 to the telecom exchange CEO round table on innovative interconnections. Both for our guests here on site at telecom exchange NYC and for our viewers joining us on RCR TV and JSA TV. Our last panel of the day is entitled Innovative Interconnections in Their Impact on Business. The panel is moderated by my friend, Richard Lukic. More than 20 years of investment banking experience, Richard has originated, structured, executed hundreds of deals totaling over 100 billion, that's with the B, of transaction value. Mr. Lukic is a founder of Bank Street and aspires with his partners to create a premier middle market investment banking franchise focused on growth sectors of the global economy. Please welcome my friend Richard. So as I understand it, we're standing between everyone and their cocktails. We intend to make this as much fun. I was about to announce that my set the bar high because they intend to provoke and cause mischief here to make this the most enjoyable panel of the day. So with that being said, I'd like to maybe, first of all, thank you everyone. I was commenting to the JSA team that this event has actually continued to grow in its success that it's fair to say that it has outgrown this room, but kudos to all of you. I might ask every one of the panelists, many of whom are well known to the crowd here, to spend maybe a minute or two introducing themselves in their organization as a kickoff on the dialogue here and then we'll start the back and forth and I'd like to also, I guess a range for a question or two from the audience or at the end even though I know we're sort of running low on time because of the mischievous prior panel that paid into our time here, but we'll try to make up for it. We've got higher bandwidth on this panel. So maybe we'll start with Richard and work our way down the aisle here. Sure, so I'm Richard Steenberg and I'm the CTO, one of the co-founders of Packet Fabric. Packet Fabric is a completely greenfield network designed around the purpose of 100% automation and making networking services more like a cloud. So kind of a complete rethinking of the way that we provide services, both transport, interconnection, everything in that space. I'm Nazar Ahmad. I'm part of the network team at Facebook. Somebody said the other day, you guys are working on everything's deep sea to space. Anything in the middle network we're working on. So building infrastructure to support Facebook, serving the billions of users like I lose count on that stuff but essentially pretty much any country in the world we have users so connectivity there. And then also Facebook as a mission wants to connect more people to the internet. So working on a lot of technology development on broadband access side as well. Frank Ray, still working for a small software company in the Pacific Northwest. And I work on all of global network strategy and acquisition, building out our cloud services, supporting all of Microsoft's 200 online cloud properties. Patrick Coughlin, I'm Chief Revenue Officer for First Light. We were founded by Bill Gates. Oh, I'm sorry, I was reading from your script. We're a facilities based operator in New England and upstate New York. 9,600 root miles, 5,000 on-net buildings. I think what we're known for is taking customers from tier two, tier three, tier four and bringing them to tier one cities. And we manage both carry a wholesale and vertical markets. We have network into Montreal and some of the things that we're doing. Hopefully we'll assist some of the other players here in terms of diversity and creating new opportunities across the board. Gil Santely, CEO and founder of NJFX, had a company called Four Connections, sold it in 2008 to Cablevision, worked at my golf game for a while, got a really good tan. And 2015, we had an idea. And the idea was, let's do something right. Let's build an interconnection facility where it's meant to go. Let's put it where the cables come across the ocean. Let's create an intersection point that's meant to do what it does in terms of building it right. So NJFX was founded. It was built inside originally the Tata cable landing station, operated in the meet me room carrier neutral style. And in 2016 last year, we opened our doors to the world's first tier three co-location facility at a cable landing station. Today, we've got 48 more acres and we're putting up buildings. We're allowing others to do what only a few in the past could do. And that is interconnect with subsea cables in a place that's what's meant to be. And we've got seven new backhaul providers with multiple fiber cables giving optionality to all those subsea cables. So in essence, we're treating the subsea network the way it should have always been treated with a proper building and plenty of capacity to leave. Thank you, Gil. So much has been said today about these pesky content and OTT folks who drive unit pricing down and are looking at cost plus models as the basis for their approach to the market. Maybe picking up on some of those themes, I'd ask Frank and the job to comment on the evolution. I mean, I think we've seen also there is a difference between content players in terms of how they approach the marketplace. We've also seen principal plays with capital for ownership versus participatory syndicate propositions. Maybe comment a little bit about how