Comcast Corporation (CMCSA) Earnings Call Transcript & Summary

We're incredibly excited to have Elad Nafshi here with us today from Comcast Cable. He's the EVP and Chief Network Officer. Elad's been at the company at Comcast for, I think, just around 17 years, and he's been involved with many important projects across broadband and video at the company. And I'm particularly excited about this fireside chat because I think, clearly, cable sentiment is so sensitive right now to perceived views around competitive intensity. And I don't know that there's a great appreciation of what's really happening with the path to 10G and DOCSIS 4.0. And so hopefully, we'll uncover some of the key debates here today. And with that, Elad, thanks for being here.

Thank you so much for having me. It's a pleasure.

Yes. Well, so let's kick it off with a high-level overview. Maybe talk about the broadband network evolution today, where you are, more importantly where you're going over the next 5 years? And what are the key milestones that you and the team are looking to hit?

Yes. Definitely. So thank you so much for coming. I really appreciate the opportunity to tell our network story. Our network is going to go through a really exciting transformation over the next 4 to 5 years. It's really around 3 main themes, if I had to bucket them. The first one is we've been working very hard over the last several years to virtualize the access network. And what that means is that we're able to take advantage of web scale and web speed from a high-speed data delivery infrastructure. We're able to take advantage of the latest advancement in digital fiber, which has benefits from a service reliability standpoint. And then because this is now a fully digital platform, it really tees up the point, which is turning a smart network into a brilliant network. A network that knows in real time, everything that goes on service-wise end-to-end and is able to pinpoint and so feel, when possible, and do that completely transparent to the customer. And all of that architecture is really in service of the first point, which is the upgradability of the network and our ability to deliver multiple gigabit symmetrical services over our coaxial network or over fiber and do that on a point-by-point location without needing to overbuild ourselves.

That's great. And I want to dive right into the key debates, and one of the big ones is, of course, DOCSIS 4.0 versus fiber and fiber-to-the-home. And so maybe you could talk a little bit about comparing the 2. Some of your -- fixed wireless is one topic of debate with investors, but looking out longer term, it's really about over the next 4 or 5 years, how does cable compete with telco fiber? Some of your cable peers have decided to overbuild their own HFC networks and GoFIBER. You and Charter have been incredibly consistent in terms of the successful path that you see ahead with DOCSIS. So maybe you could kind of talk to us about the 2 and why you're so confident that, at least for Comcast, DOCSIS 4.0 is the right way to go?

Yes. Definitely. So first of all, let's take a step back, right? The Comcast network is an HFC network. Engineering speak for hybrid fiber and coaxial network, right? It emphasis on fiber because the network has fiber leading all the way down to a few hundred feet away from the home. And it's just that last 100 feet into the home that are reliant on the coaxial cable, right? So when we talk about fiber, we're no stranger to fiber. We have hundreds of thousands of miles of fiber network across our network, okay? So now we're talking about the last 100 feet of delivery and whether to do that over fiber, to do that over coax. The reason why we don't feel that we need to overbuild ourselves with fiber is because we believe that we're able to deliver the same services, meeting multiple gigabit symmetrical services with highly reliable service, reliant on the brilliant network that I described earlier and do that everywhere through each and every one of our 60-plus million homes without needing to dig up streets and without needing to dig up your front lawn and without -- or installation with your favorite telco technicians in-house and be able to do that everywhere. And that's why we believe that the DOCSIS evolution is the right one for us.

Understood. Okay. And the next key question for me at least is network costs. And maybe you can kind of help us think about the contours of the cost profile over the next few years, especially as you ramp the DOCSIS 4.0. What are the key cost buckets that we should consider? And -- I'll try. Any chance you'd be willing to frame the expected cost profile, whether it's on a per homes pass basis or just more broadly for...?

