Lumen Technologies, Inc. ($LUMN)

Good afternoon. Welcome to day 1 of our 54th Annual TD Cowen TMT Conference. My name is Greg Williams. I cover cable, wireless, telco and fiber here at TD Cowen. Joining me this session is Chris Stansbury, CFO of Lumen Technologies. So Chris, thanks for joining us.

Thank you. Great to be here.

So Lumen has a lot on its plate, including recently closing the AT&T fiber-to-the-home deal, PCF execution now, ramping up digital revenue, the Alkira deal and cost savings initiatives. Where are your priorities at the moment?

Our priority is really on the inflection of the business. And so this year, we inflect EBITDA. There's a lot of focus on that. The cost savings help drive that before revenue inflects, but the revenue inflection is right behind that. And we're doing everything we can to try to accelerate that. As we said at our Investor Day, we expect revenue to inflect in 2028. I think -- at that time, Alkira was not contemplated in those economics. And so that -- we think that will accelerate our revenue growth, but we've got to dimensionalize that. But first, we got to close the deal. So -- but it's really about inflecting the business.

Okay. And looking forward to that. Part of the inflecting the business is PCF demand, at least from a cash flow perspective. You now have $13 billion in PCF deals to date. As a leading indicator for fiber demand, we're tracking data center leasing because these data centers are eventually going to need photons, if you will. And we saw another massive step-up last quarter, 9.4 gigawatts of leasing in the past quarter alone. So should we, in turn, see a similar step-up in PCF wins? Or is your $13 billion in wins already capturing what we just saw in -- data centers?

I think we have to think about them in different buckets. So PCF is really largely consumption for the hyperscalers internal use to train their AI models. There is more demand there. We see more demand. I think we will do more. But it's really not the big growth opportunity for Lumen. The big growth opportunity when you think about DC growth is it goes back to what we've been talking about for a few years now, which is data is getting further and further away from the point of consumption. And in a world of AI, that's a huge problem because as you have agents interacting with customers or running smart factories or whatever it is, those agents are looking for data in real time, trying to make decisions as fast as a human would make them. And if that data is across the country, you cannot have a data lag or any kind of interruption in getting it there. So what we see with the data center growth is really the growing demand in what a network engineer would refer to as east-west traffic. And that's the movement of data from anywhere to anywhere through a single pane of glass in a very programmable way. And that's what we're building, and that's what Alkira brings. So I think our enterprise focus, which is something our competitors have not had, it's really starting to show as an advantage as AI starts to move more into an inference phase.

Right. So PCF is good for the cash flow, but the programmable network is the future of...

Exactly. And so when you think about that, the PCF deals have largely been focused on just the training of AI, right? And so now we're moving into inference, and that's where the programmable network.

And when does inference really start to overcome training and maybe...

I think it's starting now. How fast it builds is hard to say. I mean I think the latest thing as we all learn our way through AI is, I think the phraseology is tokenomics, right? So you've got -- where does AI make sense and where does AI not make sense? And I think the companies that ultimately will win the day are the companies that rewrite business processes and use AI to fundamentally transform the way they go to market. It's not about taking a bad process and putting AI on top of it. It's about completely rewriting business rules and moving into the next phase of growth.

And while inference will be a huge opportunity, higher margin perhaps, lower CapEx since a lot of the conduits in the ground. On the flip side, a lot of inference will be done perhaps in the availability zones, where there's other fiber providers. So do you see more competition? Do you see pricing risk because you'll be going up against...

I don't. So let's break it down a little bit. Historically, legacy telco, enterprise telco has been focused only on North-South traffic. North-South is defined as premise to somewhere. It can be another premise, it could be a DC, it could be a cloud. That market in the U.S. is about a $12 billion market, and its TAM is growing at 1%, okay? The amount of data growing every year in that is much, much greater than that. So what is that? That's price compression. That is the cloud that has hung over legacy telcos head forever is you sell to the procurement department and it's a race to the bottom on who's going to have the lowest price the next time that the contract comes up for renewal. East-West traffic is what changes all of that because it's about the movement of data between clouds, between DCs. It never touches the prem. That TAM is a global TAM because you don't need to own the network. If you think about what happens with the closure of the Alkira deal, we will be the first telco on the planet that has married North-South and East-West traffic together. The point of intersection is multi-cloud gateway. It will work on-net or off-net, in a very programmable way, meaning through one pane of glass, I want to move data from here to here right now, you design your own network, you push go and it works. There is no one else who's doing that. And that is what is absolutely critical in terms of giving enterprise the ability to reduce their total cost of ownership because networking has only been thought of in a very static way. This turns it into a very dynamic asset. And that's where we see we can win.