you distinguish or differentiate your go-to-market strategy and maybe that is a way for us to maybe hear from two of the leading players who have been deploying capital in this space. Sure. No, you're in space now. All right. Sure. Fine. So our strategy is driven essentially by necessity. If you think about the volumes and amounts of traffic that we're pushing out of individual data centers, it's not sort of engineering-wise even feasible to buy in a lit capacity model because we're talking about turning up terabits every time that we turn up stuff and going through a ISPs deployment cycle, budget approval cycles and all of that for hardware is just impossible to do at our pace. So our environment tends to be really volatile and crazy growth at times, and so we need that engineering flexibility and then we need the control to be able to turn up stuff in a moment's notice. So outside of any of the business needs, there's a engineering need to have a lot of flexibility, a lot of capacity and some control over what we're deploying. Now, if you look at the financial side of things, it's the same conversation that once you need that much capacity it just makes no financial sense to actually try to buy in a lit capacity model. There's not a lot of value add on top of that other NISP can provide. It actually slows us down in a lot of ways to have that lit capacity model. So we end up sort of doing the dark fiber. Then the other engineering aspect for us is really that we tend to take a lot more risks and push technology a little further and that cycle benefits us. It's part of our fundamental design where if we can push the technology envelope and get a lot of excess capacity that becomes our buffer against big spikes. We get terabits or several terabit spikes all the time. But it's also redundant because now we're building capacity that is designed to be redundant and the excess capacity, as I said, becomes our buffer in that. So we have an advantage to push technology far more and take risks that a ISP may not be able to take because then they have other customers that cannot tolerate the failure rates. We can tolerate a lot more failure rates because our systems are designed that way. So there's like fundamental need for us to go build path rather than buying capacity. I'll let Frank get the list. Everything he just said. Now, we do face a lot of the similar issues and that's why we've partnered in some areas. We're not a content player. We're not an OTT. We are a cloud provider and we're providing different services to the market. So for us, a lot of the focus is around availability, reliability, resiliency, and that requires us to look at the world and the infrastructure in a slightly different way and then we run into a lot of the issues that Najem and his team faces. I think one of the things that differentiates us from a lot of the other players in our space and the other technology companies in our space is that the companies that we partner with in the market, they're also large customers of Microsoft, right? So there's a huge buy-sell relationship. So on the one hand, it creates its own challenges but it opens up its other opportunities and it creates, you know, requires you to be a little bit more creative in how you go about and working with those companies and partnering with them and trying to establish what your go-forward path is. Talk a little bit, before we leave the two of you, talk a little bit about where your organization's determined to use the balance sheet versus consumption from the marketplace. You mentioned that sometimes it's traffic-driven but there are also circumstances where it could be built by the balance sheet of a third party and you've opted to own the underlying asset in other cases, you've not. So maybe a little bit of distinction on where you make that determination. Our default is if you can buy something, we'll buy it because it saves us time. Engineering resources are precious. They are the most valuable commodity that we got. So if you can buy something that makes sense, then we will definitely buy it. And if you don't feel like we can get what we want, then we have to go try to build it. That's sort of most of the time our decision criteria. In a lot of cases, what we have seen, especially in the fiber world, is that a particular telco or ISP may have a business case that is just quite not there yet, just almost there but not quite. And then what we can do is we can come help in and say, okay, we'll get the business case across the line for you, you go build it and we'll take this much out of it. So we've done a bunch of those type of deals where we help a business case through the line and make that project happen. Often otherwise projects linger on for a long time just looking for the right amount of funding of the business case to happen. So those are really two criteria that we use. Frank, I think picking up on John's last comment, you've actually had some recent case studies as well where you all were the enabler of Express and some other projects as well. Maybe talk a minute or two about determining to do that and the approach you took in those instances or whether or not that's something you'll do again. Yes. We were just talking with Jamie about that earlier today. I think I've been saying for a while now the time lines, the