Yes. I'd love to. So when you really take a step back and look at the network, I talked about virtualization, and that's the platform that actually fits the beats down for the house, right, and takes the Internet and sends it up from the house. And that's the piece that we're working on virtualizing, and that's well on its way to be mass deployed. The network upgrade is really around digitizing the nodes. That's the local aggregation point and point of demarcation between the fiber and the coaxial network that's hundred feet away from -- a few hundred feet away from the home and then, obviously, take advantage of the latest CP, customer premise device that we could install in. There's a lot of exciting things there as well if we get a chance to speak about this. And so from a cost per passing standpoint, to upgrade our network into what we call a mid-split, which enables us to launch multiple gig services today and then lay the foundation, all of the architecture in place in order to then upgrade the DOCSIS 4.0 FDX cost us, on a gross basis, less than $200 a homes pass, right? And that's not all incremental because, in many ways, we would need to make that investment for capacity augmentations anyway. And so that's why we believe that we have a much more cost-effective way of delivering multiple gigabit symmetrical speeds in great scale and do that faster, more reliably and without the customer disruption that a full fiber overbuild would require.

That's fantastic. I appreciate that number. I think it's going to be quite helpful for us. All right. Let's talk about consumption trends. I think Comcast leaders in the past have talked about the web of today is about consumption and the web of tomorrow about creation. And so as you build your network, can you -- what are you kind of solve for? Would it be consumer demand for bandwidth? Or what's the end state that you're trying to build today for the consumer of tomorrow?

Right. So first and foremost, the network will never [ gave ] the customer in terms of the services that we're able to deliver. And whether it's through residential customers or commercial customers, we want to make sure that network will always be there to deliver the absolute best, most reliable services across our entire network, okay? That's, by the way, what makes the Comcast network unique, is our ability to do that. From a speed standpoint, it's hard to argue that these multiple gigabit symmetrical services are required primarily when you look at customer usage. When you look at the ratio of downstream to upstream consumption today, okay, meaning how many bits are consumed downstream versus how many bits are consumed upstream, the network is still highly asymmetric. The ratio is about 15:1 downstream to upstream, okay? And even at the height of COVID, right, we were all kind of not knowing whether the zombie apocalypse is going to transpire next, that ratio is about 12.5:1, meaning there's still a lot more entertainment consumption than there is content that is being generated out of the home. But I want to take a step back for a second. Rather than play a game of gigabits. To me, what's even more important is latency, latency out of home, right? And if you buy into the vision of the metaverse, okay, to have a truly immersive metaverse experience, low latency is a lot more important than the sheer bit speed. And either way, the DOCSIS 4.0 upgrade that we're doing, we'll execute on both. We'll have multiple gigabits symmetrical services with unrivaled low latency that we're already proving out.

Yes. Important for the metaverse or if you're a gamer, of course.

I -- as I shared with you, I have 4 boys. And there's that humbling moment when your child beats you in Call of Duty. Humbling. Very humbling. And it's -- and I said that as a gamer. So yes, trust me. I feel the pain.

That's great. All right. Let's talk about multi-gig speeds. So in September, Comcast announced its nationwide rollout of multi-gig Internet speeds, including introducing download speeds of up to 2 gigs and upload speeds that are 5 to 10x faster than existing speeds in about 10% of your customer base through the end of the year. From a network perspective, what has been the initial learnings from this rollout of the newly increased speed tiers in the Northeast? And how has engagement been with -- across the customer base?

Yes. So I'm going to nerd out a little bit. So when you look at the way we're delivering the multiple gig speeds today, we're reliant on something called DOCSIS 3.1, which is the current latest data over coaxial cable. And we are able to deliver multi-gig services on that service. But to open up additional upstream speeds, yes, the network is still asymmetric, but we want to be able to deliver on greater speeds. We're implementing something called mid-split. And what mid-split is, in plain terms, increasing the amount of frequencies that we're allocating for upstream bandwidth. And that enables us to effectively triple the amount of upstream capacity that we have today and increase the upstream speeds by 6 to 7x the faster speeds that we have today. Operationalizing it is really where the network becomes brilliant. It's how do you now launch DOCSIS 3.1 services on the upstream, overcoming network disruptions that are external. For example, TV antennas and the options that come from TV antennas. LTE interferences from cell towers. What you want to do is you want to build a network that is able to self-detect that and mitigate around those interferences as they come in because most of them are transient. And that's the brilliant network that we are able to build already, that self-corrects all of that in real time without needing to have any type of technician or any type of human interaction. So that has been the greatest learnings from a network standpoint on how do you deploy that in great scale and operationalize those speed deliveries. In terms of the response, sadly, I'm not upgraded yet, my home isn't. Well, I'm this close. They're going to get to me later this year. I'm part of the 34 markets that we've announced, and I can't wait for that. I really can't. I make fun of my -- the team that runs the deployments, okay, let's go. Come on. I can't wait to have it. It's enormously well received, obviously, without going through the numbers.