And with Alkira, since you brought it up, it's nice to see them moving back on the acquisition path again. It's a healthy sign. And you acquired them for $475 million. Can you just provide -- you just already did a little bit for real-world examples, like the enterprises have a pane of glass and they can ramp dynamic capacity, but a lot of it could be off-net as well as on-net.

Exactly.

It depend on -- you have SLAs, so you'll depend on other carriers, they put you on broadband here.

So if you think about a global solution, think about a large multinational bank, we could license this technology to a provider in Europe, to a provider in Asia with standards. I mean part of what the value is of marrying that incredibly powerful software layer with our network is the privacy of our network. A lot of what gets moved around for enterprise today is done over the public Internet. more network hops, slower, less secure, and that privacy element is critical from a security standpoint. If we can find partners who will provide that same level of privacy on their footprint where they have it, the orchestration layer is what Alkira brings, and that's ultimately what gets monetized. So I think ultimately, you will continue to see Lumen invest in fiber where it makes economic sense to invest in fiber. But the reality is we don't have to lay one more foot of fiber to have a global footprint that allows the complete orchestration of AI networking needs.

And the deal rationale is I feel like you needed to do or build an orchestration platform anyway.

Correct.

And maybe this is going to cost you a little bit more, but you're ready day 1 at the right place...

Yes. I mean our estimates are we were going to spend in the next couple of years, $100 million to $200 million to build a lot of what Alkira brings. And it's ready effectively day 1. I mean it's -- our team would -- our technical team would say that this builds the entirety of the foundation that we need. It integrates very easily, and then we can innovate from here. So there's already a lot of innovation that has happened with things like direct cloud on-ramps, multi-cloud gateways, the plumbing that's required to move these big workloads around. The orchestration layer now makes that very programmable through one pane of glass. And look, what do I mean by programmable because this is really important. The math that matters isn't the cost of a wave. That is no longer a point of conversation. What matters is the total cost of ownership. I need to move petabytes of data from clouds into GPU clusters to be processed. The math that matters is the cost of a GPU cluster, $2,000 an hour. If I've got 10-gig connectivity and it's slow and it's clunky or I've got 400-gig connectivity, it doesn't matter what that cost of that wave is. It matters how long it's going to take me to transport that data. And so we're talking hundreds of thousands of dollars in terms of cost differential for a petabyte of data over 400 gigs versus 10 gigs.

Right. The network cost is so small compared to...

It's so small. And so when we talk about a programmable network, think about a situation where data is trying to get into a region of the country to be processed by GPU clusters and it's reached a capacity limit. And it's reached a capacity limit either because the GPUs are already running at capacity or because there's an energy cap and the local providers are putting a cap on how many of those GPUs can be running.

You route that traffic...

What does the programmable network do? It automatically reroutes it where there's capacity. You don't have to have a telco go out and dig a trench and build connectivity from A to B. You literally log into a screen and you move the workloads or the network does it for you. So that's what we're talking about. And that's the power that we unlock with Alkira and Lumen married together.

Got it. And the network topology where data goes is a great segue to my next question, which is the edge. We recently got back from Connect (X) a few weeks ago, and the edge or the topic of the edge was all the rage. It felt like the 5G edge all over again 8 years ago. Lumen stand to win big because I remember in the older days of Lumen, they had a lot of edge connections, certain amount of milliseconds to every customer. Investors were naturally skeptical, why spend the extra millions of dollars just to shave off a couple of milliseconds. Just curious to hear your thoughts. It seems like there's some validation even yesterday when I Squared bought some of these smaller edge data centers from Cogent. So just curious to hear your thoughts of where the edge is going to go and where data gravity is going to go with AI?

