的新版本。那也沒問題。

We don't believe in the current view of you either take it out an acquisition or you lose the chance to. And a lot of our philosophy on how we think about customer flexibility, including on device choice is reflected in T-Mix.

我們不認同目前那種觀點：你要嘛在申辦時就把它帶走，否則就失去機會。而我們在如何看待客戶彈性（包括裝置選擇）上的許多理念，都反映在 T-Mix 之中。

John, do you want to talk about agents and our view on that?

John，你想談談代理以及我們對此的看法嗎？

Yeah. Thanks, Mike, for the question. We haven't really seen anything to date nor do we see anything emerging around Muse or Instinct or any of the AI agents that are out there. But as you would expect, we're watching this very closely. We're being very thoughtful about it as we continue to advance in an AI-centric world. We're going to see more and more of this. But I think it's also important to remember that were the customer champion on this? And when you think about fairness and transparency and value, we've always been the champion for this.

是的。謝謝你，Mike，提出這個問題。到目前為止，我們其實沒有看到任何相關情況，也沒有看到在 Muse 或 Instinct 或任何現有的 AI 代理（AI agents）周邊有什麼正在浮現的跡象。但如你所預期的，我們正非常密切地關注這件事。在我們持續在以 AI 為核心的世界中前進之際，我們也非常審慎地思考這件事。我們會看到這類情況越來越多。但我認為也很重要的一點是，要記得我們在這件事上是客戶的捍衛者（customer champion）。當你想到公平、透明與價值時，我們一直都是這方面的倡議者。

Transparency has always been a big, huge thing for us and making sure that customers are getting the very best value for every dollar that they spend with us, making sure that customers are getting the best product quality and, of course, getting the very best experience. One of the things, too, that I think it's also important to note is we have this asymmetrical advantage at T-Mobile, unlike our competitors, when you look at the front book versus back book, back book being much lower on average than front book. So we feel like we're very well positioned on this entire front. And just like AI search, if there's people asking about their wireless relationship and are they getting the very best deal of all the options out there.

透明度一直以來都是我們非常、非常重視的一件事，並確保客戶在每一塊錢花在我們身上時都能獲得最佳價值，確保客戶拿到最好的產品品質，當然也能獲得最好的體驗。另外我認為也值得注意的是，T-Mobile 相較於競爭對手有一個不對稱優勢：當你看前段客群（front book）與後段客群（back book）時，後段客群的平均價格要低得多。因此我們覺得在整個面向上，我們的位置非常有利。而且就像 AI 搜尋一樣，如果有人在詢問他們的無線服務關係、以及在所有選項中他們是否拿到最划算的方案。

Just like AI Search, we like those jump balls and people asking those kinds of questions because it's going to lead them to, if you're not with T-Mobile, you should be with T-Mobile. And if the AI agents are proliferating that in some way in the future, and again, we haven't seen anything of significance. But if that's happens in the future, we see that overall really, really a good thing on balance. But we're very, very thoughtful. We're staying really close to it, but I like where we sit here relative to our advantages and what this could do potentially in the marketplace.

就像 AI 搜尋一樣，我們喜歡這種「五五波」的球（jump balls），也喜歡人們問這類問題，因為這會引導他們得出結論：如果你不是 T-Mobile 的客戶，你就應該成為 T-Mobile 的客戶。而如果未來 AI 代理以某種方式把這件事擴散開來——再次強調，我們目前沒有看到任何顯著的情況。但如果未來真的發生，我們整體上認為利大於弊，是一件非常、非常好的事。不過我們會非常、非常審慎。我們會緊密追蹤，但就我們的優勢位置以及這可能在市場上帶來的影響而言，我喜歡我們目前所處的位置。

Thank you. Thanks, Mike. Operator, next question, please.

謝謝。謝謝你，Mike。接線員，請下一題。

Peter Supino, Wolfe Research.

Wolfe Research 的 Peter Supino。

Hi, everybody. A question about spectrum. Last quarter, you spoke positively about the auctions the FCC is planning for 2027 and '28. And many investors are interested in understanding how to reconcile your leading position in spectrum with your appetite for those auctions. And particularly given you've talked about FWA as a fallow capacity model. So I just wondered if you could expand on the driver of your interest in those spectrum auctions?

