Global view of what's going on.
Good.
-in the ecosystem. And so they're always curious around what other people are doing, what other operators are-
Yeah
are investing in, and how they see stuff.
Cool. I think we're gonna get started here. Good afternoon, everybody. My name is Brandon Nispel. I cover communication services for KeyBanc. This is a fireside chat. We have Ed Knapp, the Chief Technology Officer for American Tower. Ed, welcome.
Thanks, Brandon. Pleasure to be here. Nice to see everyone.
So Ed, you were here, about a year ago. Why don't you give us an update in terms of what's going on in the U.S. from a wireless carrier perspective, network investment, 5G? What's the latest?
Yeah, so 5G has had its share of, let's say, ups and downs a bit more recently, but it's been on a trajectory that's pretty typical, right? I think the way it started out was a little bit atypical, but where we are today in terms of the midpoint of the cycle for 5G is starting to take shape. So the reason why it started out a little different was, the spectrum in the U.S. wasn't really there, right? And so it took until 2021, when the spectrum was provided in mid-band. Most of the world had been looking at mid-band. It's a higher frequency than what we had typically done. And some operators had existing spectrum that they could deploy at 5G, so it created this sort of rush in the last few years.
But a lot of 5G was talked about in 2017 and 2018, but it was talked about as small cells and millimeter wave, and I think that's where the industry sort of did a head fake a bit until they got to the true spectrum that globally needed to be built out for 5G. The other part of the journey that's a little bit different is we started out, in the standards, they put a 4G brain on the 5G radio network. So this is basically what they call non-standalone. So we're using the 4G control, and the 4G core to be able to support 5G radio access services, and that creates some limitations.
And the standalone portion of 5G, the core, the true core, so the whole end-to-end system is 5G, built from 5G from scratch, the standards, that's just still starting to roll out, and there's a number of reasons for that. It has to do with devices, capability, it had to do with just some of the spectrum that was available and how you manage that. And that's led to some of the issues that people have said around how's 5G performing and where is it in its overall performance. But in the U.S., what we see as a tower company, American Tower, we see large investments that have occurred over the last few years. A lot of the operators are now, many of them are at 50% or higher of their build-out of sites, right?
Some are just shy of that, but some even further ahead. But the three main operators have all made substantial investments in 5G build-out. What we say is the first phase is coverage. You need to get the spectrum, we needed to get the right spectrum, and then we started to build that out. And we have lots of spectrum bands that we can tie together over time. That's where we are today. So then, people have built out 250-280 million POPs to population that's covered, and they've used a combination of low-frequency bands and this mid-band. Now we're at the point where we start to get to this 5G standalone core, and we start to see services demand, which is continues to grow at a CAGR of 20%.
Now, we're starting to see the devices kick in and new applications beginning to take place. We're seeing those in private 5G. You know, people are building these in factories and different types of settings where it's self-contained. But in the public wide area network, that last push is gonna be get the network to have a 5G brain on the 5G access network, make it 5G end-to-end, and start getting devices and new services and applications on it. And then we think that will drive demand. We're also seeing, you know, a little bit of price increases by operators. We've seen, obviously, interest rates come down a bit, so CapEx can start to turn around.
But we see in the second half of the year, as we've announced last week in our earnings, that there's a serially or sequentially acceleration in the application space for building out the additional coverage that's required. So that's all happening. So 5G is following the traditional model of that coverage phase, then there's this sort of pause and integration phase of new applications, and then there'll be this growth phase of capacity, and that's the part that we'll see in the second half of this decade.
Okay, so there was a lot there. Let's just start with where the carriers are in terms of the CapEx cycle. Do you think they pulled back because they're getting more efficient, they can do more with less? What do you think the main reason was for them pulling back on capital spending?
