AT&T (T) Q3 2026 Earnings Call: Fiber Growth and Higher EPS Outlook

Core Takeaways

• AT&T’s consolidated service revenue increased 3.8% year over year, while adjusted EBITDA rose 5.4% and adjusted EPS advanced 20% to $0.65.
• The company added more than 1 million fiber, fixed wireless and postpaid phone subscribers for the second consecutive quarter. Advanced Home Internet recorded 696,000 net additions.
• Advanced Connectivity service revenue grew 6.1%, supported by 5% wireless service revenue growth and more than 23% growth in Advanced Home Internet service revenue.
• Management raised its full-year adjusted EPS outlook to approximately $2.40 and now expects adjusted EBITDA growth at the upper end of its 3% to 4% range.
• Full-year free cash flow guidance remains above $18 billion. AT&T expects approximately $6 billion of free cash flow in the fourth quarter.
• Net debt to adjusted EBITDA increased to 3.13x following the $23 billion EchoStar spectrum purchase. Management expects leverage to return to approximately 2.5x during 2029.

Key Financial Data

AT&T recorded a $2.6 billion noncash impairment charge related to the planned sale of AT&T Mexico and an approximately $800 million employee separation charge, with the cash impact expected primarily in 2027.

Business and Operating Performance

Advanced Connectivity remained AT&T’s main growth engine. Segment service revenue increased 6.1% year over year, while EBITDA rose 7.3%.

Wireless service revenue grew 5%, supported by subscriber growth and a 4% increase in postpaid phone ARPU. AT&T added 379,000 postpaid phone subscribers, while postpaid phone churn remained stable at 0.92%.

Advanced Home Internet service revenue increased more than 23%. The company added 696,000 customers, comprising 416,000 AT&T Fiber additions and 280,000 Internet Air additions. Management attributed the record combined additions to expanded fiber coverage, deeper mid-band spectrum holdings, seasonal demand and introductory pricing.

Fiber ARPU declined 4% year over year. AT&T cited introductory pricing, converged-account discounts and growth in lower-ARPU Quantum customers within acquired Lumen territories. Management said its priority is total Advanced Connectivity service revenue and EBITDA growth rather than optimizing a single operating metric.

Nearly 43% of Advanced Home Internet customers also had an AT&T postpaid wireless account. The company added approximately 350,000 converged customer relationships during the quarter. Converged penetration in the acquired Lumen territories has increased by 250 basis points since the transaction closed in early February.

Business Advanced Connectivity service revenue grew 2.5%, while Fiber and Advanced Connectivity service revenue within the business operation increased nearly 9%. Management highlighted improved sales execution and stronger demand for secure, high-capacity connectivity.

Legacy service revenue declined approximately 25%, and legacy EBITDA fell about 27%, reflecting copper-service retirement and customer migration to newer technologies. AT&T now has more than 100 wire centers with no customers supported by legacy services.

Management Guidance

• Full-year consolidated service revenue is expected to grow in the low-single-digit range.
• Full-year adjusted EBITDA growth is now expected at the upper end of the 3% to 4% range.
• Full-year adjusted EPS is expected to be approximately $2.40, compared with the previous $2.25 to $2.35 range. Management attributed the increase primarily to lower depreciation expense from held-for-sale accounting for AT&T Mexico.
• Full-year free cash flow guidance remains above $18 billion, with capital investment of $23 billion to $24 billion.
• Fourth-quarter free cash flow is expected to be approximately $6 billion, versus $4.2 billion a year earlier.
• Advanced Connectivity service revenue and EBITDA are expected to grow by more than 5% and 6%, respectively, for the full year.
• AT&T expects fourth-quarter fiber ARPU to decline at a rate similar to Q3. Over the longer guidance period, management continues to expect relatively stable ARPU, with most growth driven by customer additions.
• Capital investment is expected to total $22 billion to $23 billion annually in 2027 and 2028, approximately $1 billion below the prior plan. The company still targets more than 60 million fiber locations by the end of the decade.
• Management reiterated its expectation for more than $19 billion of free cash flow in 2027.

Risks and Points to Watch

Fourth-quarter adjusted EBITDA growth will face higher marketing and customer acquisition costs from an extended device launch cycle and expanded fiber sales activity. Unfavorable vendor dispute settlements of approximately $100 million will primarily affect the legacy segment.

Higher device costs create a working-capital burden because AT&T pays suppliers upfront and collects from customers over time. Higher gross debt and interest rates are also increasing cash interest expense.

Advanced Home Internet additions are expected to moderate during the fourth quarter because of seasonally weaker holiday demand. Fiber ARPU remains pressured by introductory offers, convergence discounts and the acquired Quantum customer mix.

Leverage increased following the EchoStar spectrum acquisition. AT&T’s path back toward its approximately 2.5x target depends on adjusted EBITDA growth, debt repayment, excess cash flow and proceeds from the fiber joint venture and planned AT&T Mexico sale.

Analyst Q&A Highlights

Fiber penetration and wholesale strategy: AT&T evaluates fiber investments using a 40% terminal penetration assumption, but management believes penetration can exceed that level and continues to view 50% as achievable. Wholesale access could eventually be used selectively by geography after higher-value retail opportunities have been pursued.

LEO satellite competition: Management views low-Earth-orbit connectivity as useful at network edges and in less densely populated areas. However, it argued that satellite service does not replicate the indoor coverage, capacity and performance of a scaled terrestrial network. AT&T said 98% of traffic occurs where it has already built network capabilities.