各位好。我有一個關於頻譜的問題。上一季你們對 FCC 計畫在 2027 與 2028 年進行的拍賣表達了正面看法。許多投資人想了解，如何把你們在頻譜上的領先地位，與你們對那些拍賣的興趣（胃納）做出一致的解讀。尤其是考量你們談到 FWA 是一種「閒置容量」（fallow capacity）模式。所以我想請你們多談談：你們對那些頻譜拍賣感興趣的驅動因素是什麼？

Sure. Thanks for that, Peter. So the way we think about spectrum is there are 2 factors that drive our appetite for spectrum. And I'll come back to kind of C-band and 2.7. One of them is a commitment to continue to drive network leadership, not for its own sake. But because of the opportunity that it has opened out for us historically and the opportunity leaning in. If you just go back to 5G from the point we made the decision to go all in on 5G. It has had a dramatic transformational impact on this business. In terms of our small markets and rural areas share going from 13% to 24% in terms of being able to drive a fallow capacity business to 15 million FWA customers by the end of the decade.

當然。謝謝你，Peter。我們看待頻譜的方式是：有兩個因素會驅動我們對頻譜的需求（胃納）。我等一下會回到 C-band 與 2.7。其一是我們承諾要持續推動網路領先地位，但不是為了領先而領先。而是因為這種領先在歷史上為我們打開了機會，而且未來持續加碼也會帶來機會。如果你回顧 5G，從我們決定全面投入 5G 的那一刻起，它對這個業務帶來了戲劇性、轉型性的影響。例如在小型市場與鄉村地區的市占，從 13% 提升到 24%；也讓我們能夠推動一個「閒置容量」業務，目標是在本十年末達到 1,500 萬 FWA 客戶。

In terms of the opportunity, it's opened out in enterprises. As we look forward to things like 6G and where that goes, the same set of opportunities exist. And you also have, in addition, this whole idea of what the TAM could be for Physical and Edge AI and how that drives it. So that's one set of considerations. We're just looking at the overall impact of network leadership and its multifaceted impact on different parts of our business. And we look at that very rigorously.

就機會而言，它也為企業市場打開了空間。展望 6G 以及其發展方向，同樣的一組機會依然存在。此外，你還有另一個概念：實體與邊緣 AI（Physical and Edge AI）的可服務總市場（TAM）可能有多大，以及它如何驅動需求。所以這是一組考量。我們是在看網路領先地位的整體影響，以及它對我們業務不同部分所帶來的多面向影響。而我們會非常嚴謹地評估這些。

Now in parallel to that, there's also a very analytical build or densify versus buy piece that determines the value of that spectrum. And you've seen us at points in time, walk away from spectrum because the build versus densify -- sorry, the densify versus buy clearly pointed us in the direction of densify. Good example of that is the EchoStar spectrum, where we walked away from that deal because it didn't make economic sense.

與此同時，還有一個非常分析導向的「自建或加密（densify）對比購買」的判斷，用來決定該頻譜的價值。你也看到我們在某些時點會選擇放棄頻譜，因為「加密對比購買」——抱歉，是「加密對比購買」的分析結果很清楚地指向應該去做加密。一個很好的例子是 EchoStar 的頻譜，我們退出那筆交易，因為在經濟上不合理。

So keeping these two in balance, making sure that we understand the upside of network leadership and its multifaceted impact on the business, while being really rigorous about thinking through analytically the right price to buy that spectrum is the way we approach it. So as we look at C-band 2.0 or the next C-band auction and 2.7, that's how you should expect us to approach it. We, in general, in principle, always welcome the coming of more spectrum to the market because it opens up a ton of possibilities. That doesn't mean we're going to pay an uneconomic price for it. So that's the balance of how we think about it.

因此，把這兩者維持在平衡：一方面確保我們理解網路領先地位的上行空間及其對業務的多面向影響；另一方面在分析上非常嚴謹地思考購買該頻譜的合理價格——這就是我們的做法。所以當我們看 C-band 2.0 或下一次 C-band 拍賣，以及 2.7 時，你應該預期我們會用同樣方式來評估。原則上，我們一向歡迎市場上有更多頻譜釋出，因為它會打開大量可能性。但這不代表我們會為它支付不具經濟性的價格。這就是我們思考的平衡點。

Thanks, Peter. Operator, let's go to the next question, please.

謝謝你，Peter。接線員，請下一題。

Michael Ng, Goldman Sachs.

Goldman Sachs 的 Michael Ng。

I just have two as well. Could you talk a little bit about some of your cost and digital flow initiatives, just given the success that you've had with T-Life and then the network optimization across your RAN footprint.

我也有兩個問題。鑑於你們在 T-Life 上取得的成功，以及你們在整個 RAN 覆蓋範圍內的網路最佳化，能否談談你們的一些成本與數位流程（digital flow）計畫？

And then just as a housekeeping item, I was just wondering if you could talk about the drivers of the wholesale and I think the prepaid outlook changed for the full year? Thank you.