Well, I think it was, it's all relative. So when you think about it, because of that weird start, it was a little bit of a delayed start. There was this, you know, millimeter wave, small cell, and people waiting on spectrum. There was a surge in 2021-2022, which overshot, right? So we hit north of $40 billion in CapEx. We're back down around $32 billion-$34 billion in CapEx in the U.S. for the main operators, and they've really all announced their earnings in the last week. So the pullback is really more of a resettlement. It's still north of where we were in 4G-
Sure
... in the cycle, and I think it's a normalization that we're seeing rather than a massive change. Now, when they... You gotta remember, two operators were a little behind another operator, and they were playing catch-up. So they were throwing money at trying to get out there, that build-out, and their spectrum came in late 'cause satellite had the spectrum. They had to refarm that, get that in place in late, and the auction took place in 2021, late 2020 to February 2021. Then we had the FAA, if you remember that, then there was all this fear 5G was gonna cause planes to crash, so with the altimeter.
So there was a little bit of delay on the front end, but so they had a surge of amendments and changes to their antennas to get these new massive MIMO solutions out there to try to drive, you know, the wide area coverage for, for 5G. So I think it's a little bit more of the normalization rather than a, a massive falloff, and that, that will go back to a progression of continuing to build out more POP coverage and then continue to add new features and capabilities. But just the radio part, while it dominates 60%, you still have to deal with upgrading the core, software, and fiber for the infrastructure.
What used to be a backhaul of 1 Gbps, which satisfied us pretty well in early days, late stages of 4G, you need to move to 10 Gbps, you know, backhaul, in order to handle some of the additional capabilities in, in fi-
So you mentioned a bunch of things there in terms of build out a mid-band spectrum. As these carriers are building mid-band spectrum, you also mentioned massive MIMO. How prevalent has something like massive MIMO become on cell sites?
It's fundamental to that spectrum band. Now, it's different. And again, without getting too much detailed, every area of geography you build out needs a different tool in the toolkit. It's not a one-size-fits-all. And in certain areas like urban and heavier suburban, you really want what they call Full Dimension MIMO, which allows you to create beams in three dimensions and create a lot more capacity and reuse. So you end up with things like what they call 64T 64R. Large numbers of antennas, larger aperture, and a lot more performance than maybe what you can get in a rural area, where you may only have, you know, let's call it 4T and 4R, as opposed to 64T and 64R.
Now, that all scales accordingly based on the power and what you're trying to achieve, but in rural, you're really trying to get lower frequencies and longer range. So the physics doesn't allow that many ports anyway. And that's what you'll see some people use 600, some people use 700 MHz, lower frequency bands. But we're fundamental to 3.5 in mid-band, massive MIMO unlocks that spectrum. You gotta remember, long time, most of the cellular was below 2 GHz. And there was some 2.5 available for the longest time, and carriers struggled to, you know, Clearwire and a bunch of folks, they struggled to monetize that because the technology wasn't there. Once we got to the solutions where massive MIMO was more cost-effective, that opened up that spectrum, and it opened up the mid-band spectrum.
We're gonna see even higher mid-band spectrum in 6G.
Well, I'm gonna ask you about that in a second. I suppose as we're moving towards, you know, from a coverage build-out perspective, more towards densification, what does that mean for you? How do you see that sort of progressing over the next couple of years?
So we see. Well, operators will first do amendments to modify their current, what we call rad centers or the space they use on towers. They'll add the additional antennas. They may modify some to compact things that they might have to, to sort of maintain their space. But what, a lot of the densification will require more sites, right? So what you see is. Well, let's go to the basics. There's three ways to sort of add capacity. First is more spectrum. That's the easiest thing, and we see that about, in the U.S., there's about 1 GHz of spectrum allocated to the three main operators. There's another probably 200+ to other folks that have it.
90% of that roughly is energized and deployed, you know, at some, not everywhere, but roughly those bands are being used. There's a ton of spectrum, you know, in fact, there's a whole logjam right now with the government and the FCC and auction capabilities, but that's a different issue. But spectrum is basically the lifeblood of the business, right? The more... Each G, we need new spectrum, and that spectrum needs put to work, and we need technology. So the second piece is, how do you improve the capacity of the system? How do you get more spectrally efficient? How do you get more, what they call bits per hertz? That's the second tool in the toolkit.