So edge will continue to grow because, again, think about East-West traffic, where you've got data sitting in multiple cloud environments and you need to do API development, you can pull that data down to the edge, manipulate what needs to be done and send it back wherever it needs to go, the data wherever it needs to go rather than trying to pull it all on to a campus somewhere and do it yourself. So it's -- that proximity matters. And latency really matters in a world of AI. I can't think of a better example than, say, a smart factory or how about a customer service agent where you are engaging with a voice agent or a chat agent. What happens if it takes 90 seconds for it to respond, right, versus if it's responding instantaneously as it sources data from everywhere to answer your question that matters. And so latency is a huge, huge issue for AI. It is -- it's ultimately limits AI's capacity. And so we talk about our network being within 95% of U.S. businesses within -- under 5 milliseconds. And that benchmark needs to continue to tighten.

And then I want to talk about the NaaS and digital revenue that you spoke of. Your digital revenue was, I think, $37 million, so en-route to the $500 million to $600 million target.

That just in the quarter, yes.

Yes, just in the quarter alone. And you're assuming a linear ramp in your projections, but you actually think it might be more of a J-curve. So help us envision the scenario and the rationale for this J-curve adoption. Like is it one of the situations where a couple of companies do a proof of concept and they see it and then the adoption of skyrocketing like can you help us time the J-curve, any insights...

Well, I would love to time the J curve, but it's a hard one to call. I mean, look, we're developing a new category. I think the best reference point that we have is what happened in cloud. And if you think about cloud adoption when these -- when cloud companies came into existence, what an enterprise say, well, I don't need that, right? I can -- I've got my on-prem data centers. And then what ended up happening is over time, people realize, hey, wait a minute, the variable nature of that capability, the ability to provision compute and storage on demand, all of a sudden exploded. That was the J-curve. So we think that's ultimately what happens here. Alkira could be a J-curve moment. We'll see. I can tell you this, we're doing everything we can to make it a J-curve moment. So we've got to obviously close the deal, but we're already thinking about the types of engagements that we can have with customers. We obviously can't do much until the deal closes. But at that point, we will have a target of companies in various industries that we want to go into and use as use cases that we can then share with other customers. Because the selling motion is very different, we're changing the inside of enterprise as we go and sell these solutions, right? Again, today, you sell to procurement departments and network engineers. What do procurement departments and network engineers not want to happen? They don't want you to move their cheese. They don't want you to say, you know what, we don't need you guys doing this anymore. We don't need you to figure out how to cobble together a network solution to move data everywhere. We don't need you to figure out if you can get a lower cost on a wave. It's a very different selling process...

Evangelize it.

Well, and it's C-suite driven. When we talk to CTOs about the capabilities that we're bringing, they're like, wait, what, you can do this when. This changes everything. And so they will rewire the inside of their organizations to adapt to the capabilities that we can bring. But the way the network works today, the network architecture never envisioned AI. And the way companies access cloud data today through neutral third-party facilities with cross-connect fees and slower speeds, that doesn't serve AI, and it's expensive. So it's slower, it's expensive, it's less secure. Our belief is we come with a solution that addresses all of that, and they start to rewrite the way they buy and manage network services inside of their company. So it's very much going to be a C-suite sale.

Right. And you mentioned cross connects are expensive on the on-ramps and the meeting rooms. So speaking of expensive or not expensive, how are you envisioning pricing the NaaS services then?

So I think it's going to look very much like the way cloud prices, right? You will -- depending on what your commit level is, you'll get varying degrees of price discount, right? So...

Based on data like...

Based on volume. Exactly. And by the way, that's not just connectivity because there's other services that we'll continue to bring to the market that can be delivered digitally through this motion. So that will be a primary component. But again, one of the beauties of what we have available is the ability to turn up and turn down capacity on demand as you need it. So there'll be kind of base levels of volume that every enterprise needs, but there's also going to be those peak moments. You don't have to buy to peak anymore, right? You can scale up, you can scale down. And so our belief is because of that flexibility of that as we do more East-West volume with corporations and fix their networking problems, we actually start to take share in North-South.

On Analyst Day, you talked about the P times Q scenarios. I think your balanced view was like $1,400 per port. Is that where we are today? I'm just trying to understand.

We haven't disclosed that. What we've said is, look, if you really think about the full use case of NaaS, right, in the combination of North-South and East-West traffic, we were talking about as much as $5,000 a port. We're obviously at the very early stages. But again, Alkira accelerates a lot of that. So the answer is we don't know. I wish we did know. I wish I could give you better modeling information today. But what I can share is that as we talk to more enterprises about the capabilities we will have once the deal closes in the second half of the year, they're very excited about the kind of problem solving we can bring to their enterprise.