另外一個是例行性問題（housekeeping）：我想請你們談談，為什麼你們對全年的批發（wholesale）以及我認為是預付（prepaid）的展望有所調整？謝謝。

So let me pick up the cost piece and more broadly, I think the AI and digital transformation, and then Peter can come back to the housekeeping stuff. So, we're really pleased with the progress we're making on the AI and digital transformation. Now this is multifaceted. It starts with things like what we've seen in care. We boldly said three years ago that we would reduce the number of calls coming into care by 75%. We're well on track on that journey. We've seen real progress there in terms of our AI voice assistant in terms of not just call deflection, but proactive call prevention by solving the problem before the customer has it, which comes back to our broader philosophy on digital and AI, which is -- this is not about cost.

那我先接成本這一塊，以及更廣泛的 AI 與數位轉型，然後 Peter 再回來回答例行性事項。我們對 AI 與數位轉型的進展感到非常滿意。這是多面向的。它從我們在客服（care）看到的成果開始。三年前我們大膽地說過，我們會把打進客服的來電量降低 75%。我們在這段旅程上進展順利。我們在 AI 語音助理方面看到實質進步，不只是把電話導流（call deflection），更是透過在客戶遇到問題之前就先解決問題來主動預防來電（proactive call prevention）。這也回到我們更廣泛的數位與 AI 哲學——也就是：這不是為了成本。

This is about improving the experience so that people don't need to call, so that people don't need that service. We use it extensively in driving our network optimization with our self-optimizing network, but also all of our CapEx is directed by an AI-driven tool that we call customer-driven coverage and that's providing really, really strong results as well. I think the one place we've not talked about as much in the past is how much AI is fundamentally changing our tech capabilities, especially on our IT stack and the extent to which you look at something like T-Mix, and you should go have a player with the floor.

而是為了改善體驗，讓人們不需要打電話、不需要那項服務。我們也大量用它來推動網路最佳化，透過我們的自我最佳化網路（self-optimizing network）；同時，我們所有的資本支出（CapEx）也都是由一個 AI 驅動的工具來指引，我們稱之為「客戶驅動覆蓋」（customer-driven coverage），而且也帶來非常、非常強的成果。我認為有一個我們過去談得比較少的地方，是 AI 正在多大程度上從根本改變我們的技術能力，特別是在我們的 IT 技術堆疊（IT stack）上；以及當你看像 T-Mix 這樣的東西，你應該去找一位在現場的人來示範。

Something like that is fundamentally enabled by a very different AI-enabled tech stack that's resulted in significantly higher throughput and also significantly lower cost to own inclusive of the cost of tokens. So we're seeing a broad impact of AI and digital across the board.

像那樣的能力，根本上是由一個非常不同、AI 賦能的技術堆疊所支撐，帶來顯著更高的吞吐量（throughput），同時也顯著降低了持有成本（cost to own），而且包含了 token 成本。因此我們看到 AI 與數位化在各個面向都帶來廣泛影響。

Obviously, things that happen in T-Life and greater penetration further helps reduce care volumes further helps reduce the amount of service that's required and makes it a more frictionless painless process to be -- to be a T-Mobile customer. So it's all going really well. We're feeling very good about it and we're doubling down on how we drive this transformation going forward. Peter?

顯然，在 T-Life 中發生的事情以及更高的滲透率，進一步有助於降低客服量、進一步減少所需的服務量，並讓成為——成為 T-Mobile 客戶的流程更順暢、無摩擦且不痛苦。所以一切進展得非常順利。我們對此感覺非常好，並且會加倍投入，推動這項轉型在未來持續前進。Peter？

Excellent. Yes, Michael, on the house cleaning items, there's really a couple of things. On the wholesale and other service revenue line, when I'm thinking about what is the drag 2026 relative to 2025, it is, while it was a long-standing expectation that both TracFone and Boost would come off, we make estimates at the start of the year at the midyear and those have rolled off at a little bit faster pace than we anticipated. So that's all that's happening there when I think about '26 relative to 2025 and the downward pressure.