development time lines and time to market cycle in today's digital economy continues to get faster and faster but the timelines to develop the kind of infrastructure that we all collectively need, it's not really getting any faster. In some cases it's gotten a lot slower. So that becomes a real key differentiator for us. Anywhere where we can help to try to accelerate what's being developed in the market but there needs to be a market for it. If we're just gonna develop something that only Microsoft needs, then we have to question it. We have to do our own fiduciary responsibility question whether or not that's actually something that's gonna be of value because ultimately we're all either customers of the people that are here or of Najum's company, of my company. So we all have to have a need for it and if we don't then we have to question of whether that's the right thing to do. Richard, maybe switching to you for a minute. Much has also been said today about new infrastructure and the proliferation of additional mesh networks on the global stage. Your business along with a couple other players in your space sort of is taking advantage of that in a way that's quite innovative and driving the cost paradigm, particularly for content users, but also others as well. Maybe talk a little bit about how you're approaching the marketplace in a way because it's sort of a nouveau approach to the marketplace that we're seeing more of. So I think a real key differentiator between what we're trying to do and other people in the space and by in the space I think most people mean has a portal and an API and does some type of automation is there's a lot of folks that are out there focused on connecting to public cloud, connecting to Amazon or Microsoft and it has its place, right? There's a value there. But what we're looking at doing is making networking more like a cloud. So I know these both involve the words networking cloud but they're very different concepts, right? We're looking at, in the same way that, you know, an Amazon kind of changed the model for how people use compute and storage by removing the need. You know, if you look historically, it wasn't that many years ago that every single company needed to have their own infrastructure, have their own racks, have their own people to go out there and configure it to now anyone anywhere can go to a website, put down a credit card and start spinning up cores. Why can't you do that in the field of networking, right? So that's, we've really taken the approach of how do you virtualize the network and how do you make it consumable? The other big aspect of this is, you know, as much as technology advances on the transport side, it's still fundamentally limited at the interconnection side. So a Microsoft or a Facebook will push terabits of capacity. I don't think people realize how much capacity these content guys push and probably content's not the right word, but how much these guys push between one data center and another to make web scale services work. That is driving the economy. That's driving the scale. That's why whenever I see people talk about, you know, there being a bottom of price compression, I laugh because that these guys are absolutely driving that scale to make that work, but it stops at the point where their network ends because now you've got to get to someone else's network and now you've got to figure out where's the right place to be to do that? What's the right infrastructure look like to do that? It's no longer under their control. It's someone else needs to do it. So our vision is kind of taking the technology that they've pioneered and turning it into not just connecting yourself to yourself, but connecting to a broader ecosystem and really making that bandwidth massive, easy to consume, very scalable, very quick. And talk a little bit about how you use technology as well to drive your unit pricing ability as well. Yeah, so like I said, you know, we really focus on kind of leveraging the technologies that have been built for building these large data center networks. You know, there's a lot of work going on in just data center networking and the packet side, where you're able to now build these massive terabit capacity devices that are very scalable. You're able to get the optic prices much, much lower by focusing on the volume that they've been using and the technology they've been using. Technologies like EVPN have come along to kind of enable these more scalable networks. DCI on the optical side, right, has really driven a large amount of business where it's now very easy to turn up, you know, today 200 gig waves and tomorrow 400 and by the end of the year, 600 gig waves. That type of technology is really driving the new evolution of a next generation network and then the real trick is how do you put software on that? How do you automate that? How do you make that virtualized? That's right. Exciting times in that space. Maybe switching to Gil for a minute because he touched on an element of the interconnection play that takes these hard network, whether they're subsea or terrestrial in nature and aggregates them at a point where traffic is distributed all over the world in many cases. Maybe Gil, talk a minute or two about what you see in this proliferation dynamic away from sort of the traditional core data