So give me the spectrum, then give me the underlying technology to create more capacity in that part of the spectrum. When those things run out of gas, then you've got to go, and you've got to densify. So that's when you end up with smaller sites, more sites, and you end up going to even potentially higher frequency bands and then start to start to cycle again. So those are the three ways that you would do it. Now, for us, it comes back to colocation is a new... And as an operator saying, let's say I had one tenant on a tower. It's a good business, but two tenants is a great business 'cause you get operating leverage.
So with the more tenants we can get on a tower, the more people sharing that passive infrastructure, the better the overall economics are for our business. So colocation is driven by these capacity requirements, which force operators to look at more sites. Now, it's hard to build sites. In the different countries, it's very different. In the U.S., generally speaking, zoning laws and the FCC and the government try to do some regulation to help get small cells out in the early stages of 5G. But we'll see a lot more colo, so that tenant ratio should go up over time. People are obviously building more sites. They have different approaches to doing that, but that's what we would see fundamentally.
So, okay, if we're gonna go towards densification, I want to ask about that, but, how far can carriers sort of push spectral efficiency in terms of bits per hertz in the near term before they start to densify?
So if you go just some numbers, so like... I remember I was reading something about 3G because it was the same thing. People were saying, "Oh, 3G is gonna fail. These operators are not seeing the business." And it was a voice transition to data. They were going to maybe 1-2 bits per second per hertz, right? That's basically was the transition. We got to 4G, we got a lot more technology capability. We added MIMO for, not massive MIMO, but sort of basic MIMO. And we started to look at carrier aggregation and ways in which we can start to aggregate more channels together, different ways, uplinks and downlinks.
That got us to say, you know, you could say it was 3-5, you know, bits per second per hertz, and even higher in certain cases because we changed underlying, let's call it, technology. I don't want to get into the weeds on it, but there's a lot of stuff that you could change to improve how many bits you could send per unit time over the channel. Now we're looking at tens of bits per second per hertz. And particularly, if you look at fixed, like fixed wireless, you can get more capability, and but you don't always see that everywhere. It's not something you see on a regular basis. It's sort of the peak of what's achievable.
The way that's done today is very different, and why you need the 10 Gb on the backhaul is, we used to send a signal direct to one mobile, and that would consume, for that particular time, that radio or that sector. Then we would say, "Okay, we need it to then multiplex more of those folks together." Well, we could send double the capacity to a single individual, or we could start to allocate and chop up the spectrum into smaller pieces, which we did with OFDMA. So now we could send all people at the same time, their own signal with different bit rates. Now, what we can do with massive MIMO is we could say, we could send double the capacity, because we could use what's called cross-pole technology, and we could send that to an individual. That's called single user MIMO.
So we give you a higher data rate. But now, with massive MIMO, we call it multi-user MIMO. So now I can spatially talk to someone over there at the same time I'm spatially talking to someone over there, using the same resources.
Sure.
So we can pair people in the room, and we can further get to those higher. So capacity then, in that technology step, is always an engineering battle, right? People are trying to invent new things, but at the end of the day, you still have the physics of the environment. You run out of the capability to do that, or it becomes expensive, and you need to revert to just building more sites.
When you say building more sites, how do you quantify that? How do you quantify what the carriers need to do from a grid perspective, over-
So there was a debate recently, and it was in... Actually, not recently, but there's always a debate. This is an example. So one of the problems with the coverage is always in the uplink, right? The uplink is a challenge, and as you move to higher frequencies, you've got to figure out, how am I gonna deal with the fact that new services and applications want higher performing uplinks? You know, think of what people might be doing by capturing information and sharing with the network, as opposed to just consuming information down from the network. So when we look at how does the overall technology change, they're putting higher-powered uplinks, and they're putting more now even carrier aggregation, as I mentioned, putting channels together, and also MIMO on the uplink. You need the standalone 5G core to do that.
If you don't do those things, then the cell distance, what they call the inter-site distances, those have traditionally been built for 4G at 2 GHz and below. If you want to stay on that same grid, you've got to throw technology at it in order to use the higher freq, higher frequency spectrum. If you don't throw the technology at it, then you're forced to build more sites, right? No matter what you could do with all those bits per hertz that I just talked about, they're not gonna happen unless you constantly are upgrading the devices, you're upgrading the radios in the network, and you're upgrading your base stations. Those require changes and amendments and new touch points if you haven't been able to configure for that day one.
