Listen: Next in Tech | Episode 64: Digital infrastructure – towers and datacenters unite! Sort of...

View Full Transcript

Transcript provided by Kensho.

Eric Hanselman

Welcome to Next in Tech, an S&P Global Market Intelligence podcast where the world of emerging tech lives. I'm your host, Eric Hanselman, principal research analyst for the 451 Research arm of S&P Global Market Intelligence. And today, we're going to be discussing digital infrastructure with Lynnette Luna, senior analyst; and Dan Thompson, principal research analyst. Lynnette, welcome to the podcast. And Dan, welcome back.

Lynnette Luna

Thank you.

Dan Thompson

Yes. Thanks, happy to be here.

Eric Hanselman

So I want to extend this conversation we've had around digital infrastructure fairly broadly. I mean, if we think -- we've talked a lot about the digitization of organizations, better use of data. Last couple of episodes have been talking about data, data handling, data application, data security. But we haven't really talked about the infrastructure that's necessary to support it. And there's a lot of activity that's going on in both of your spheres.

I mean, Lynnette, you cover a lot of cell tower infrastructure. Dan, you're talking about data centers. There's a lot that comes into play trying to bring both these together. And if we look at organizations, they're getting to this journey towards greater digitization, they're thinking more about integration of computing and connectivity, how you bring all this together. How do each of you see this playing out? Lynnette, why don't I toss the question to you to start?

Lynnette Luna

I think with the advent of 5G, it's very interesting, I actually had a good discussion with some -- a carrier and some other people involved in the enterprise market, really talking about the type of 5G and how it was supposed to transform everything immediately. And I think what we're seeing instead is more of a transformation of existing processes, so overlaying new experiences on top of those.

So I think that's going to kind of create a whole transformation for data centers and for operators to kind of start to get involved in the edge. Every operator has like an edge computing plan to get there. A lot of them are partnering with hyperscalers or trying to do it on their own. So it would be interesting to see how this plays out in the tower industry.

Eric Hanselman

Well, wait a minute, you're saying 5G hasn't already transformed everything? I say that tongue in cheek because to your point, there's a ton of hype around this. But you've touched on what is one of those really important aspects of this, which is the concept of the edge and the idea about being able to bring together the connectivity and compute pieces. And Dan, you're on the compute supporting infrastructure side.

Dan Thompson

Yes, sure. I mean, I would echo a lot of what Lynnette said. I mean, in the data center world, 5G is obviously a hot topic as data center providers are working with the network providers and trying to decide what does this look like kind of longer term. I think nearer term for us, this tight integration between compute and networking and specifically the data center has kind of played out in a number of ways.

So for one data center providers have really doubled down on their interconnection play, trying to make this as automated as possible. So that enterprises, you don't have to pick up a phone and call somebody and then wait 30 days for something to get implemented. They've all -- or at least the major ones anyway, have all rolled out an online portal that allows you to provision networking right there just on your screen and then have that roll out.

In some cases, real time, and in some cases, there's still kind of a man behind the curtain or one behind the curtain, moving cables and things like this. But the goal is for that to happen kind of real time and for it to be software-defined in the future. We kind of coined a term a couple of years ago, software-defined interconnection. And that's kind of the idea there.

The other way that this has played out is connectivity to the public clouds. As companies have been leveraging the public cloud more and more, that connectivity into those public clouds for a hybridized infrastructure now becomes that much more critical. And companies are looking for faster connectivity into the public cloud but also more redundant connectivity into the public cloud.

So that, as again, that connection between the data center and the public cloud is more critical, I've got to have more of those options available to me in case one happens to go down. And so that's really kind of what we've been seeing. But you just can't ignore what Lynnette was saying, which is this whole edge concept as we look to the future.

Eric Hanselman

Well, what applications are we looking at, where distributed and dispersed computing could be put to work? You talked about the need to be able to interconnect data centers and public cloud. But we've got to get those end users. And presumably, that's where the tower piece comes in, right? So what are the applications where tower connectivity and data center really come together?