center propositions to more of these interconnection rich type of platforms and you're obviously addressing one of the corridors with the largest density of landing station opportunities but there are other places in the world as well that have some similar opportunities. Tell us a little bit about what you've learned on the Jersey Shore that is something that we could see as a trend maybe on the world stage as we think about these kinds of opportunities. Sure. So over the last 18 months, we've spent time traveling the globe getting to meet all the international carriers within their countries and in the US and you'd be amazed to find out that these international carriers believe their subsea cables land in New York City. We try to explain to them there are no cable landings in lower Manhattan made to refuse to believe us but we finally had a few of these folks come and said there's the cable, walk them to the cage. That's a wet cable, see what it looks like? We're in New Jersey, right? Here is US cables, they go to Ashburn. You can pick and choose how things work. So what we've really done is disrupted and educated the marketplace. We've made the world aware that New York City is not a place where cables actually land. It's a great city. We all love to travel and come here but our network doesn't have to go through lower Manhattan. We don't have data centers left in New York City anymore. It's a great place to walk around, barely drive but a network really should consider the destination it's going to. So the interconnection point we created was a purpose bill facility that lets you pick and choose how you leave and come to the US and if you wanna travel between continents in one building with one cross-connect we can connect you between Brazil and Europe as well as the Caribbean. So we've got optionality for you that the market's never seen before. It's just really the next step in the evolution of interconnection. Great and I think we're seeing some of the themes that Gil just referenced in other parts of the world as well as we've been active in some additional geographies where you have concentrations of landing station capacity and terrestrial needs that actually need to access it. I think there'll be more opportunities to talk about it on the world stage. If I would add, we have another location that's being developed at the moment and it's down in Virginia, in the beach and they can do what we just did. They can try and put a data center right there and have a neutral facility connect between the Moray and the Bruisa Cable and for the first time you'll see that then, another interconnection point. The only other one we ever had in the US was called Nap of the Americas and everyone thought including me that Nap of the Americas where the cable actually landed but that's in Miami. The cables were landing in Boca, 50 miles north. So again, we're just educating a marketplace on how things really work and making sure the awareness is there with the marketplace. Yeah, to add to his point, I mean, we really are not that innovative with our interconnections. We've done everything traditionally. You come into either a head end that someone developed or a carrier hotel and then you're at the mercy of that carrier who owns that facility and some of them are very restrictive in terms of, A, the cost of cross-connecting or the cost of even getting you in there. A lot of them just wanna meet you somewhere and then they'll pull you in. I think we need more, whether it's a Gils company or another company, we need more companies that are building true carrier neutral facilities because to me, if you're holding either a Facebook or a Microsoft with a ransom note that says it's gonna cost you $5,000 a month to cross-connect. That's not innovative. And I think it holds back some of the things we're trying to do because these guys are doing, if they're hitting terabytes' worth of information, they wanna be able to quickly and seamlessly get rid of it and have it go around the world. I don't think we do that today. It's something we need to describe. Not to take over your panel, but what happened recently was... I... No, no. There was a fiber cut six months ago and the cable was down hard for days and the subsidy provider said, hey, I need some help with the deal out with another subsidy provider to transit over a bunch of traffic. It was gonna take 200 cross-connects to do it. The provider said, sure, that's gonna cost you X amount of dollars per month for a three-year contract. I need it for like a day, two days. In our model, we don't charge recurring fees. How can we help you? Let's do all these cross-connects in the next 24 hours, get them done for you. So, the industry needs carrier neutral where you're not doing predatory pricing on cross-connecting people to each other. And that's what we're saying. Let's get back to basics. The carriers need help, right? They need to work together. It's a collaborative industry and they have to have a marketplace where making a dollar is not the most important thing. We've got a higher calling in this industry, right? We've got to make this all work. Because these guys will just go out and build their own facility. We made it prohibitive. You could do that. 5,000 to cross-connect, yeah. Sorry. We just started a new business today. Free. I don't know