AI connectivity opportunity: AT&T is seeing higher cloud workloads and demand for capacity, latency management and consistent performance across fixed, mobile and multi-cloud networks. The company is working with AWS and Microsoft on high-capacity fiber connectivity and sees future demand from inference, robotics and warehouse automation.

Fiber acquisitions: Management said it would consider acquiring additional fiber assets if pricing, network standards and geographic fit support AT&T’s convergence strategy and local operating scale.

AI agents and customer switching: Management said customers should be able to authorize agents to manage accounts, subject to identity verification, account controls and transaction authentication. AT&T expects product quality, value and customer trust to remain central as agent-based purchasing develops.

Full Earnings Call Transcript

Complete Earnings Call Transcript

Management Remarks

Operator

Good morning, and welcome to AT&T's Third Quarter 2026 Earnings Call. [Operator Instructions] And as a reminder, this conference is being recorded.

I would now like to turn the conference call over to our host, Brett Feldman, Treasurer and Head of Investor Relations. Please go ahead.

Brett Feldman

Thank you, and good morning. Welcome to our third quarter call. I'm Brett Feldman, Treasurer and Head of Investor Relations for AT&T. Joining me on the call today are John Stankey, our Chairman and CEO; and Pascal Desroches, our CFO. We're also joined today by our Deputy CFO, Jennifer Biry.

Before we begin, I need to call your attention to our safe harbor statement. It says that some of our comments today may be forward-looking. As such, they are subject to risks and uncertainties described in AT&T's SEC filings. Results may differ materially. Additional information as well as our earnings materials are available on the Investor Relations website.

With that, I'll turn things over to John.

John Stankey

Thanks, Brett, and good morning, everyone. I appreciate you joining us today. We continued the momentum we established in the first half of the year, delivering another strong quarter that reflects durable growth from our investment-led strategy.

During the quarter, we added over 1 million advanced connectivity subscribers from fiber, fixed wireless and postpaid phone, our best quarter in more than 4 years and our second consecutive quarter with more than 1 million net adds across these services. These are high-quality customer relationships and we continue to see growth in accounts that select AT&T for both their wireless and Internet connectivity.

Our improved gains in subscribers and converged accounts are driving strong financial results. In the third quarter, we achieved 3.8% year-over-year growth in consolidated service revenue, our best performance in more than 4 years. It was also our second consecutive quarter with adjusted EBITDA growth above 5% and adjusted EPS growth above 20%.

Our strong operating and financial performance reflects balanced growth across our portfolio of 5G and fiber services in both the consumer and business sectors. We're achieving these results through consistent execution against our core priorities, which are growing converged customer relationships, operating with greater efficiency and investing for durable growth and attractive returns.

At the end of the third quarter, nearly 43% of our Advanced Home Internet customers also had a postpaid wireless account with AT&T. We added approximately 350,000 customer relationships to our base of converged accounts in the third quarter, which is a faster pace of sequential growth than we achieved in the second quarter. This includes momentum in territories we acquired from Lumen, where converged penetration has increased by 250 basis points since we closed the transaction in early February.

Our fiber reach has never been greater. Our spectrum position has never been stronger. And we're reaching into this large underpenetrated opportunity with new plans and pricing strategies designed to bring more converged customers onto our network.

As we expand our fiber reach, we're using tiered introductory pricing to build penetration and anchor the home early across a wide range of customer cohorts that we can grow our business with over time, by attaching wireless and other high-value services and features. With fiber, this is a differentiated and effective go-to-market strategy that we're able to execute efficiently as a result of our local operating scale and owners' economics across 5G and fiber.

We're also removing friction for consumers by constructing plans built around simplicity, transparent value and convergence by investing in digital experiences that allow customers to do business with us the way they want to, and by offering peace of mind through America's Best Guarantee. By continuing to put our service and customer experience at the forefront of our customer value proposition and shunning uneconomical device offers, we are well positioned to drive durable converged relationships.

We saw improved traction with business accounts in the quarter and achieved our second consecutive quarter of year-over-year growth in business Advanced Connectivity service revenues. Our performance reflects improved sales execution across direct, indirect and digital channels, greater penetration of fiber and fixed wireless, and better positioning with business customers that need reliable, secure and high-capacity converged connectivity across multiple locations.

Our broad and deep fiber network also has us well positioned to be a leader in AI-grade connectivity. Over 90% of mapped U.S. data centers are within 10 miles of our fiber and over 80% are within 1 mile. This is one of the reasons why the nation's leading hyperscalers are working with AT&T to support their rapidly-growing high-capacity connectivity needs. This includes AWS, who is using AT&T to connect their locations with high-capacity fiber. We're also working with Microsoft to support its high capacity needs as AI growth drives demand. These moves are simply foundational to drive sustainable value by layering high-margin services and network intelligence on top of that fiber to connect growing AI workloads across the entire ecosystem, including business endpoints, metro aggregation and managed multi-vendor cloud ingress and egress.

I've said this before, but it's worth repeating. The strength of our performance comes from the structural advantages we've established after years of consistent and targeted investment. In the dense urban and suburban areas where we deploy both fiber and 5G infrastructure at scale, we believe that our competitors cannot and will not match our operating proficiency, network performance or exceptional converged value proposition. By the end of this decade, we expect to operate the most advanced and technologically open communications network in the U.S. built on a foundation of dense interconnected metro fiber and a nationwide wireless network, that I should mention already functions at scale indoors and out. This asset base puts us in a leading position to serve converged consumer and business customers today and well into the future as AI becomes a key driver of their connectivity needs and requires ever-improving high-performance networking.