Lynnette Luna

So I don't see that today. But I think it's more planning for tomorrow. Video has turned into this huge data traffic. The majority of traffic on mobile networks is data today. And as 5G gets rolled out more and more to the enterprise, that's just going to move it faster. And we're talking about more like HD type of video. And then we start getting to VR and AR type of applications. So it's worth to start network planning.

And I think we're starting to see some tower companies looking at that, especially American Tower, which bought CoreSite last year. I believe that was in November. And they really see it as companies push more data into the cloud, they see these more as a cloud on-ramps that will need to become much more diverse and spread out toward the edge up to and including cell towers.

So I think that they're thinking about with the evolution to mobile edge, that network deployment will likely require some tower and tower space because carriers will join those facilities to reduce the latency of the traffic, especially when you have heavy applications like AR and VR.

So they want those cloud on-ramps and compute power close to where their base radios are, which are at the tower sites today. Other tower companies are just kind of dipping their toes in the water, maybe more willing to invest in edge companies or make strategic investments. And that's what we're seeing today.

Eric Hanselman

That makes sense.

Dan Thompson

Yes. I think from our perspective, what we're seeing -- I can't say that it's specific to cell towers. But if I can zoom out just a little bit and kind of expand the discussion just to edge deployments, I mean, we have seen play out in our Voice of the Enterprise research, some use cases in the manufacturing world, where they're maybe doing analytics on consolidated data from a number of different manufacturing equipment, and they're trying to decide, do we need to run maintenance or things like this?

We've also seen some use cases from smart city top applications and some more advanced 911-type services. So there has been some demand for workloads to be kind of closer in to end users or in devices. More often, I believe, in devices necessarily than end users, although Lynnette kind of pointed out some interesting use cases around video. I would also throw like video gaming as we have some of these more massively interactive-type games that are kind of coming out that require compute to be near a big grouping of users.

So that has kind of been one interesting use of this. So far though, the edge from our perspective though has been maybe downtown areas, which we wouldn't have -- when we kind of started this edge discussion, no one would have thought downtown L.A. was an edge. But for some of these applications, it is the edge. Some of it is Tier 2 cities. Some of it then, as Lynnette was talking about when we get down to the cell tower usage, can be a specific -- a very specific spot on a map.

So far though, like the use cases that I've seen, like that hasn't come up, right? There's been lots of discussions around autonomous vehicles and these airplane engines that generate tons and tons of data. But so far, like neither of those applications has demanded data center space at all the cell towers all around the country. What has happened in both of those cases actually is compute has been put in the airplane or compute has been put in the smart car.

And that's how those industries are kind of dealing with it. Future-looking though, perhaps there will be some other application from those, where it's more less real-time analytics that needs to happen, but somewhat near-term analytics. All of those could play out in the future. But again, it's a lot of guessing at this point.

Eric Hanselman

Well, I guess, we're in a situation now where we've got enough compute capacity close enough to handle the use cases that we're dealing with today. And there are pieces that can potentially -- I mean, Lynnette, you talked about video. And one of the big uses of -- I'll use the "edge" definition is actually video ingest processing, so to reduce the total amount of bandwidth that's loading up the upstream version of video. Downstream already has a whole set of controls and things that are already there.

But -- and Dan, you also mentioned gaming. I mean, if we look at things like the Niantic Harry Potter games coming out, that it takes that sort of Pokémon Go approach of individual users seeing that sort of augmented reality view to now actually having players interact with each other, which means that you have to have some amount of compute close enough to the players that are actually doing the interaction to coordinate the interaction.

So maybe that drives it a little bit. But I guess, a lot of this really starts to get back to the question of some of those market considerations about what actually drives this. And it seems like that's something where we've got to get to consumption models where the end users can actually consume the union of the connectivity and the compute in ways that make sense.