why anybody would want to go to Virginia Beach anyway. It's crazy down there. But, you know, I'm just listening. I was just listening to Gill. Maybe part of the innovation needs to be that we start using different terminology for what we're talking about because I think a lot of the buildings that we're talking about, not charging cross-connects, they're not really... They are already carrier neutral, but the problem is still the same. So, maybe carrier neutral is not the word that we need to be using anymore. I don't know. That's just me. Yeah, and I think also maybe a good segue to Richard, again, on talking about virtualizing a lot of this interconnect activity as well because the infrastructure layer is certainly a big part of it. But I think with a preponderance of innovation around the virtualization of the entire function and the sharing of those interconnected facilities, it provides a whole new set of opportunities as well. Yeah, so I think if you look at the cycle of kind of the evolution of carrier neutral data centers, what you see is there's a lot of places that come along and say, we want to be competitive. We want to give away cross-connects for free. We just want to enable stuff, and the problem is no one's there. And it takes years, it takes a five-year cycle, right, for really the economy of different carriers to all shift to this is the new place to be. And by the time that happens, they suddenly figure out, oh, this is a great revenue stream. We need to monetize. And it goes up and up and up and up every little, every year. So the problem is that the interconnection itself is dependent upon being in a particular physical place, right? So I can go start the best data center tomorrow, and I can say interconnection is free. Why isn't everyone using it? Because it takes time and business and reasons to be there. So our approach is to virtualize it. So we're providing the interconnection without being tied to a particular physical space. So the data center itself isn't necessary. You still need a data center somewhere, but the goal is to make it so that it's completely agnostic as to where you're at. It should be as simple as easy. For people who are doing less than terabits, they probably don't need to be doing it themselves, right? And that's my definition of cloud and hyper-specialization is if we're able to do this and do it better and do it cheaper, more reliably than everyone else, if you're not doing these terabit scale connections, you probably don't need that physical infrastructure yourself. We're better able to virtualize that and then not be tied to that physical space. I want to come back to Pat for a second because when we saw him and his colleagues come into first light some years ago, company was growing sort of at 50 to 100% a year and that really was quite slow by their appetite and expectations. So they decided to inorganically juice that and today have actually created one of the most impressive platforms in the Northeast. Pat, talk a little bit about the motivations that have been behind your aggressive assembly and organic growth in the Northeast and how you distinguish that from an ordinary or organic approach that actually was performing quite well even before you started down this path. It's a great question. Scale means everything in this business. Frank won't talk to me if I have a small network that really doesn't do anything for him. So, you know, and that goes for the wireless, carries as much as picking on Frank because he's sitting next to me, and the job might talk to you. But, you know, scale means everything to my customers. You know, being able to, in a much larger fabric, solve problems for them, whether it's cell tower backhaul, whether it's diversity, whether it's large enterprise customers looking for, you know, internet access. Again, you know, as a bailiwick against the ILEC, those are all rationales for us expanding the footprint inorganically. And, you know, we can see it now. The conversations that we're in today, we would never have been a year and a half ago. And it's because of the scale and the size of the network. So we saw it as a great opportunity to, you know, expand the footprint, but also now, you know, develop a little bit of an international flair. We go into Canada in two different fashions. We're actually working on a third that will offer some diversity out of Western New York. Again, we couldn't do that if we didn't compile those assets, you know, today. And eventually we hope to creep down. I get a tan that's somewhat like Gil in the future, but interconnect with him, because again, we believe that the more points we touch, you know, the greater the opportunity field. Make sure Gil buys you a t-shirt at his golf club before you go into the facility. So I'm gonna probably turn to the audience in a minute, but I guess as we sort of all look out, we chatted a little bit about this whole data center infrastructure ecosystem, and we're starting to see an increasing orientation towards a plit proliferation of more and smaller sized facilities closer to edge infrastructure. And that obviously has implications for how people connect to it and how it is utilized both by content players as well as those pushing data to the cloud. Obviously that also has a profound impact on networks and Pat I think touched