Over the past few weeks, we've announced several agreements that further advance our goal of being the leading provider of advanced connectivity in the U.S. This includes the launch of our joint venture to help close coverage gaps by making satellite-enabled connectivity easier to deliver where traditional service is limited. We announced an agreement with Global Infrastructure Partners, a part of BlackRock, and CPP Investments to form a JV that includes GigaPower and the fiber assets we acquired from Lumen, in order to accelerate fiber deployment outside of our traditional operating territories.

We also reached a multiyear agreement with Corning to supply the fiber we need to support our goal of reaching 60 million plus fiber locations by the end of this decade. We signed an agreement to sell our wireless business in Mexico. Combined with our industry-leading capital investment and recent acquisitions of spectrum from EchoStar and fiber from Lumen, these agreements sharpen our U.S. investment focus, strengthen our convergence strategy and position AT&T to have the leading share in advanced connectivity service revenue by the end of the decade.

On the cost and simplicity side, we continue to make significant progress in our work to exit less efficient copper-based legacy services. We're putting our customers at the center of this transition and we're working with them to upgrade to newer, more reliable services like Phone Advanced, Fiber and Internet Air. Our progress here is picking up momentum. We now have more than 100 wire centers with no customers supported by legacy services. This is an industry-first at scale, and we're on track to reach our target of discontinuing copper-based services in 50% of our wire centers outside of California by the end of 2027 and substantially all of them by the end of 2029.

We still have a lot of work to do here, but we've put ourselves on a path to remove significant cost and complexity as we remove outdated technologies from service. As we accelerate this work, we believe we have a meaningful medium-term opportunity to monetize the copper embedded in our legacy network.

As I close, I think many of you are aware that this will be Pascal's final quarterly report. I know he'll tell you that his tenure has been a relaxing walk in the park and that nearly 6 years has raced by, but it's fair to say much has been expected of him over this period. It's also fair to say he delivered on those expectations. I could not have asked for a better partner and leader to support the transformation of AT&T's business model. His story is an inspiring story of solid values, quiet determination, respect and care for others, and a daily focus to solve problems on behalf of shareholders and the company. I know I speak for the broader team at AT&T when I say we'll miss his daily presence, but we'll be forever grateful for his leadership and outsized contributions.

And now for the final time, Pascal, over to you.

Pascal Desroches

Thank you, John, and good morning, everyone. At a consolidated level, total revenues in the third quarter was up 5.3% year-over-year, with service revenues growing 3.8%. Adjusted EBITDA grew 5.4% year-over-year and we achieved an adjusted EBITDA margin of 38.7%. This is our highest third quarter consolidated adjusted EBITDA margin since we refocused our business on Advanced Connectivity at the beginning of this decade. And we achieved this record despite 12% growth in equipment revenue, which is dilutive to margins. This highlights the continued improvement in our underlying operating leverage as we grow our base of converged customers and achieve efficiencies from our transformation initiatives.

For the full year, we continue to expect consolidated service revenues to grow in the low single-digit range and now expect consolidated adjusted EBITDA to grow in the upper portion of the 3% to 4% range. During the fourth quarter, we expect our adjusted EBITDA growth will be impacted by higher marketing and other customer acquisition costs compared to last year, driven by an extended device launch cycle and our expanded fiber go-to-market opportunity, as well as unfavorable vendor dispute settlements of approximately $100 million that will primarily impact our legacy segment.

Adjusted EPS of $0.65 was up 20% from $0.54 in the third quarter of 2025, driven primarily by growth in adjusted EBITDA and lower depreciation expense. Our third quarter results reflect 2 notable adjustments, including a $2.6 billion noncash impairment charge related to our agreement to sell AT&T Mexico and a roughly $800 million employee separation charge that will impact cash primarily in 2027. For the full year, we now expect adjusted EPS in the $2.40 range, compared to our prior outlook of $2.25 to $2.35, primarily because of lower depreciation expense related to held-for-sale accounting for AT&T Mexico.

Capital investment was $5.6 billion in the third quarter, up roughly $300 million year-over-year. Consistent with our outlook, free cash flow was approximately $5 billion, essentially stable compared to the third quarter of 2025. For the fourth quarter, we expect free cash flow of approximately $6 billion, compared to $4.2 billion in the fourth quarter of last year. This primarily reflects an outlook for higher cash from operations compared to last year when our fourth quarter results were impacted by discrete items, including a large contribution to our pension and legal settlements. For the full year, we continue to expect $18 billion plus of free cash flow, with $23 billion to $24 billion of capital investment.

We now expect cash taxes in the $750 million range for the year, compared to our prior outlook of $1 billion to $1.5 billion, as we continue to benefit from the provisions in last year's One Big Beautiful Bill Act. Compared to our initial outlook, this is offset by higher cash interest payments, which primarily reflects higher gross debt and interest rates as well as increased investments in working capital driven primarily by higher upfront spending on devices and other customer acquisition costs as we drive higher levels of customer additions across our Advanced Connectivity services.

In our earnings release, we reiterated our long-term outlook for consolidated growth in service revenues, adjusted EBITDA, adjusted EPS as well as our outlook for free cash flow through 2028. As a reminder, our long-term outlook is presented on a continuing operations basis and excludes our Mexico wireless business beginning in 2027.

As we continue to extend our industry-leading scale in fiber, we are becoming more effective in how we plan, build, install and calibrate the pace of our network deployment. We believe this positions us to reach 60 million plus locations with fiber by the end of the decade with annual capital investment that is roughly $1 billion lower than we previously planned. For 2027 and 2028, this represents annual capital investment in the $22 billion to $23 billion range.