很好。是的，Michael，關於一些整理性的事項，主要有幾點。在批發及其他服務營收這一行，當我思考 2026 年相較 2025 年的拖累是什麼時，情況是：雖然 TracFone 和 Boost 會退出一直是長期預期，我們會在年初與年中做估算，而它們的流失速度比我們預期稍快。所以，當我思考 2026 相對 2025 的情況以及下行壓力時，這就是全部正在發生的事。

On prepaid, again, we foreshadowed this. This is a continuation of a trend that we've seen and talked about this a little bit at Q2. For Q4, I expect that sequential change might be just a slightly bit higher sequential change. But when you step back, I think what's really happening in prepaid for us is there's a subset of prepaid customers.

至於預付費（prepaid），同樣地，我們之前也已預告過。這是我們看到並一直在談的趨勢延續，我們在第二季也稍微談到過。對第四季而言，我預期環比變化可能會略微高一些。但退一步看，我認為我們在預付費真正發生的是：有一部分預付費客戶。

I would call them, the higher value, more prime customers that we've seen trend-wise start engaging more directly with postpaid brands. Yes. We've long had an established small migration program where some of those higher value customers and a CLV accretive way for us would move from prepaid to postpaid. And I think as an industry and as T-Mobile, we're seeing some more direct engagement with postpaid.

我會稱他們為價值較高、更偏主流（prime）的客戶；從趨勢來看，我們看到他們開始更直接地與後付費（postpaid）品牌互動。是的。我們長期以來一直有一個既定的小型遷移計畫，讓其中一些高價值客戶以對我們 CLV（客戶終身價值）具增益的方式，從預付費轉到後付費。我認為，無論是整個產業還是 T-Mobile，我們都看到與後付費更直接的互動正在增加。

And so when I think about that small subset of prepaid customers, it's important probably to not delineate them between prepaid and postpaid, but think about it a little bit more fluidly. And so I step back and think about that in the context of the total postpaid service revenue growth, what we're seeing because the fundamental question for you would be, okay, is the postpaid service revenue growth healthy then? Are those accounts healthy then if you're seeing some of that direct engagement from pre into post across the industry and us. And I think you've seen both in our 3.5% full year ARPA guide increase as well as the 7% organic postpaid service revenue increase that it's a fabulous business that we're seeing in postpaid.

因此，當我想到那一小部分預付費客戶時，或許重要的是不要把他們嚴格劃分為預付費或後付費，而是更流動地看待。所以我退一步，將其放在整體後付費服務營收成長的脈絡中思考：你看到的是什麼？因為你根本的問題會是：好，那後付費服務營收成長是否健康？如果你看到整個產業以及我們這裡，確實有一些從預付費到後付費的直接互動，那些帳戶是否健康？我認為你已經從我們全年 ARPA 指引上調 3.5%，以及有機後付費服務營收成長 7% 看到了：我們在後付費看到的是一門非常出色的業務。

It's very clear. Thank you.

非常清楚。謝謝。

Thanks, mike. Operator, next question, please.

謝謝，Mike。接線員，請下一題。

Sebastiano Petti, JP morgan.

JP Morgan 的 Sebastiano Petti。

Thank you for taking the question. Just a quick follow-up. I guess, Peter, on your response to John Hodulik's earlier question on just seasonality in the fourth quarter. I mean, should we anticipate that there is in addition to typical seasonality on churn, should we anticipate that there is some level of maybe upward pressure -- residual upward pressure related to the rate plan modernization within the fourth quarter? That's my first question.

謝謝讓我提問。我想快速追問一下。Peter，我想就你先前回覆 John Hodulik 關於第四季季節性的問題。我的意思是，除了流失率（churn）的典型季節性之外，我們是否應該預期第四季還會有某種程度的上行壓力——與資費方案現代化（rate plan modernization）相關的殘餘上行壓力？這是我的第一個問題。

Okay. Go ahead, ask your other one.

好的。你繼續，把另一個也問了吧。

And then yes, let me just get out of the way. Sorry about that. And then as you think about, obviously, lots of focus on converged footprints and fixed footprint. I mean has your view on the merits of incremental fiber M&A or accelerating the fiber build through additional on-balance sheet investment? Has that changed at all, particularly, I mean it seems like some of the spectral efficiency gains you're talking about maybe suggested there's more run rate to FWA above and beyond the $15 million? Just trying to discern between those two? Thank you.