Understood. Okay. Well, I'll borrow your phrase from earlier and let's continue nerding out a little bit and talk about the network upgrade path. As you begin testing with consumers next year, what are the key operational priorities around the network that you're focused on to begin deployment of DOCSIS 4.0 in the market in the back half of 2023, which is the target that not a lot of other cable companies have really shared or -- and so we're excited that you have and look forward to that?

Well, trust me, I'm excited that we have as well because I can't wait, right? But at the same time, as important as the speed is, we have to make sure that the service always works, right? Customers -- I use this example all the time. We were shielding in place, me and my wife and the 4 boys, during COVID, right? And on those rare occasions that the Internet would be down, I would have an average of 1.5 kids within 15 seconds of it going down in my home office, letting me know that the Internet is down. It's like, okay, I'm [indiscernible], I could see that, too, right? And that's just the reliability expectations that consumers have today. And so the time spent between now and that second half of next year is really around truly hardening and operationalizing the service quality and making sure that we are able to consistently deliver the multiple gigabit symmetrical speeds. The second one is operations. We need to make sure that our technicians have the meters and all the visibility that they need in order to be able to do that. And then the third, which I'm mostly excited about, is with DOCSIS 4.0 FDX, we get real-time visibility into every edge of the network like never before. And I can't wait to see what my data sciences team is able to do with that and how do we make, again, the network, even smarter in terms of real-time responsiveness to any type of reliability challenges across the footprint. So that's really what we're going to do between now and now.

That's perfect. And let's talk about full duplex a little bit more. Can you help investors better understand why Comcast network was best suited for an upgrade to full duplex DOCSIS 4.0 and what were the network realities or requirements for Comcast? And what operationally does this path provides for the company going forward as you kind of target the 50 million-plus homes and businesses pass by the end of 2025?

Yes, yes, yes. So if you take a step back and look at historically how cable upgraded its network, okay, the first cable systems were like 330, 350 and then they went to 550 and then it went to -- some of you kind of get the picture, right? And when the networks became 2-way, they allocated a small number of frequencies for upstream speeds because the fastest cable modem service at the time was 1 meg, and that was really fast, okay? And over time, as speeds increased, what happens is that the network still is a bidirectional network, meaning there's frequencies that are allocated for upstream and there's speeds that are allocated for downstream, most of them are allocated for downstream. And it's really kind of separated, and therefore, not very efficient when you look at the actual usage, okay? Because there's actually very little sustained speed at 2 or 3 or 4 gigabits, right? It just doesn't exist. And so in order to launch the -- DOCSIS 4.0 gives cable operators 2 options. They should continue to expand the frequencies all the way up to 1.8 gigahertz and -- but keep the upstream and downstream separate as they do that, okay, which means there's a lot of spectrum that's going to be sitting there idle most of the day, but it's still hard separated from a frequency standpoint. Or you don't need to do that. You could keep the network as it is today with our network virtualization and be able to share the spectrum with complete power where you're able to overlay upstream and downstream transmission right on top of each other and be able to do that more efficiently. Now the essence here is really around speed upgradeability, okay? In order to expand the network all the way up to 1.8 gigahertz, we need to go out to the network and find every splitter, every cable, every device that connects the home that is not currently capable of passing 1.8 gigahertz of signal, okay? With FDX, you don't need to do any of them. All you need is software download to a virtualized architecture, new generation of DOCSIS 4.0 FDX electronics in our nodes and amplifiers, and we're ready to go. And so the key to our strategy [Audio Gap] the reliability that comes with and the scale across our entire footprint like no other into DOCSIS 4.0, which is why we chose the DOCSIS 4.0 FDX. Either ways it would work. We believe that this is the right way for Comcast.

That's great. And let's talk about FDX just a little bit more and kind of the successful test that you've had over the last year or so. You announced the successful test with the newly designed FDX amplifiers. Can you talk a little bit more on the amplifier side? And when you think about the innovation with the amplifiers, what does that ultimately enable for the network and for consumers?

