AT&T Inc. (T) Earnings Call Transcript & Summary

Good morning, everyone. Hope you're having a good conference. It's an exciting agenda, cutting-edge topics and fantastic set of speakers. My name is Shashank. I'm the Managing Director and Partner with our telco practice -- with BCG's telco practice. With me is Rony Abovitz, a Senior Advisor at BCG, Founder of Magic Leap in a previous life. And Rony and I are joined by Igal Elbaz, Network CTO at AT&T. He needs no introduction, but I'm going to try anyways. He's sort of been at the forefront of innovation at AT&T and, frankly, the entire sector for more than a decade. We spoke to him at this conference last year where he painted a picture of incredible delivery of optimism and excitement around AT&T's 5G plans. And today, we'll hear a bit about how those plans have unfolded and of Igal's vision for the future. Igal, welcome. It's a pleasure.

Thank you, Shash, and good morning and glad to be here again with you guys always having a pleasant conversation with both of you.

Awesome. Thank you so much, Igal. So look, let's start with where I ended my little intro setup, right, which is accomplishments over the past year. There's been so much in the media just about coverage, about how quickly you all in the entire sector have been deploying. Would love to hear sort of some initial reactions from you on sort of how the last year has gone.

Sure Shashank. Great year. Just to remind everyone, our core connectivity business is based on Converged Access or converge broadband. We are building -- pretty much the only one who's still building [ 2 scale ] platform, 5G wireless network and a multi-gig fiber network. They're typically talked separately but not for us, mainly because customers want broadband, not necessarily they care if it's delivered to them over wireless or over fixed networks. So we've been investing in those 2 platforms and continue to invest over the last year. We are the largest fiber Internet provider in North America, will be covering 30 million living units by the end of '25. We're also operating the largest wireless network in the U.S. Our 5G network covers 290 million POPs . You refer to the build and the aggressiveness of the build that typically comes to fruition over the last year around mid-band, which is the spectrum that we've acquired a couple of years ago, and [indiscernible] on deploying. We already covered 190 million POPs with our mid-band and we're going to be at 200 million at the end of the year. That's really what gives us the combination of coverage, capacity and speed. We recently launched our home broadband solution over fixed wireless AT&T Internet Air, very well received in the market. We are migrating thousands of customers daily into our 5G stand-alone core. That's the cloud native version of the mobility core that brings all kind of capabilities. Just to remind everyone, all of our 5G customers are already served out of a cloud-based virtual platform over our nonstand-alone core, but we are now in the process of migrating the customer into a stand-alone. I typically like to talk about -- when I talk about 5G, I always remind everyone the importance of the edge and is an acceleration for low-latency use cases. And what I mean by edge is typically is the convergence of compute, storage and networking closer to the customers. And we have a little bit of different approach to that. And rather than bringing the cloud players into our data centers, we are taking advantage of the fact that many of the cloud players are building more and more regional sites in the U.S. So what we're doing is we're extending our network and optimizing routes into their regional sites and that's what create the -- that convergence close to customers. At large, most of customers in the U.S. have a 5G device. We also had a couple of firsts this year and typically -- these are typically around areas that we have interest in. So we've done the first 4G and then a couple of months after that 5G satellite direct to cellular cores. We've done this with AST SpaceMobile. That's an area that we are keenly interested in. And the other one was a first RedCap data call that we've done earlier this year. And again, given our large IoT business, RedCap is of high interest for us. So when you bring all of this together, other things that you mentioned, you now can imagine how we think about this. This is an extraordinary platform for innovation and mainly that can help us to unleash the next cycle of mobile economy innovation. So we're pretty excited about that.

Yes. No, that's awesome Igal. You mentioned platform, and I was going to go there actually because there's been so much conversation about use cases on the 5G network and the rate at which they will manifest themselves and a range of operators, both in the U.S. and beyond have either taken a use-case-first approach and sometimes, in your case, for example, you've taken a platform-first approach, et cetera. So just talk to us a little bit about that distinction and sort of your philosophy behind like how you're actually building this platform to scale?