Based on the attractive opportunity we see to grow fiber and converge customers as we expand our fiber reach, we expect to reallocate this capital into the business, primarily through working capital in support of customer acquisition investments as well as higher interest payment. We will provide more color on the drivers that support our outlook for $19 billion plus of free cash flow in 2027 as part of our fourth quarter earnings report early next year.

Our consolidated growth continues to be driven by momentum within our Advanced Connectivity segment. Advanced Connectivity service revenues grew 6.1% year-over-year with EBITDA increasing 7.3%. Wireless service revenues grew 5% year-over-year, driven primarily by growth in our subscriber base and a 4% growth in postpaid phone ARPU. This reflects great execution of our wireless pricing strategy, including the impact of our new and refreshed rate plans, as well as our consistent approach to managing back book pricing across the base. We remain focused on adding the right customers on the right plans with the right offers. This approach resulted in 379,000 postpaid phone net adds during a competitive third quarter, with stable year-over-year postpaid phone churn of 0.92% and strong margin contribution.

Advanced Home Internet service revenues grew more than 23% year-over-year in the third quarter, driven by the increase in our customer base. We achieved 696,000 Advanced Home Internet net adds in the third quarter, including 416,000 AT&T Fiber net adds and 280,000 Internet Air net adds. This was a record quarter for combined net adds across these services, driven by the significant increase in our fiber reach, our deeper mid-band spectrum position, positive seasonal dynamics and strong uptake of our introductory pricing. During the fourth quarter, we expect that our Advanced Home net adds will moderate as we typically see lower demand in this category during the holiday season.

Fiber ARPU declined 4% year-over-year in the third quarter. This reflects a range of factors, including traction with our introductory pricing, our growing base of converged accounts that earn discounts, and growth in lower-ARPU fiber customers on the Quantum brand in territories we acquired from Lumen.

In the near term, we expect fiber ARPU will continue to be impacted by these dynamics, including during the fourth quarter when we anticipate a similar year-over-year decline compared to our third quarter results. But as we've discussed before, our goal is to optimize growth in total Advanced Connectivity service revenue and EBITDA, not specific operating metrics. And we believe that our third quarter results highlight how we are improving overall margins while driving balanced and profitable growth across our portfolio. This includes another quarter of improved momentum in Business Advanced Connectivity service revenues, which grew 2.5% year-over-year with Fiber and Advanced Connectivity service revenues growing nearly 9%.

We achieved solid growth across wireless, fiber and fixed wireless, which is now consistently offsetting structural pressures on transitional services such as VPN and wholesale. We exited the third quarter with great operating momentum and we continue to expect full year Advanced Connectivity service revenue growth of 5% plus and EBITDA growth of 6% plus.

In our legacy segment, service revenues declined approximately 25% year-over-year, consistent with our expectation for continued 20% plus declines as we power down copper-based services and migrate customers to more advanced and reliable technologies. Legacy EBITDA declined about 27%, and we continue to make progress taking cost out of the business as we ramp the retirement of legacy wire centers. As I noted earlier, our fourth quarter results will be impacted by unfavorable vendor dispute settlements that will be primarily reported within the legacy segment. Copper retirement remains an integral part of our overall cost transformation, and we remain on pace to achieve our target of $4 billion in annual consolidated cost savings by the end of 2028.

We ended the third quarter with net debt to adjusted EBITDA of 3.13x, up from 2.68x in the second quarter, which reflects our $23 billion purchase of spectrum licenses from EchoStar at the end of July. The transaction materially deepened our low and mid-band spectrum holdings and puts us in a great position to execute our strategy. We continue to expect net debt to adjusted EBITDA to return to a level consistent with our target in the 2.5x range during 2029. This will be achieved through growth in adjusted EBITDA and debt repayment using excess cash flow and proceeds in 2027 from closing our fiber JV and the planned sale of our Mexico wireless business.

We returned $4.8 billion to shareholders during the third quarter, including $2.8 billion of share repurchases. Through the end of the third quarter, we repurchased over $7 billion of stock this year, and we continue to target approximately $10 billion of share repurchases in 2026.

Before I move on to the Q&A, I want to take a moment to thank everyone on the call. Over the years, your thoughtful questions and candid perspective have sharpened my thinking and made every conversation more valuable. I'm grateful for your engagement. I'm really proud of the work we have done here over the past 5-plus years. And as a shareholder, I'm very pleased that Jennifer will be stepping into the CFO role after the new year.

Brett, we're now ready for the Q&A.

Brett Feldman

Thank you, Pascal. Operator, we are ready to take the first question.

Operator

[Operator Instructions] The first question today comes from John Hodulik with UBS.

Question-and-Answer Session

John Hodulik

Great. And before we begin, I just want to say congrats to Pascal on the retirement. It's been great working with you. And welcome to Jennifer.

Maybe for John, if we could start first with the fiber strategy. Could you talk a little bit about your sort of goals for -- in terms of penetration rates? You have been talking about a 40% penetration rate. And now I think you have aspirations to get to 50%. Just based on what you're seeing in the market and from competition and fixed wireless penetration, is 50% still sort of the expectation? And could it go higher? And maybe what are you seeing in the Lumen markets?

And then as a follow-up, John, to your commentary about the signal -- the wireless signal radiating internally. Obviously, a lot of questions about the Starlink mobile strategy. Just can you -- any high-level thoughts on the use of LEO constellation to provide mobile service, and maybe what it would take to build out a terrestrial footprint to complement that service and match what you get from AT&T from a mobile standpoint?