Yes. So let me just demystify what an amplifier is, okay? So on those few hundred feet of coaxial cable, you need to make sure that the signal gets all the way to the homes or the end of the other coaxial line, okay? And to do that, you need to amplify the signal, right, just make it stronger. And so amplifiers are installed in the network in order to do exactly that. They basically take the signal, amplify and continue to send down the line, okay? And so when you talk about the upgradeability, being able to overlay those upstream and downstream frequencies for DOCSIS 4.0 FDX, you need a new generation of really smart amplifiers that are able to do that and do it seamlessly to the customers. And what we're able to show in Philadelphia just over a month ago, and I'm so happy that you were there for that, is a new generation of smart amplifiers that not only are able to do that, but are also enabling us to have real-time visibility and auto correction of any type of RF interference on the network. We don't need to send -- once we have those in place, we won't need to send technicians to deal with RF impairments. They don't need to go out there. There's something called pads in there, they're like modules that you need to have to swap out. That amplifier is basically a computer that self-corrects all of it and works in tandem with the rest of the AI and the virtualized architecture and auto corrects all of it. And so when I talk about a brilliant network, delivering multiple gigabit symmetrical speeds, that's what's coming. That's why the amplifiers are so important. And that was really the missing piece to be able to deliver the multiple gigabits symmetrical services throughout the entire network without needing to do any type of special construction to do that.

That's great. That's great. And let's talk a little bit more about the network virtualization because I know that's been a big focus as well. And maybe you could help us better understand what the operational benefits are of that processes?

Yes. Totally. So if you look at the kind of traditionally how cable modem services were delivered, they were delivered on something called appliances, right? Those are dedicated hardware that were custom made to be -- something called the CMTS, cable modem termination system. Engineers have acronyms, right? And what it is, it's basically a device that just aggregates traffic from the home into the Internet and from the Internet down to the home, okay? And that's great. And we've used those for many, many years. But the problem is around the speed of innovation, right, because they're custom-made devices, they're custom-made chipsets that feed those devices. And so the speed of innovation is in the cycle of the custom-made [indiscernible], right, versus if we were able to rely on the same concept that the web scalers, the hyperscalers do, which is the latest and greatest Intel-based hardware, and we abstract the hardware from software, hence virtualizing it, right, and now we could innovate at the speed of Moore's Law, meaning a server that I buy today will be twice as capable at the same cost 2 years from now. And that's not just saying it, that's the essence of Moore's Law, right? Andy, Moore was one of the founders of Intel, and he came up with that law almost 40 years ago now, and that has held true throughout those years. And so, as an example, as we speak today, we're transitioning to a third generation of this Intel-based hardware. And we're literally doubling our capacity by virtue of doing that, with that virtualized architecture. Now the really exciting piece is by separating the hardware from the software, we're now able to innovate at the speed of software. We don't need to wait for this advancement to that advancement. I have teams of developers, who are building and innovating and rolling this out with a highly automated, highly reliable deployment of apps that we are deploying across [Audio Gap] Virtualization is really the essence and what enables us to deploy multiple gigabit symmetrical services over coaxial cable or over fiber. Again, we have many homes connected with fiber and be able to do that completely seamlessly without needing to have a new platform, without needing to have a new architecture, without needing to have new operations, and all of that with the real-time visibility that comes with the virtualized architecture.

All right. I want to go back to something you mentioned earlier. We were all obsessed as consumers at least on speed, how fast was download, was upload. We talk about the importance of latency. Talk to us about importance of reliability or what other factors are you and the network team are focused on to deliver the brilliant network?

Customers pay us for speed with the expectation of reliability. And we take that very, very seriously. The advantages of the virtualized architecture is that real-time visibility, and what the real-time visibility gives us is the ability to act on impairments as the customers don't want to know that we knew what happened to them an hour ago, right? They want to know that, hey, my service is impaired right now. What are you doing about it? And where we can do something about this small platform standpoint, we want machines, artificial intelligence and the brilliant network to take care of that. And where we can't, we want to make sure that we're able to dispatch the technicians to the pinpoint location of where the issue is, so we could quickly resolve it. And that's the exciting piece of that network transformation that we're after. Yes, speed is important. Yes, we'll be able to deliver multiple gigabits symmetrical with the lowest latency, but it has to must always work. And that's what's built into everything that we do across our network.