Lynnette Luna

Yes, I agree with that. Things like AR/VR are kind of pie in the sky right now. But when you think about like Meta, formerly Facebook, and others really looking at that as like the next big computing platform. I think that will provide some interesting need for the edge. And I was talking with some of these executives about how COVID has changed business cases. And it's really accelerated a lot of things in 5G that probably would have taken a lot longer. Health care is one. You have remote or telemedicine that probably wouldn't have taken off as fast as it has with COVID.

And you have things like sharing high-quality images. I was even talking to somebody about the whole rideshare industry and how maybe you can overlay AR on top of that, so for greater security to make sure you're getting in the right car with the right driver or you're the right person they're supposed to be picking up, or in my case, this weekend, when I had to talk to the driver and explain what landmarks were around me because he couldn't find where I was. How cool would that be if I could just like show him...

Eric Hanselman

Point the phone and it's like, "Here's where I am." Yes. Well, and the same thing, you could see, "Hey, here is your car arriving," and identifying that this is the right car.

Lynnette Luna

Right. So I think we'll see a lot of things that will have high computing power in the enterprise market start trickling down into the consumer market, so where it kind of gets spread out and where you start needing more edge locations. When that happens, I can't tell you. It could be 5 years from now.

Eric Hanselman

Well, I guess, we see some of it today. I mean, you've got startups, the likes of MobiledgeX, which is a Deutsche Telekom spinoff that has edge-centric orchestration. So you can say, "Hey, put my workload this close to this particular place, thing, activity." So you've got pieces of that.

I guess, that's maybe what I was getting towards was this thinking about we've got to have consumption models that understand how to actually let people consume edge capacity in ways in which they don't have to figure out where the user is or where they need to put the workload or bringing all these things actually to get...

Dan Thompson

Yes. I mean, I think that's one thing that has kind of struck me is as we've gotten into this, the very first question that everybody always asks is, "What's the enterprise use case here?" And really, the companies that we've seen come to the forefront first to the edge are actually service providers, who are laying their services on top of edge, and then they are the ones actually reaching out to the enterprise.

And it's just exactly what you just said, Eric, it takes the complexity out of, "Okay, great, there's a box at a cell tower down the street. What am I supposed to do with that?" A service provider then came in, put compute in it, put a hypervisor in it or whatever and then now makes that available so that I can reach out to it from my cloud infrastructure or my virtual infrastructure or whatever so that I'm not having to manage all the hardware.

It's a classic scenario of the cloud providers took the headache of dealing with data centers and dealing with the whole physical hardware stack out of the way, right, and they just handed you the compute. It's the same thing that these service providers are doing. They're taking the headache of dealing with the box and the cell tower and the connectivity, also the physical infrastructure stack of it and then just handed you connectivity. So I mean, it seems like it was a very logical evolution of this model.

And then I think the next step will be enterprises start kind of putting all this together and thinking, "Oh, hey, actually, I could put my own VMware stack in there, and it might save me a little bit of money or perhaps I could have a better performance if I did it all on my own or I could do some more custom things," all the classic reasons for why you would now use colocation as opposed to using cloud or whatever. So lots of options, but I think the evolution of it happened in just the same way we saw it in cloud, as I said a minute ago, rather than the enterprise jumping in feet first.

Eric Hanselman

Well, this is the point at which the audience is going to heave a collective grown, and so to say, "Oh, now Eric is going to say abstraction again." But again, this is -- the great thing about the power of tech is that what you just described, Dan, is the ability to have this level of abstracted service capability and something where it helps drive consumption. If you can go build the platform, now all of a sudden, you make it easy to consume and it drives use, simple and straightforward.

Dan Thompson

Absolutely. I mean, if you're an application developer, like you don't want to have to deal with the box at the bottom of a cell tower, right? You just want to be able to push your -- whatever part of the app needs to be close to the user or close to the device, you want to be able to push that out to wherever you need it and not have to worry about all the pieces and parts. So it's a question used quite case of the cloud and kind of what makes it so nice is it removes all that headache and just gives me the ability to do work.

Eric Hanselman

And gets to that point of driving that -- pointing that connectivity towards towers and the tower infrastructure and helping to blend those two together.