Sure. So I described all of the components of it. And now -- we've done both. At the same time, there's use cases that are emerging in the ecosystem, whether it's the fixed wireless, whether it's private 5G, everyone talks about gaming and immersive experiences that we can start seeing. I personally always like to give the automotive example is a great manifestation of 5G. If you think about the vehicle of today, that's pretty much a software-defined vehicle. So we moved from connected car to electric car and over time, we're going to move to driverless cars. And inside of the car, it became more and more of a software-defined vehicle. And if you think about the different use cases that runs over connectivity inside of a vehicle, it starts with voice service. You have the all idea of entertainment inside of the vehicle hopefully in the back seat, obviously, not when you're driving. And over time, as you move into electrical, the amount of data that comes out of the vehicles, we talk to updates as well as the transitioning to driverless cars, whether it's the secondary system that can be supported out of the cloud with remote resistance into the vehicles that might be part of the regulation in some of the areas. So if you think about this, these are different use cases inside of this mobility-based vehicle and those use cases have different connectivity characteristics. And all of them can be supported with 5G. So these are the kind of things that are happening at the same time that we are building the platform. And I think you touched on the point of the rate of technology and the pace of technology that introduces it. I always like to give this GenAI example. A year ago, we were here together. I don't think we talked about OpenAI. I don't know if it was a topic in your conference last year because it was relatively new. I don't know how many of us knew about OpenAI and how many of knew what GenAI is. Now I'm sure most of the people in the audience know the differences between the number of parameters in ChatGPT 3.5 versus ChatGPT 4 or Gemini or other models. That's the rate of innovation and that's the rate of technology. And we want to make sure that when those use cases -- because there's a large amount of capital invested in those industry, whether it's driverless cars, immersive experiences, we all have seen the Vision Pro coming out of Apple. In other use cases, money is invested, big companies are invested in this. So we know that those use cases will show up. We know that those experiences will emerge, and we just want to make sure that we have the right platform and right capabilities to support that.

Yes. I mean, perfect segue into the next theme, right, which is AI, also the theme of this particular conference. I know, Rony, you probably have a couple of questions for Igal on that.

Yes. Thank you, Shashank. So, Igal, we'll call it AI, AI and more AI. So a few questions on the path. What are your general thoughts on Gen AI, but I guess the other flavors of AI that are emerging? And, I guess, from your perspective -- from the AT&T perspective, how game-changing or world changing is it and when do you think that happens?

So probably, it's -- a good place to start is to remind everyone. We have the data team or we actually called it Big Data team for over a decade. This is how it started, right? Big data, that was the term back them. And over the years, we had a very established chief data office that was able to introduce machine learning and AI across AT&T. So that is not new to us. We have use cases around customer care and chatbot and how we serve our customers. We have intelligence behind how we're dispatching vehicle and installation with our customers, how do you route that. We've applied this in Ford, things like Global Call. We apply this in the network. And lot of our network planning comes with machine learning and AI. So that domain is not new to us, and we've seen a lot of benefits across the company for years. What was unique about a year ago is because we had that team and because we had -- and we still have great relationship with Microsoft, and we have a great Chief Data Officer, his name is Andy Markus, and he immediately said, okay, like GenAI there's something there, we need to be a first mover. And what we've done is -- what -- we created what we call Ask AT&T, which is a private secured environment in Azure that allows us to securely put our data up and allow us to create our own environment for GenAI. Obviously, ChatGPT is the best model, but not only there's other models that we can bring in. 68,000 across -- people across AT&T have access to this. Clearly, the initial use cases are around code generation and around software development and testing. But it's growing across the company. In fact, I believe we already have a list of over 500 potential use cases where we think we're going to use GenAI across the company. Additional efficiencies in care, marketing, obviously, in network, people don't know this, but what you're doing in the network [ MOPs ] and things like these are all document basis, right? How much can we -- so it is transformational. I don't have to educate the audience here. Everyone understands how our transformational AI and specifically GenAI can be for different industries for sure. For us, we saw this as something that we want to tap in quickly. Again, we think that we were a first mover, not just in our industry, I think in every enterprise. And again, we were able to take advantage of investment that we've done for years and relationship that we have to actually move forward and create that environment.