And what's the margin profile versus like traditional telco margins on these services?

It's much higher. If you think about traditional telco, one of the reasons why there's really been a lack of innovation in the space is everything is infrastructure-based. So every service required its own infrastructure layer, its own set of ports, people driving around in trucks, plugging in the next thing. And because that was so capital-intensive and expensive, it really thwarted innovation. What we're bringing to market is software-based. So as Alkira is a software layer that will be able to talk to multiple ports. As we install more NaaS ports for connectivity into Lumen's network, those ports can receive more services. Those services get delivered digitally. No more trucks, no more infrastructure layers every time you come out with a new service. So it's a very scalable model. So what you will see over time, and we committed to this in Investor Day, I think Alkira only furthers that is significant margin growth in the coming years as well as much lower capital intensity.

And do you see competition flowing to the NaaS market? Who is your competition? I feel like it could make a Megaport, but they only have the network or the telcos, but they're doing fiber-to-home and wireless.

It's a good question, and it's one that is always difficult for us to answer because we don't mean this in any kind of an arrogant way. There isn't a pure-play competitor. There is no other competitor that owns the network and has the software connectivity. Megaport is a good example of a provider that can bring a software layer to a company, but it's limited in a couple of ways, their ability to scale is driven by how much network they have to buy to support the customer, right, because they don't own the network. And the second piece of it is that they can't deliver services deep in the network. You can only deliver it at the highest levels of the network. So think about all the layers of the network, our ability to digitally bring services over time to every layer of that network architecture is meaningfully different. So it's really the marriage of the 2 that makes this incredibly powerful.

Got it. I want to switch gears and talk about some finance stuff, the CFO of the company. The cost savings plan for one. You went from $400 million of savings at year-end. I think the goal is $700 million by the end of this year and $1 billion by year-end 2027. Can you elaborate on the efforts, what you've done so far? I think you've flattened the network and the systems. And then what's in store for the next...

It's really more of the same. So again, like a lot of others that have been in the historical space, the way these companies grew is through acquisition. So there's an enormous amount of tech debt. There's things that were never addressed. The efficiency of any kind of M&A activity historically has really been limited to spans and layers, right? It's not about fixing the underlying problem. And we're dealing with that. So we've just completed our ERP. We went from over 30 GLs to 1. We went from multiple procurement systems to one. There's efficiency that comes with that. We still have work to do on things like CRM as an example, that will allow us to unify order entry systems. So that work will continue. But ultimately, what we're doing is all of the digital products are being built in go-forward systems, one system. The old stuff, we're not going to try to lift and shift. It's just too much work for too little benefit. We're going to slowly retire those products. And as those products retire and convert to a digital product, the IT systems retire with it. So a lot of the work is really around that. And then again, just driving more efficiency. The remaining copper assets in consumer are being managed by the same operations team that manages enterprise. There's a lot of efficiency there. There's copper mining. There's a number of activities. So it's really across the board.

But copper decommissioning that's not part of the plan? Or is that...

It's part of the plan.

It's part of the plan. And any updates to sizing that savings in particular, even the timing of that?

No, we haven't -- it's small today. And over time, I think it will take us well into the future and therefore, isn't contemplated completely in the $1 billion because it's going to -- I mean, that decommissioning is going to take place over many, many years.

Right. Well, interestingly, would LEO threats actually accelerate copper decommissioning and then fixed wireless too?

I don't know. I mean the reality is, I think on the copper side, we're going -- barring something completely unforeseen, I think we're going to continue to see that business kind of attrit at the levels that it has been. And it's just a long, slow decline. Now one of the things that we continue to look at is are there opportunities to more aggressively create that churn by converting customers to a digital product that we can deliver in a much more efficient way because we would love nothing more than to not have those customers attrit, but to cannibalize them ourselves and have them stay in the family and be able to consume other services from us longer term. So as we go forward, we may do some things very intentionally to take some of that legacy business and convert it, but we'll obviously communicate that if and when we do that.

Got it. I want to talk about the EBITDA cadence for the year. You usually beat EBITDA in the first quarter, but you had the help of the $32 million sort of onetime PCF help. 2Q is typically flattish. But then again, in the first quarter, you had a month of fiber-to-the-home. So a lot of moving parts here...

A lot of moving parts...