Great. And I know there's a lot of details over here, but I think it's important for investors to better appreciate these. So maybe we could talk a little bit about the XMF platform. And you recently introduced the platform. It monitors hundreds of thousands of local broadband optical links every minute across the network. What does this mean for you, the engineers, and more broadly with the network?

It's a great story of Comcast innovation. So the challenge with a fiber network, right, is when it works, that's great. But when it doesn't work, why doesn't it work? And how do you detect it? It's just light on a piece of glass [Audio Gap] where the impairment is. And this is a technical challenge that the cross-country fiber folks have solved because they have to send signal 2,000 miles away, they need to make sure that they know if there's any type of impairments, okay? The problem with that because of the distance, the tolerance of the chipset that actually makes that mechanism work needs to be really tight, right? So the yield on these chips is very low. They throw -- the people, who make the devices have to throw entire cost because it's just not good enough. The quality is not fair enough, okay? Now our access fiber is no more than 80 kilometers away from our locations, right? That's the benefit of a cable network. And so I can effectively take the rejects of all those chips that would otherwise be thrown away to the garbage, they're good enough of our businesses, why don't we build a device that is able to now connect to the network upgrade and the network virtualization that we're executing and is able to constantly monitor each and every one of the hundreds of thousands of access fibers that we have out there and be able to do all of that in real time, and comes with built-in capabilities that when there is an impairment, it auto detects it, and it is able to pinpoint on a map exactly where it is. And that's what we do. And so we're better engineered [Audio Gap] [indiscernible] meter for fiber, okay? And that's something that is being installed as we speak nationally. And we're able within less than 2 minutes to pinpoint exactly where fiber impairment occurs. And now this AI engine is built into our customer-facing tools, we're able to actually auto dispatch a fiber restoration crew exactly to that location to repair it, which reduces the mean time to repair by hours because otherwise, you're going to have to do all of that manually. And now we're actually able to message to the customers in real time that says, hey, there's a fiber disruption in your area. We're aware of it, our technicians are on their way. And the current ETA for restoration is 4 hours, and we'll update you as the repair of the network progresses. And that is a brilliant network, and that's where a brilliant network comes together.

That's great. And where it is -- when you talk about the customer experience, where does Xfinity Assistant come in?

That's a great question. So the Xfinity Assistant is that customer side of the AI or the real-time visibility that is able to message the customers exactly what is going on with their services and be able to do that in real time, right? And so rather than say, hey, reboot your modem, right, if there's a fiber cut, what are you rebooting? There's a fiber cut, right? And so how do we take that information, that real-time visibility and turn it relevant, not just to our platform, not just to our technicians, but to our customers and do all of that in real time and do all of that great scale? And that's really the customer-facing side of the reliability of the brilliant network to be able to message at that level to our customers when we're getting [ waver ] from a feedback standpoint, but even more importantly, from a Net Promoter score, that's how we measure the satisfaction of the customers, and customers that we're able to interact with at that level.

Great. And I think, unfortunately, we only have time for one more question, and I want to go back to what you said, which I think will be probably the headline of today when you gave out that $200 homes pass number. Can you just unpack that a little bit more in terms of anything more you could share on what's in that number? How you expect that to evolve? Can you just talk about how it's not all incremental? How do we think about what parts of the $200 are incremental? And what's maybe an opportunity for rationalization and efficiencies elsewhere?

Yes. So I can build your models for you, man. That would be fair.

That'd be great.

I would say this, the number that we share today is an all-in number, right, and I kind of walked you through the molecules on what makes that up. I would say that from a capacity standpoint, take a step back and think about this, right, what we're able to get with the upgradeability, whether it's with the mid-split or whether it's with DOCSIS 4.0 FDX is tremendously more capacity that is readable, right? We augment capacity in the network all the time. When they say that it is not all incremental, it's because we're able to learn [Audio Gap] in order to augment our capacity where we need to, right? I think that it will stay at the less than $200 for now. And I'm super excited, I really am, to be able to have that opportunity to bring this brilliant network, to bring these multiple gigabit symmetrical services, to bring a network that is truly like no other to so many homes across the U.S., and we're just getting started.

That's great. Well, Elad, thanks so much for delightful conversation.

Thank you so much for having me.

I appreciate it. Thank you.