Igal, that's super helpful. So you're clearly seeing -- you have seen value from AI already and you're seeing new explosion in value. What about your connectivity stack? If you think about the ecosystem of your network and the potential explosion or ongoing explosion in traffic that could be AI-driven, how are you handling that? And is that something you are seeing?

Sure. Just to give everyone perspective, on a daily basis, 650 petabyte of data crosses our network, every single day. That's a big amount of data. But again, in terms of how we -- so there's 2 levels to this, one, you need to make sure that you have a robust network that allows ourselves and our customers to move -- takes around efficiently because the traffic is growing 25%, 30% year-over-year. That tells you the reliance of customers on connectivity. And that tells you about the explosion of data that traverses the network. So you need to build a network that allows you to move those bits effectively. At the same time, for us, again, in terms of how we are managing the network or operating the network and obviously, data that we own or we have access to, then we always ask ourselves, okay, how can we add intelligence into what we're doing across the company? And again, this is part of what that Chief Data Office is -- team is doing for us.

That's great, Igal. And how do you view the risks of AI? Like what are the things you're watching for? What's AT&T watching for? There's all the positives, but how are you balancing some of the risks and risk mitigations they have to put in place?

We have strong governance that, again, that same team has built to make sure the -- first of all, the basics, can we actually create a secure and a private environment in the cloud because you do need a large amount of compute to run models, to run AI. So first of all, how do we make sure that our data that we upload is secure, say, it doesn't leak. We know about cases like this that happened in the industry or in general. So we're very cautious about this. In addition to this, obviously, there's strong governance that the team has put together to make sure that we use this wisely, that we have the right guardrails to -- there's -- and I don't want to get into this. There's all kind of views about where AI is going to go and are we going to have control of that? Are we going to lose control of that? I'm not there yet. Obviously, there's people that are smarter than me in that space that can already -- can provide a better view about where this is going, but I think we put the right guardrails and governance around the company to make sure that we use it wisely, we use it safely. There's humans that are involved to make sure that there's no bias and that the action that AI is taking on our behalf is actually in the -- is in compliance of what is it that we want to do as a company.

Igal, thank you and one more question on AI, then we'll go back to Shashank. When you think about this the next maybe 5 to 10 years. You described some of your readiness from the past, but how do you see AT&T winning in this blend of network, edge and AI? Like do you feel AT&T is poised, you're making the right investments going forward, not just what you've done to get here, but like look at the next decade at this blend of Edge AI and network?

Absolutely. First of all, and we may talk about this later on, if you look at wireless networks, then the next thing that is happening is that AI starts to get embedded in every layer of the wireless network. All the way from the air interface to how we're doing load balancing and traffic steering, all the way out to close loop and management of the network. And the more the network gets more open, there are more capabilities of AI in machine learning that you can apply. So that is happening. And again, it's a very small scale at this point in wireless networks and now we translate itself into the standards. But there's no doubt that this is where it's going. The way I characterize this is that, we're moving from a cloud-native 5G into AI-native [ Next G ]. And so -- and our ability to continue to build distributed networks and distributed edge and understanding the importance of the convergence of compute, storage and networking very close to where the data is, so we can keep the integrity of the data. So we can do what needs to be doing on the data without again traversing all over the network. It starts to be very critical and very important. So networks anyways are starting to looks like more and more like software running on a cloud infrastructure. Obviously, AI is going to be part of this, whether it's part of the stack of the network or whether it's coexisting workload on that same environment. Time will tell, but there's no doubt that edge AI networking is convergence that will happen, and we are very uniquely positioned to support that.

That's excellent. And Shashank, do you want to bring us home?

Absolutely. I mean, look, Igal, I think, answered half of what my question was going to be right around just the future of connectivity because if I go back a decade ago and you thought about, well, what would happen in the 10 years after that, I don't think anyone would have imagined sort of where we are today with the state of just the intelligence revolution, the state of the data processing capabilities in every part of the network and so on. Obviously, Igal, you mentioned AI is being embedded into every layer of the network, and that's one part of it. But what are some other big pillars that you believe the future of connectivity will rest on. What is your vision for the future, Igal?