Lynnette Luna

Right. Yes. I think tower companies want to increase their value to carriers. So any added thing that they can add to towers, and I think a lot of them are playing in fiber as well, increases their value.

Eric Hanselman

Well, that's a really important point, it's -- towers by themselves are good, but you've actually got to have something that connects everything together. And that's the fiber connectivity piece.

Dan Thompson

Yes. And actually, I want to -- it's actually a great kind of jumping-off point to something that Lynnette brought up earlier. So she was -- she mentioned this American Tower deal, American Tower acquired CoreSite. It's been some time ago now at this point. But to me, if we can circle in on that deal for just a second. What was interesting about that is there are a couple of companies out there who have been pitching this whole notion of containerized data centers.

So -- and when I say containerized, I mean literally like a shipping container top thing. And the idea was placing this at the base of cell towers. And then you have a number of data center providers who have data centers kind of spread all over the place, right? But to me, the missing piece has always been as a company, as a service provider, I need potentially both of those, right? And I don't really want to have to go source at all.

And so to me, in the back of my mind, when I'm hearing these sales pitches, the missing piece of the data center provider is, "Okay, that's great, but you don't have like a truly edge solution." And when I'm listening to the edge, people I'm like, "Okay, great, but you don't have the actual core solution that someone would need." And what was really fascinating to me about the American Tower deal is it put those two together now.

So now it's -- you can go to a one-stop shop. I can go to an American Tower and say, "Hey, I'm using CoreSite already in downtown L.A. I need something over in El Segundo. I see that you have a tower that's like 2 blocks away from a manufacturing facility. That would be really, really helpful if I could leverage that." And then now suddenly, "Oh, okay, well, now we can string all this together," American Tower can put the fiber between One Wilshire and that box down in El Segundo or whatever, whatever the scenario is. And now I can actually have a solution rather than having to piece it all together.

Again, I mean, this comes back to your abstraction, Eric, not quite totally abstracted yet, but it's removing barriers to entry for companies that, at the end of the day, they just have an application that they need to make work, and they really don't want to get into the hardware business if they can help it, this American Tower deal. And I'm guessing that we'll see more like this, kind of helps kind of solve for dealing with multiple vendors and all that kind of thing.

Lynnette Luna

That's a good point, Dan.

Eric Hanselman

Well, I guess, the question in all this then is so what should enterprises be planning for? I mean, with greater integration coming into play, and how should tower and data center owners address this? I mean, is this -- does everybody have to consolidate? Is this happy partnerships? Where do you see this going? And I guess, that core piece is what should enterprises be thinking? Should they be expecting abstracted services? And is it all going to be abstracted ponies and rainbows someday soon?

Dan Thompson

Well, I mean, interestingly, look, if we can go back to the American Tower deal, so American Tower had actually struck a partnership with Flexential before they bought CoreSite. I don't mean to suppose that there was anything wrong with that partnership. But clearly, American Tower wanted something more. And they went and bought it. So I mean, to me, I was excited about the Flexential deal actually for the exact reason that I just described a minute ago with the CoreSite deal.

But again, clearly, CoreSite felt like they needed to own the assets rather than partnering. Could the partnership model work? I think so. I mean, if you could integrate this tight enough, I mean, I've seen other data center providers work out partnerships to where it was seamless from a billing perspective to their customers. Basically, myself, the Flexential customer, just goes to them and says, "Hey, I need tower space, I hear you have a thing," and they can sort it all out and I just get one bill. As long as it looks like that to the customer, to me, I think it's a winner.

To take this one step further though, a company like a CoreSite or a Flexential, both of those companies offer private cloud solutions, why not go ahead and deploy that private cloud in the box for the customer? And then now I've got a one-stop shop and I'm just deploying applications now rather than I still have to kind of think about hardware. So to me, that's the consolidation or where this consolidation can shine. And from an enterprise's perspective, I think we're still -- and I don't know. And I would love to hear Lynnette's thought on this as well. Like I kind of still feel like from an enterprise perspective, they're still trying to figure out what to do with this or how to leverage this.