Third quarter is typically higher spend, maintenance OpEx. And then the fourth quarter is that large step up. But then you have cost savings going through the whole entire balance of the year, just layer that on top of it. So just with all that said, help us with the general puts and takes as we think about the balance of the year?

So in the first quarter, it would be a mistake to model the first quarter by -- and then forward by just seasonally adjusting it because between the onetime nature of some PCF revenue as well as the consumer business, between those 2 things, you're talking between $60 million and $70 million of really kind of high-margin revenue. We also had -- and not prepared to call this as an ongoing benefit, a little better performance in legacy than we thought. And what we think is going on there is that as customers are buying more digital services, they're keeping the old for a little while to make sure everything is working the way they want it to. So I mean, again, we'll take that. That's a nice tailwind, but I don't think that's a trend line change. And we've also -- we're seeing some higher medical costs like everybody else is. So I think you need to adjust first quarter pretty heavily. Where we see it is we will inflect EBITDA this year. We said it wasn't going to be a significant inflection. Next year is where we start to see, I think, a more significant build in the EBITDA, and that's our guidance. We feel good about the guidance for the year.

Got it. And you did mention input costs are up. The cost of resin is up, optical equipment and DRAM within that is up, I think, as much as 70%, cost of fiber, we've seen reports up to 75%. So your deal with Corning is proving out excellent.

Yes.

Still, could we see higher input costs or cost pressuring the operations...

I mean, yes. But again, a lot of that we're able to pass to customers where it truly exists, and we'll do that where we can. So I'm not as concerned about input costs as we are really about making sure we drive the inflection in revenue. And that's the key focus.

And then how about labor? I mean how much of is insourced versus outsourced?

A lot of the labor is outsourced, and we've got great partners. Two of them are publicly available. They've been very public about what they do, but it's really a West, Central and East. They kind of meet each other where those boundaries touch as we're building these networks for ourselves and for others. And they've been fantastic partners. They've really helped us keep things on time and on schedule.

And then just moving to the balance sheet. You've done a tremendous job on the balance sheet. You paid down the super priorities as promised. I think pro forma leverage is now below 4x. So what's next? Are there any other balance sheet initiatives from here?

The big one is it's publicly, it's out there right now. I mean we've -- there's a few things that we needed to do on our capital structure. The first is reduce the quantum of debt that's happened. We've gone from $20 billion in debt to under $13 billion. The second is fix the maturity curve, right? 50% of our debt due in 1 year a few years ago, 2027 is what it was. And now we've got a maturity curve that looks normal and boring. And quite frankly, we really don't have anything to do for the next number of years. The last thing was to simplify the structure and not have 3 borrowing entities. So when we close Q2 with the debt exchange that we're doing right now, we will have one reported entity. So equity investors, credit investors, management, everybody is going to be looking at one balance sheet and one P&L. And beyond that, I think it's just more regular course of business. It's nice to be here. There's been a lot of hard work, but the focus is on growth.

Got it. And at the last minute we have, I just want to bring it all together when I think about Lumen a few years from now when the PCF dust settles and then your revenue does inflect in 2029. When I look at your long-term targets from Analyst Day, I'm trying to put them together and extrapolate from what I saw, it seems like you'd be like a $500 million free cash flow company. And there's a lot of numbers here, but EBITDA maybe [ $3.6 billion ] there's some $400 million in noncash EBITDA in those slides, $2 billion in CapEx, $700 million in interest. So is that directionally correct?

Yes, let me answer it this way. So first of all, we think -- what we said at Investor Day was business revenue growth in '28, total Lumen in '29. And again, we'll see how much Alkira can pull that forward. We'll get back to the market on that after we close. What the model showed us at Investor Day was that even if you strip out PCF, the investments we need to make are fully funded. And we would hit at the end of a 5-year window, a leverage target of 3.25 and we'd have extra cash flow beyond that. So we're in a very healthy spot where we are generating great free cash flow. EBITDA inflects this year, and now it's a question as to how fast we can inflect revenue, and it's happening. I mean just look at the rates of decline on revenue, obviously, just saying that out loud has a negative connotation. But look at those rates of decline vis-a-vis others in the space. And we are way, way ahead because our business mix is already over half of that in growth buckets versus legacy declining buckets. And nobody else can say that. Yes, so it's that -- we're not looking back on that.

Great. With that, we're out of time. Thanks, Chris.

Great. Thanks a lot.