John Stankey

John, so our fiber actually -- my fiber goals would be to have 100% penetration, but it may take a little while to get there. I don't think we have to step back and say we have limitations to this over time. We clearly, when we think about how we invest in an area, we do our economics assuming we can get to a terminal penetration rate of 40%. And we don't want to go invest any place where we don't think we can get adequate returns at that level. That's not what our internal objectives are. We believe we can exceed that level.

What I've shared with you before is we have a pretty straightforward process that gets us from 0 to 40% in terms of tactics and timing. The 40% to 50% range is a little bit different. It takes a little bit more time, takes a variety of different tactics, and it comes with persistence and consistency. And frankly, it will lean a lot on us being very successful on more converged customers to get there. And so still fully believe that's achievable. And I would expect that as we go forward, you've heard me say that we have a lot of different plays that we can run. Obviously, in all markets, I'd love to be able to get to that 50% penetration with a direct retail relationship. But I've also shared with you that in this company and in this industry, it is not beyond any expectation that you wouldn't have some wholesale at some point in time.

And we got a lot of capacity on that network. We could literally handle 100% of the customers in an area if they all came to us. So I want to make sure that we're putting all that capacity to use over the long haul. And if we hit a place in particular markets, for example, where we think we've reached our ability to penetrate under our brand, as this matures over time, is it possible that we might open up the network and have others sell on top of it to drive penetration and returns up? I wouldn't rule that out over the long haul. So I think there's more innings to play in this game as we move forward.

The Lumen markets, as you know and we shared when we acquired the asset, were underpenetrated comparatively to what AT&T markets might have been for their same build tenure. And so what we've been doing, as you're aware, is trying to convert all those markets to AT&T branded fiber. We're just about through that process. We have a couple left to go to finish that up. As we convert those markets to the AT&T branded fiber, as we put the business processes in place that we have honed in our other operating regions and apply them to those markets, we're seeing penetration march up.

That penetration is marching up in a manner that's consistent with the business case that we outlined in our goals. I think we shared with you when we did the transaction that we expected the terminal penetration levels of our out-of-region markets to be somewhat lower than the terminal penetration levels of our embedded traditional markets. That again was done to put some conservatism in the business case.

With what I'm seeing right now, maybe that is a little too conservative. We'll see as time marches on. And as I shared in my remarks, the other advantage we're seeing is increasing in the convergence rate, we're up 250 basis points since the March, end of February acquisition. And that is part of the reason you're seeing the strong margin performance in the wireless business as well. So I feel really good about where those are and where we're going to go over the long haul.

In terms of what it takes to build a converged -- scaled converged network, which is what I believe everybody is marching into here and what some of the new entrants are talking about needing to do, it takes a lot. I may be not the best person to reference on that. I might -- next time you bump into Charlie, you should talk to him and ask him what it took and how many years and complications that come up in doing those types of things. It's not something that gets done overnight.

I wouldn't over-rotate on what I would call 1 carrier of low-band spectrum. Let's put this in perspective. In our scaled network, we run, depending on the metro, somewhere between 6 to 8 carriers of low-band spectrum. In addition, that spectrum is not just housed on macro antenna sites. It exists on every indoor building system and high-rises in hospitals and stadiums. I've shared with you that you can go into a metropolitan area like Chicago and look at the traffic on the network, and somewhere between 40% to 45% of the traffic on any given day is traffic that's originated from inside of a building. And as a result of that, how you perform indoors, of course, is extremely important. And we shouldn't think about those as being nominal loads. They're significant loads even with robust WiFi networks that are out there.

So I think it takes a lot of years and a huge level of investment to get to the kind of performance levels that customers have become accustomed to in scaled networks that are running on the infrastructure that most of the incumbent providers have today. So I think there's going to be continued advances. We're certainly excited about the continued advances in LEO technology and how it might help us at the edges of our network and preventing folks from walking off the network, how it might help us in places where we have less density and maybe we can lower our cost structure, rather than putting less-used macro sites in place, serving those differently.

But we've got the scale where the scale matters. 98% of the traffic is where we've already built and where we already have capabilities, and customers have become accustomed to a level of service that is pretty high. And rightfully so, they pay us a lot for what they get from us and they should expect it to work.

So that's my point of view on it. I feel really good about where we're at. I feel really good that we're going to have the best converged set of assets out there, both with fiber and our robust wireless network to meet those expectations moving forward, especially in the upstream.

Operator

The next question comes from Michael Rollins with Citi.

Michael Rollins

I also want to send my congratulations to Pascal on your upcoming retirement, and a welcome to Jennifer. Two questions, if I could. So first, if you drill down further on the broadband ARPU, and can you share what you're seeing in terms of the performance of broadband ARPUs and revenue in your more established mature fiber markets relative to the expansion markets? And do you see broadband returning to some degree of annual pricing power, whether it's through the speed, [ share ] mix, future upsells or just simply changes in retail pricing over time?

And then secondly, John, you referred to some of the AI opportunities earlier for AT&T. And just curious, what are the signals that you're currently seeing in the network from enterprises, hyperscalers and consumers in terms of early uptake of data traffic and consumption and how customers may be engaging with you differently in terms of how they're looking to invest in connectivity with AT&T?

John Stankey

Michael, so the way I think about broadband ARPU is what we want is a portfolio of customers that basically buy from us at all discrete price points. And I think you've heard us talk about the fact that we've been somewhat underpenetrated in maybe more of the value segments. That's true both in our wireless business as well as our fixed business. And I see no reason that, especially when you're building a network that has the kind of capacity that a fiber network has, that we shouldn't think about we want to play in every segment. We just want to play in those segments at the right price point.