好的，那我就一次問完。抱歉。另外，當你思考——顯然市場非常關注融合覆蓋（converged footprints）與固網覆蓋（fixed footprint）。你對於額外的光纖併購（fiber M&A）的優點，或是透過更多資產負債表內投資來加速光纖建設（fiber build），這個看法有改變嗎？特別是，我的意思是，你提到的一些頻譜效率提升（spectral efficiency gains），似乎暗示 FWA（固定無線接入）在 1,500 萬（$15 million）以上可能還有更多的穩態空間（run rate）。我只是想分辨這兩者之間的取捨。謝謝。

Maybe I'll take the fiber M&A question, and Peter, the other one. I'll come back to -- we don't believe in convergence when you define it as there is a very strict form of convergence, which is fixed line, along with wireless. We're huge believers in bundling, which is multiple line phone accounts, which is FWA with our phone account, which is more and more often actually our credit card and which can also be fiber with that. It's just one form of bundling, right.

也許我來回答光纖併購的問題，Peter 回答另一個。我先回到——我們不相信你所定義的那種融合，也就是一種非常嚴格形式的融合：固網加上無線。我們非常相信「綑綁」（bundling）：例如多門號手機帳戶、把 FWA 與我們的手機帳戶綑綁、而且越來越常見的是也把我們的信用卡納入，當然也可以把光纖一起綑綁。這只是綑綁的一種形式，對吧。

There is no panacea convergence nor do we see a sharply differential churn benefit other than having more relationships. In terms of -- and that sort of informs our view of fiber because we don't approach fiber from a -- we need this because of convergence. We look at fiber as a stand-alone business based on the return on equity that it lands up giving us. And so we will be thoughtful and choiceful and not chase a number of homes passed. We'll look at it from a very rigorous return on capital employed. And what is this land up, meaning in terms of accretion to equity value before we head into those. So if those deals do show up, we're very interested in them.

並不存在什麼萬靈丹式的融合（convergence），而且除了擁有更多關係（relationships）之外，我們也看不到顯著差異化的流失率改善效益。就——這也在某種程度上形塑了我們對光纖的看法，因為我們不會以「因為融合所以需要光纖」的角度來看待光纖。我們把光纖視為一個獨立業務，取決於它最終能帶給我們的股東權益報酬率（return on equity）。因此我們會審慎且有選擇性，不會去追逐某個「覆蓋住戶數」（homes passed）的數字。我們會以非常嚴謹的投入資本報酬率（return on capital employed）來評估。以及在我們投入之前，這最終會如何落地——也就是對股東權益價值（equity value）的增益（accretion）會是什麼。所以如果有這類交易出現，我們會非常有興趣。

What we've seen is, typically, these tend to be first to fiber, near first to fiber type deals. They tend to be specific footprints and we're always interested in them. And we've seen and shown that with the existing acquisitions, the T-Mobile brand and the T-Mobile distribution adds more value to it. So if those do show up, we will play, but we're not going to go chase a certain number of homes passed.

我們看到的是，通常這些交易往往屬於「第一批建光纖」或「接近第一批建光纖」的類型。它們通常是特定的覆蓋區域（footprints），我們一直都感興趣。而且我們也已經從既有的收購中看到並證明：T-Mobile 品牌與 T-Mobile 的通路（distribution）能為其帶來更多價值。所以如果這些交易出現，我們會參與，但我們不會去追逐某個特定的覆蓋住戶數。

Great. Yes. On the churn question, so as a follow-up, I mean there's, of course, the regular seasonality, you'll see the kind of the math, as I said, of just the makeup of postpaid accounts and 5G broadband and fiber only accounts churning predictably as they always do at a higher level structurally than postpaid phone churn itself.

很好。是的。關於流失率（churn）問題，作為追問，我的意思是，當然有一般的季節性；你會看到我剛才提到的那種「數學」——也就是後付費帳戶、5G 寬頻以及僅光纖帳戶的組成結構，會如同以往一樣可預期地以結構性較高的水準流失，相較於後付費手機本身的流失率。

And then you're right, there is a little bit of -- we did mention there will be some residual, typically when you see something like a rate plan optimization it's a couple quarter impact, although the Q4 impact will be much more muted. But we do anticipate a little bit of impact in Q4. All of that is contemplated in the guide that we gave you, though.

然後你說得對，確實會有一點——我們也提過，通常當你看到像資費方案優化（rate plan optimization）這樣的事情時，會有幾個季度的影響，雖然第四季的影響會小很多。但我們確實預期第四季會有一些影響。不過，這些都已經納入我們給你的指引之中。

Thanks, Peter. That's all the time we have for questions today. Thanks, everybody, for joining us. We look forward to connecting with you soon. For any additional questions, please reach out to the Investor Relations or media department. Thank you so much.

謝謝你，Peter。今天的提問時間就到這裡。謝謝各位的參與。我們期待很快再與各位交流。如有任何其他問題，請聯繫投資人關係部門或媒體部門。非常感謝。