So we recently started to be a little bit more public about our effort around network APIs. And the essence of this is -- so first of all, if you go a decade back, LTE was introduced, mobile economy was created, value was created. And let's be honest, we operate those at large, didn't participate this in a way we should have. And I think fast forward, many things have changed that allows us now to participate in the next cycle of innovation in a much different way. First of all, we feel much more comfortable in cloud environment. We understand software, we understand APIs. It's actually part of the standards. If you look at some of the 5G standards, APIs and API's functions are part of this. Networks are much more open than they used to be. So that's the -- it's an excellent ground now to say, okay, how can we put a software layer on top of the network that allows developers and people who are working on experiences to actually take advantage of the network in a way that potentially they didn't or couldn't do in the past. And we'll try to bring 3 things -- 3 ingredients into this to that next level. One is, again, converged scale network. Two is all of the things that we've done around open and disaggregation, virtualization of our network for the last decade, like, you actually need a programmable and open network to start to apply some of those intelligence on top of this. And we're pretty experienced in the API space. We have buildings of course, every month, easing provisioning, new device when there's big launches, all of the entitlement and provisions are all based on APIs. If we onboard a new MVNO, this is all based on APIs. But there were always in the area of what I call network periphery. And now we want to apply all of these into the network itself and think from an eyes of a developer. There is -- if you look at wireless network, there's some level of unpredictability in networks because they may have congestion, they may have some latency differences. If done right, then we can expose the right insights from the network in near real time. The question is, can we change the paradigm where developers are actually -- instead of building intelligence into their app to manage that unpredictability, can they develop their app to the network. And I think this is what we want to accomplish here and make sure that we remove some of the complexities that comes with networks and how you build networks and allow developers to say, okay, there's a great platform out there. Let's take advantage of this, let's build into the network. So we can focus on our experience. So we can focus on what our app is supposed to do versus taking care of other stuff that is less relevant to us. So for us, that's the -- the next thing that we've been vocal about lately, and we are investing in that area. We also think that industry specification and standards are important because in most cases, as much as we want to be unique and differentiated, developers don't want it -- in most cases, don't want to build an app specifically for one operator. So we are participating in the open -- the GSMA Open Gateway or in Linux Foundation CAMARA and TMForum. These are all standard and specification bodies that helps all of us operate to kind of having the same standard, same ways of we're exposing APIs. So pretty exciting because it goes back to the platform that Rony discussed, which is, okay, now it's not enough just to build the platform, how do you expose it in a way that you make it very easy for people out there to consume it.

Yes. I'm probably going to take a question from Rony right now and ask you Igal if the network is ready for the metaverse?

That's a great question. And I think there's -- people are talking out there about metaverse-ready networks. So first of all metaverse is many things. I think there's some good definition that came out lately in a couple of books, which I like. And it's a journey to get to metaverse right? Like what exactly -- how do you move from immersive experience to metaverse, what's the definition...

You have to mention the books to the listeners?

Metaverse, I think that's the name of the book. And if not I'll tell you the name of the author.

Matthew Ball?

I believe it's yes. And why I'm not sure that right now the bottleneck for metaverse is our networks. We have great fiber networks. We have great wireless networks. I think what we're going to have to see is a highly, highly distributed cloud or compute to make sure that this can happen where enough computation and data storage can be done very close to the customers. And I think as that evolve and as the experience on top of this evolve, I think networks at the same time will continue to evolve and we'll be able to support this. Again, even part of the standards, if you think about things like 5G Advanced, there's a theme there that's called split-rendering, which is all about XR, and now we're thinking about exploring this from a network perspective. So when metaverse will show up and be ready, I believe that networks will not be the limiting factor.

Awesome. Wise words Igal. Igal, I think we're running out of time here. So I'll just probably just close it out. Thank you again for joining us. Thank you for sharing sort of the incredible progress and your vision for the future. Very curious to see how things play out in the coming months, and we look forward to having this conversation again with you in a year. Thank you very much.

I have seen that I'm saying stuff and then the year after, I need to come to report on the progress, but that's fine. We like it. So...

We like your vision, right?

Yes. I appreciate the time and the opportunity to talk to you and happy holidays to you and everyone in the audience. Thank you.

Thank you, Igal.

Thank you Igal. Bye. Thanks, everyone.