And I think in our, what I would call our more mature markets, if you went in and looked at them, what you would see is a very nice distribution of folks that are at higher price points because we went and marketed largely as premium, and it was premium. And we're now coming back and doing the cleanup on convergence to pick up additional households and, in some cases, we're going in with different price points and a different performing product. We want to make sure that the price point is matched to the performance of the particular product, think rate and speed. And we think that getting that consolidated account in makes sense. And as a result of that, you get some deterioration in average ARPUs.

And I don't want to focus on that exclusively. What we want to focus on is the fact that you're seeing great service revenue growth in our business. You're seeing great customer account growth. You're seeing a margin equation that is fantastic, in my view, based on the kind of growth we're going through. And when I have that kind of equation showing up, that's how I want the team to move forward and how we should think about these things.

And in terms of your point -- your question on what happens in the future, I think what happens in the future is what's happened in the past. That is, you have those relationships and they're good and customers like the product and they stay with you, and we have every reason to believe that a converged customer that's sitting on fiber is going to be a sticky and high lifetime value customer. We shared that with you. They're going to naturally, based on their needs and capabilities that they desire, they'll probably buy up over time.

And so just because we bring somebody in at a more value-oriented price point early on and we can offer at a price point that matches fixed wireless a better performing product, I think the future is bright. Because at some point, that household is going to take advantage of the things that you just referenced, like more capabilities around AI or they have more children or whatever it might be that cause them to say "I need to tick up to a higher speed tier," and we'll be able to do that without breaking a sweat because the network is there and can easily do that on a software adjustment, and we don't have to add incremental capacity in the infrastructure to do that. So I expect there's going to be good stability in that ARPU over time.

In the enterprise segment...

Pascal Desroches

Well, Mike, one other point I'd remind you is as recent as last December, we had a pretty meaningful price increase that impacted much of our fiber base. So consistent with John's response, we have a history of being able to manage the base of pricing across both products.

John Stankey

In the enterprise segment, on AI, what we're seeing right now is what I would call there's just increased workloads that are going on, mostly into the cloud, that people need capacity for to deal with. What the discussions are moving to is a reality that most people are in a hybrid environment where they're running traffic over both their fixed and mobile infrastructure, they're running traffic across multiple cloud layers. And what they need to be able to do is to ensure that certain applications that have higher requirements or performance are consistently managed across both those networks and across multiple cloud fabrics.

And so the discussions right now of the next application are how do we basically tag and mark performance level packets to ensure that particular applications that run in both environments can maintain a high level of service, that's consistent with what the performance is if it's an AI-based driven application or capability. Oftentimes, those have more stringent requirements on latency, round-trip performance. If you're doing inference, you need quicker response times. And people want that performance. And they're doing more of it, so they just need more share capacity. And that's part of what's driving the lift.

What I expect, based on other conversations that are going on is we, in fact, will start to see more applications pop up that are even more demanding. And when you start thinking about what happens in the robotics space, what happens as more autonomy increases at the warehouse levels, those things are going to drive another layer of requirements and performance. And when that happens, I believe there'll be probably another investment level that will be required. It will be managed networks that will be necessary given the performance requirements that are essential to allowing that to happen. And once you have that foundational infrastructure, that what I would call fiber-fed infrastructure that has the ability to track packets across both fixed and mobile and deliver a high-quality service, I think you're going to be in really good shape.

I just might add for clarity, the reference I made to some of the work that we're doing with the hyperscalers, where that fits into this picture is, as you know, we've invested heavily at the edge and endpoints both in our wireless business and our fixed business. We have a fantastic metro-to-metro infrastructure, as you might imagine, based on the nature of our network and how we operate. What we want to make sure is that our backbone in and out of those metros that hit those cloud fabrics, we have absolute visibility on how we do that ingress and egress.

And so by involving ourselves with the hyperscalers that are driving most of the workloads in that space, that makes sure that we have the right expressways or on-ramps in and out of their clouds, so that is our metro and our endpoint networks that we've built and invested in so heavily can have that end-to-end visibility and end-to-end performance. And we want to do enough work in the long haul, and then the data center -- the data center space, and partner in the right ways where we ensure we have that right infrastructure in place and we're at the right points of presence to be able to offer that back to customers.

Operator

The next question comes from David Barden with New Street Research.

David Barden

Pascal, it's been a pleasure. And welcome to Jennifer as well. So I guess, John, I agree with you, I think that the idea that SpaceX and Elon are going to invest a tremendous amount of resources to take a business that's very well functioning and tear it down is probably the wrong way to think about the ambitions in mobile, which is to kind of create new connectivity opportunities in AI and robotics and cars and all these things. And so I think that maybe the market is getting that wrong.

But I do think that there is this question about their ability to address the broadband industry, maybe at the value end of the market where fixed wireless access has been so successful and how that presence might alter your thought process around fixed wireless access, its growth opportunity, its role in nurturing how you build and bridge to the fiber business. If you could kind of elaborate a little bit on that, that would be great.

And then Pascal, just quickly, we threw the word working capital around quite a bit in the prepared remarks. It feels like the intention has been to reduce subsidies on handsets as a method of growing the wireless business. But as we reduce CapEx in fiber through efficiencies, we're reinvesting it in a combination of working capital interest expense. Could you elaborate a little bit on how this working capital investment is allocated to building the business?