So the process is still the same, and it's been that way for every G's. We get new spectrum, we build a lot of new technology, we engineer that. Some of it's cost-effective, some of it's required, some of it's expensive. Operators will choose their CapEx plan on how to leverage their existing sites as much as possible. When that runs out of gas, and there's no new spectrum on the horizon right now, they're gonna have to build more sites and be on more sites.
I realize we went down a rabbit hole with some of that, but
I went into a little more detail, maybe, but that's-
No, thank you.
Hopefully, that's all.
Let's talk about spectrum then. You mentioned the log jam that there is right now at the FCC. They can't issue any more spectrum.
Well, they're not even authorized to auction.
They're not authorized. So what's your thoughts in terms of when the U.S. will get new spectrum? What bands are those eventually gonna become? And what's your thoughts on when carriers actually start to maybe refarm some of their 4G spectrum?
So there's a lot. So there's the National Spectrum Strategy. It was announced late last year, it was pushed out in March. Talks about 4 pillars and then a whole bunch of strategic objectives. Most of it's like global competitiveness and things like that. The government owns most of the remaining spectrum, right? So to the extent that you have to share it, operators want dedicated spectrum. They want high-powered, dedicated licensed spectrum because that's, that incentivizes capital deployment and ownership, right? If you start sharing spectrum, like in CBRS, there's been a lot of battles over that. There's 2700 MHz of additional spectrum that was allocated, that was identified. 3.1-3.4 is actually the band that's of most interest, but the government wants to control that with a spectrum access control algorithm.
They want to be the sole provider of how that gets managed because there's always some DoD or some other need that's there. My view is that that's the best spectrum because it's not just where you pick the band, it's like, does the ecosystem have semiconductor components? Does it have radios? Does it have devices? That takes years to get through, including the standards and all the interference studies and things like that. So picking anywhere in those, I think there was five areas they pointed to. Some were for drones at, like, five gig. There's another band, which is really interesting, at 7-8, which is where the world's going with 6G.
So there was always this WARC conference that happens every 4 years, and they're studying anything from say, 7, or 6.5 GHz to 8 GHz, but the U.S. has already allocated some of that to unlicensed. So now that has to be looked at again. There's a part of the spectrum in upper 4 gig. Not a lot of devices and components, but there are people asking to make that a public safety band. So the National Spectrum Strategy has a bunch of ideas, but they're all half-baked, right? And that's not gonna get into the hands of the operators anytime soon. It's gonna take to 2027, 2028 or later. Then you need to have the devices, then you need to have the components, and then that'll take time to roll out.
So that's why the first bucket of how you expand the network is gonna be stuck for a while. The second bucket is the standards process, and we're in Release 18 was standardized. The industry is working on what's called Release 19. There's a lot of tricks in there, but the operators are saying: "I don't know if I want all these bells and whistles." Like, get more realistic on where we are because the next step is 5G Advanced, then 6G. So the toolkit is getting really complicated and more sophisticated to get more bits per hertz, right? But there's not enough spectrum coming. So the spectrum policy in the U.S. has to be tied to where the world's gonna go with 6G. And it's looking like there's two buckets in the U.S. right now.
You can talk about stuff at higher frequency bands, like 28 and above. Sub-terahertz is stuff that people talk about. We can go into that at another point. But the main two bands are this, let's call it upper mid-band, they call it. Some of it's around, say, 7 GHz-8 GHz, and the other part is around, say, anywhere from 12-15, but in the U.S. it could be 13 or 14. So 7-15 is this next sweet spot. That'll drive the entire ecosystem for towers, for build-out, and anticipation of 6G. And what's gonna happen is, you talk about massive MIMO, now they talk about super massive MIMO. So instead of doing, like, 64 antennas, we're talking about 4,000 antennas, right? In an array, but the array size is comparable.
It's just that because you go to higher frequency, you need a lot more antennas to create the types of gain. Now, will that support the types of mobility services? Studies from different ecosystem, OEMs saying, "Yes, that's something we can do." Remains to be seen. A lot of work and testing will go on, but that's really where the next sweet spot will be.