To me, I'm thinking, "How do I get this piece of the application," or even before we get there, "What applications might need to be closer to users or devices," and then having my app developers take a look at, "Okay, yes, we can split these pieces out, these pieces really need to be at the core. These could be at the edge. And what does that do for me from a latency perspective, from an end user experience perspective," things like this. But I don't know, Lynnette, do you see things differently? Do you see this from the networking side of the world? Is there already kind of thoughts there?

Lynnette Luna

I agree with you because I think the same thing is true for the mobile operator. They know the edge is important to the enterprise and to their networks. But they're still trying to figure out how to deliver it. So I think that's why we're seeing a lot of hyperscaler partnerships. You have Verizon, has partnerships with Amazon Web Services and Google Cloud Platform and Microsoft.

They all have some sort of partnership there just to get to market. So I think that will be a big part of this early on is just a lot of partnerships, just to kind of figure out where the heck does the edge go. And I think you're right, I mean, everybody realizes that they need this, but where does it go? And what's the most cost-effective way to do this?

Eric Hanselman

So maybe a few more toes dipped in the water here and there, but we'll see what shakes out.

Dan Thompson

Yes. I mean, you made a great point there. I mean, Amazon and Microsoft have both made partnerships with network providers for their edge-type services. Everybody is thinking in this direction. I think it's really -- what's happening in the background is companies are looking for the "killer app" of this whole scenario.

And once we find that, I think you'll really see this thing lurch forward. I think right now, we're very much in the exploration phase. I mean, I know there are apps that are kind of moving and shaking right now. But we've not seen it just explode yet. And I think that's coming.

Lynnette Luna

No. And I'm not sure if we're going to have some sort of killer app. But I think we're going to have -- I think what we're seeing today is 5G overlaying existing processes and making them better. So you keep enhancing that. And that's when you start needing things to move faster, right?

And that's when you start needing the edge. You think of like employee safety or fraud layering and things like that in the enterprise. And then I mentioned video earlier. Just maybe even just being able to overlay video over maybe a lot of customer service use cases and things like that.

Eric Hanselman

it sounds like some of those early-stage AR uses and some of those integrations. So to your point, Lynnette, maybe some incremental steps rather than the big bank.

Lynnette Luna

And like we've already seen that with LTE, right? I mean, this was the same thing. LTE came out about 2007, deployed -- I think it was Verizon who's first deployed it. It's supposed to totally transform the market. And of course, it took like 6 to 10 years for it to mature.

But then we saw brand-new businesses coming out of it. These rideshare apps would never exist without LTE. So I think we'll see the same thing. It will just take some time. But everybody likes to write about how the latest technology is always overhyped.

Eric Hanselman

Are we hyping or overhyping?

Lynnette Luna

Maybe.

Eric Hanselman

Well, we'll see whether or not we get wonderful transformation from 5G or whether or not it really is a repeat of history. "Ontogeny recapitulating phylogeny," my geologist mom would always say.

Lynnette Luna

There's also hardware considerations, too. AR glasses and different computing capabilities that need to happen to be able to handle these type of applications as well has to kind of be developed at the same time.

Eric Hanselman

All those things have to come into play. Well, we will have to see where all this goes. But that is it for us for today. We are at time. Thank you both. This was a great discussion. We'll have to see where it shakes out.

Lynnette Luna

Thank you.

Dan Thompson

Yes, thanks.

Eric Hanselman

And that is it for this episode of Next in Tech. Thanks to our audience for staying with us. And thanks to our production team, including Caroline Wright, Caterina Iacoviello, Ethan Zimman and the Marketing and Events team and, of course, our studio lead, Kyle Cangialosi.

We hope that you'll join us for our next episode, where Nick Patience is going to be back in the podcast talking about AI and ML advancements and some of the pieces that he's looking at as AIOps and all the integration pieces that actually happen for putting AI and ML capabilities to work. I hope you will join us then because there is always something Next in Tech.