John Stankey

So Dave, I would tell you this, that the most durable customer combination in our industry is a fiber-fed broadband customer with a great wireless service on top of it, bar none. It's going to be the most sustainable, longest lifetime value. In my view, when I talk about service revenue share leadership in the industry, the math and calculation on the footprint that we're building for fiber and how we think about expansion and our partnerships is all driven on a foundation of how do we get enough footprint enough converged services in a way that, that puts us in a share leadership position for consolidated service revenues moving forward. And over the long haul, whoever steps into my chair after me, that's, in my view, kind of the race you're running for, and those are going to be the most durable customers, whether they're consumers or businesses moving forward.

We have, as we always have in this industry, those transition times of technology. It's not to say that there wasn't money made off of digital subscriber lines and ADSL and other capabilities. But they are transient and they run their course if they're not the most sustainable technology. And to your point, I think fixed wireless access, as I've said, we'll have certain customer bases and applications that are going to make it a sustainable technology, but there are going to be other ways that it's used and are being used today that ultimately will be outstripped by investment in other and newer technologies.

As I've said, I think there's some -- I think LEO constellations are fantastic for delivering bandwidth to rural areas. And they may well be the most efficient and effective way to get bandwidth into less densely populated areas moving forward. Where there's a motivation to invest in fiber, it's going to be really hard for people to beat the consolidated value proposition of fiber and wireless, especially as the technology matures or the service capabilities around it mature to allow less friction for our customer, offer more effectiveness with it and make those products work even better together.

So I think over time, your point is correct that there's going to probably be some kind of a peak and then a downward trajectory of fixed wireless as those technologies and the better technologies settle into places. But that's not inconsistent with the set of expectations that we've been building in this company and that we've used to drive our guidance to you over the long haul. And it's not inconsistent with what I've articulated as this company's goal to be the service revenue leader in the industry as we exit this decade.

Pascal Desroches

Dave, you are correct in saying that the level of promotions remains the same. And the customer ultimately is going to bear the cost of higher device prices. With that said, there is a timing element. So generally, we pay for the devices within a relatively short term. So think about phones, tablets, gateways, we pay upfront and we collect over time, typically the life of the customer. Therefore, we are investing upfront and recovering it over time. So that is the working capital drain that I described in my script.

Also impacting cash is the disposition of our Mexico asset. And as I mentioned in my comments, higher cash interest, principally related to rates.

Operator

The next question comes from Sean Diffley with Morgan Stanley.

Sean Diffley

Congrats to Pascal on an awesome career. Two, if I may. The first on agentic and the second on CapEx and free cash flow outlook. So John, just an overarching question on how you're thinking about potential rise of AI agents like Muse impacting the connectivity sector or not. Do you see this as a deflationary force in the industry and a catalyst for switching? And then talk about the risk and opportunity that you see for AT&T specifically.

And then on CapEx, I think it came down by about $1 billion. For '27 and '28, you kept the free cash flow guide, the same. I think you talked about improved efficiency on some of the fiber builds. I think the Mexico sale might be in there. But anything else we should be thinking about on the CapEx and free cash flow swings?

John Stankey

Sean, so look, I think there's a lot that still is unpacked on what is technology in general, agentic search, agents, et cetera, going to do to everybody's funnel management. And I don't pretend to know exactly how that's going to play out. I will tell you, we're spending a lot of effort and time on it and there's many things we've been doing over the last several months and quarters to fine-tune our strategies on how we make sure we stay relevant.

I would tell you as different as it's going to be, what I'm encouraged by is we seem to be doing just fine in making the adjustments we need to make because there's been quite a bit of change already and how consumers are searching and looking for information and what they're doing around it. And I think we're probably in the early innings of that, but we're aware of it. We're adjusting to it.

I think at the end of the day, what's really important is you have good products and those good products have the right reputation and the right value in the market. And then ultimately, you will be successful as you move forward. And when you're operating in a position where you're putting technology out that's the lowest marginal cost technology that performs better, I think that puts us in a really good position to make sure we stay in that conversation moving forward.

Specifically to your question about Muse and some of the other chatter that's going on right now, what I would tell you is, philosophically, we believe the customer gets to make choices on what they do with their service. That's always been our point of view. We also believe customers need to dictate their path forward on things. Customers have been pretty clear with us over the past number of years about account security, account management, account control. Before agents ever showed up, there's all kinds of scenarios that go on today in real life that cause people to have to be really specific on who the decision-making authority is on their account. We run family plans and have many, many people on them. We've got people whose life circumstances change where there's a responsible party and there's maybe a separation or marital discord that changes things.

The notion of an account having an owner and a responsible party is not a new issue. As long as that responsible party is authorized and allowed to go out and make a transaction and can confirm that in the way that we would normally confirm those things, such as validating their identity and ensuring they're authorizing it, they should be able to do to their account whatever they want. If it's their agent that decides something needs to change, and they've authorized an agent to do transactions on their account, and we can authorize that transaction in the same way we would normally confirm such as when there's a SIM swap or a major change on an account today, we use things like 2-factor authentication to ensure that it's valid and confirmed, those same principles and practices will be in place moving forward with agents.

But the important thing is that the customer will control. They'll say whether or not their account is authorized to be managed by an agent. And then if there is a large transaction that takes place, there will need to be a validation and authentication of that to ensure that we're not doing something that encourages fraud or is inconsistent with what the account owner would want to do.

Over the long haul, my point of view, just like today, is we want to make sure the customers feel like they're getting value for what they're buying. We want to make sure they're rightsized to what they need in their products and services. We do that as a normal course of business when people come into our stores and talk about what their needs are or when they're working with us online and our algorithmic approach, I think agents are probably going to be another way for people to do that and be able to get insight as to what works for them. And I think over the long haul, with good products out there, treat people well, it will be fine for our business and we'll have accretion of new customers who find us who wouldn't have found us other ways. And we'll have some customers who choose to be a little bit more efficient how they buy from us.