So it seems like what you're saying is because there's no spectrum, and the spectrum that's coming is very high frequency, the grid needs to get built, and it needs to be much denser.
That's it. Yep, we're gonna need more density.
Okay.
Back to your other question,
Refarming.
The second part on refarming 4G, it's fundamental. When you get standalone, like, you can't do carrier agg. You can't bundle 5G channels without putting the standalone core, which limits the performance because you're using a 4G channel to set up your 5G experience. We want 5G channels to set up the 5G experience and be together. So things like 700 MHz, 850 MHz, you need to get those channels from 4G to 5G, and that's gonna have to happen. The problem is you still have 50-something% device penetration on 4G, on 5G, so you need to get that higher because you, you're gonna squeeze the 4G, let's call it, users to to less spectrum. But I think that that flip is happening.
We're starting to see, obviously, 5G devices. I'm assuming everybody in this room has a 5G device, but it's the rest of the country that you need to deal with, including all the suburban and rural communities.
Sure. We have a few minutes left. Does anybody in the audience have a question? Good. I wanted to ask you about Open RAN. You have one major customer in the U.S. that's sort of starting to pursue an Open RAN strategy with their network. Help us understand what's going on there.
So every generation of wireless has wanted to open up interfaces, right? And it goes back to 2G, where, you know, the switch and the radio wanted to be just, let's call it different vendors. Then we came up with small cells and said, "I could stick a small cell vendor in underneath another vendor," called... It was called HetNets at some point. O-RAN is just an extension of that. What we really wanted to do with that, and it started out before O-RAN, it was called xRAN, and some operators forced it. There was a thing called CPRI and eCPRI. These are the different protocols. I wanna be able to pick my own radio vendor separate from my baseband vendor. These are the different components of the radio access network.
O-RAN was a way to do that, and clearly, there's a lot of investment that's been made to make the radio access network more cloud native. So that was the transformation that opened up the platform to be more, let's call it, friendly to separating the radio vendor from the baseband vendor, from the core vendor. So now you can have three different vendors on the end-to-end. In the old days, you just had one vendor. Then you moved the core separate from the RAN, but the RAN was a closed system, and it was done so for many reasons. Most of it was performance and optimization. So now you're saying: I want to be able to get lower-cost platforms so I can save on my deployment CapEx, 'cause radios are expensive. What if I can pick and choose different vendors?
So opening that interface requires a lot of tedious specifications, which the O-RAN Alliance did, and what the deployment in the U.S. is saying is that we see a future of software-defined radio. We wanna be able to mix and match, and we wanna be on the beginning of that with one of the largest players who has the software stack today that can target the underlying, let's call it, Dell servers, Intel processors, things like that.
Got it. Do you see, I mean, since one major customer is going that route, will that sort of push everybody to go that route? Why or why not would they not choose that path?
So a lot of folks at our greenfield have chosen that path, right? Dish has chosen that path. There are other folks who have chosen that path. But when you look at brownfield scenarios, it's, it requires a commitment to, to make that transition, and I do think that the operators that do that can start out in different pockets and then go... But I think it's gonna be fundamental to setting the table in the, let's say, the next few years, converting your entire footprint on the RAN to that architecture, to set the table for not only 5G Advanced, but 6G. Because you'll now have the computing you need, and it may not be CPUs, it could be GPUs that are out there. 'Cause a lot of 6G is really trying to introduce AI into the protocol stack.
We're trying to introduce new capabilities like communications and sensing, and there's a whole bunch of other, you know, let's call it, societal benefits that 6G is trying to target that will be beneficial. And so having an open platform, having it cloud native, having it software-defined, that's the future, and I think everybody will ultimately get there. It'll be a question of how much they change out their equipment and from a, let's say, what's on the books from a depreciable life standpoint.
But from your perspective, it does seem like everybody will have to sort of-
It won't be as open-
Reconfigure and have it open.
Yeah, it won't be as open as everybody would say in every spec, but it'll be open enough to get operators to achieve their objective.
Got it. With that, I think we're just about out of time. So, Ed, thank you, as always, for sharing your knowledge with us.
Thank you.
Really appreciate it.
Thanks, Brandon.