Pascal, do you want to touch on the cash flow issue?

Pascal Desroches

Sure thing. On CapEx, here's the way to think about it, Sean. Overall, CapEx, we said, is going to be down about $1 billion. $300 million to $400 million of that is going to be Mexico wireless CapEx. The balance is from the efficiencies I described in my comments.

Think about this, we've been doing this for a very long time. And with that comes the advantage of scale and improve ability to get it done faster and cheaper. We have long-term contracts with partners we've been doing business for a very long time. All those things are allowing us to do -- to build a little bit more efficiently and effectively.

In terms of overall cash flow, the pieces to think about are the absence of Mexico probably costs you $400 million to $500 million annually. And beyond that, with higher device costs upfront that we're paying for and collecting over time, there will be a working capital investment that we have to make. Those are the piece parts along with the interest cost.

Operator

The next question comes from Craig Moffett with MoffettNathanson.

Craig Moffett

You talked about the fiber led strategy. Given the rise in interest rates, you've seen some of the smaller independent fiber builders starting to struggle. Brightspeed issuing a going concern warning and that sort of thing. Is there an opportunity for you to pick up attractively priced assets in a consolidation cycle given the rise in rates and what that does to some of the smaller independents?

John Stankey

Craig, yes, there are some -- I think part of the reason why we're comfortable where we are in our investment levels is one of the trends that you cited, which is, mistakenly, some people went into the space thinking that they could build little pockets of fiber in places and never get to scale in a metropolitan area, and be successful. And the reality is that's very hard to do. And in some cases, there's been maybe more concentrated builds with the business model on legacy services and other things hasn't been aggressively managed to come up with an overall profitable equation. And so we see the same thing.

I think I'd answer it the same way I've answered it in the past, tell me the price and I'll give you the answer to the question. And I think you've seen up to this point in time, I've been willing to go and acquire footprint and do things when I feel like we're getting a price that's an appropriate price. And we think if the infrastructure has been built in a way that's consistent with our standards in our practices that we can operate effectively over the long haul, and the price is right, sure, I'd be interested in picking up more footprint if it holds to my point of view of us getting scale in the right geography. I don't want to just buy a footprint for the sake of buying it. I need to make sure it bridges to other parts of our network where we can have the kind of efficiencies and scale on advertising and the growth in our wireless business and all the things that we think kind of are the magic and the economics around that.

Lumen clearly fit that bill across the footprint that we picked up, in my point of view. If other opportunities were like that, that were priced that way, showed up, sure, I'd look at them, is options and possibilities for us. Right now, I would say, in some cases, maybe there's expectations in that market that aren't consistent with what I just articulated.

Brett Feldman

Operator, I think we have time for one more question.

Operator

The next question comes from Sam McHugh with BNP.

Samuel McHugh

Congrats Pascal, and welcome, Jennifer. On the fiber ARPUs, helpful color on the fourth quarter. The '27 guidance, do you anticipate a similar decline in fiber ARPUs, as we're seeing at the moment?

And then secondly, on the wholesale fiber point, John, I think you raised this maybe a year or so ago first. Obviously, you lose a bit of kind of wholesale -- sorry, cross-sell potential in wireless if you wholesale. Would you look for reciprocal buybacks from others as a precondition of wholesaling fiber?

Pascal Desroches

The way to think about Q4 is, as I said, I expect Q4 to be down year-over-year around the same rate as we saw in Q3. That would imply like moderate -- a sequential moderation. Overall, the guidance that we've given, if you recall, over the 3-year period is to have most of the growth. The vast majority of our growth is going to be driven by customer additions and not ARPU -- not pricing. And that continues to be our expectation. So overall, we continue to expect relatively stable ARPU over this period.

John Stankey

Sam, on your question about wholesale, your question -- and first, you're thinking about it the right way, which is wholesale doesn't mean you have to open it up to all comers. Wholesale could mean that you wholesale it to those that are more strategically positioned to help you in aggregate. And an example of that would be potentially there's a wholesaler who wants to, for example, have an exclusive MVNO relationship with you, or maybe there's [ a paired ] footprint somewhere else outside of existing fiber footprint that we have that we gain access to that we wouldn't have otherwise.

And also, to your point, and as I indicated in my comments, you do this after you think you run all the plays that are your highest value plays at retail. So our goal is to do everything we can to pick up every high-value customer we can to serve at retail. And we're having great success doing that, as you're seeing with the results that we've been sharing with you. And we'll continue to run that play until we hit a place where we think we've topped out.

And I think what's important to understand is you don't have to make these decisions on a national basis. These can be very geographically driven. And I think as, over time, what we know, and I think what the cable industry would show you, is a lot of these decisions and a lot of your approaches to market are low, and they have to be. And really if you look at what we've recently done in reorganizing our business, I don't know how many of you picked up on it, we did a structural change in our consumer space in our go-to-market structure, the whole reason we did it is a realization that we're at a place where not all markets are created equal. There are differences in infrastructure. There are difference in tactics that need to be used in particular geographies of the country. Depending on what your asset base and the tactics you want to move forward, we needed to organize our business, our management teams or incentives in a way that drove that. And I think you're going to make those kind of decisions on a location by location and circumstance by circumstance basis.

Brett Feldman

All right. Thank you for the question. Operator, that was our last one. You can go ahead and close out the call.

Operator

Thank you. The conference has concluded. Thank you for attending today's presentation. You may now disconnect.