on scale. Scale sometimes is also importantly in the fiber and broadband space often determined by the mesh of network density that you have in a particular geography. How do you each see your companies adjusting to this migration path, if you will. And it's not to suggest by the way that it's away from, or predominantly away from core facilities. I feel like to some degree, the edge facilities are doing something incremental in a large measure, but I'd like to hear each of your own thoughts about how you see that proliferation affecting your business. We're starting with you, Richard. Yeah, so I think you're right. They do very different things, right? So the edge facilities are there to distribute content to the internet, things like that. And they're very different from the centralized compute where what you care about is low cost power and scalability of infrastructure. So very different things, but at the same time, you still need to think about interconnection as you start to do these edge facilities. A facility that's close to the customers is no good if the customers can't connect to it, can't get to it. And the problem is as we add more and more and more locations, you know, it used to be, it's really simple if you think about New York and you think there's only one place to go. If I go to 60 Hutts and I've got everything I need, well, that's not true, that doesn't work that way. And it definitely doesn't work that way as you start to go to these more edge facilities and you try to figure out where is that critical mass. And that's where I think there's a role for, like I said, virtualizing the network piece of it, is not to be tied to that particular physical space, but to be tied to how does that space interconnect with other networks? How does that, is there a location that it does make sense for people to start to put their edge content? You still, you need innovative ways to interconnect the backbone network, the users that get there, and it's something that hasn't really been tackled before. That's I think one of the challenges we all need to address if we're gonna make that type of model work better. Good job. Okay, I think I'm gonna try and say what Ras said in different words. All right. A lot of physics, you can't change. That's, physics is absolute. Buildings run out, doesn't matter what you do. And if all of us want to be in business, it'd better run out of capacity. That's a good thing. That means the internet's growing, the network's growing, we're all using it. So get with it, it's okay. Building's running out of space. So to Ras's point, if you're tied to a building and you will have to be in one building, then it's just not gonna scale, you're never gonna grow up, you're never gonna be big. So if you wanna be big, build those abstraction layers, get away from the physical constraints of anything. As I said, you can't change the physics of things. So, figure out abstractions that you can actually not worry about where you physically are, worry about what connectivity you need and do that. And to my carrier partners, entirely, even sort of in this conversation, it's entirely too much about cost. I would argue that if you make it a lot easier for somebody to use it, you will have more customers. I will show you this. We all go pay 800 bucks and get in line to buy this thing. Seriously, it's not worth 800 bucks, but we do. Because it's fun to use, it's easy to use and you get there. Make it to connect that way that if your network is growing that you can actually point and click and build network on a heartbeat and move on and serve your business. So instead of having to worry about focusing building network, you're focusing building your business and the network can go. So the carrier partners have to think about that is instead of worrying about the cost of, hey, what the cross-connect I can charge, what I can do with it. It's making it so easy and so usable that a four-year-old can take this and run videos on it in 10 minutes. And you don't have to have a manual for it. Can you make the carrier interconnects that way? That's sort of what it takes and that's sort of the racist point as well is that you don't really care about where connectivity is or where physically you are as long as you do the connectivity. The rest of the stuff, the scale folks will figure out a lot of other things because we're putting 100 racks at our edge. We obviously can't put it in 60 Hudson's. We have to split it out and build different buildings and all of that. And which is also a good thing because that also helps to our disaster recovery conversation. One building gets flooded, you're not screwed in that region. You still have opportunities to do a bunch of other things. So you'll have more buildings, you'll build more stuff and we'll figure out how to put where. But that's sort of how interconnection market needs to go. I think you owe him a commission. That was a great commercial. I think I'm gonna go over there to get a check. He's a tech guy, just to be honest. All right. Respecting the hierarchy. Right. Yeah, I'll just try the 3.0 version then. Ooh, no Microsoft jokes. No, yeah, so I guess, that's all right. It probably won't work the first time then. You know, the thing, the thing is you mentioned now a lot of data centers are going out and proliferating in the metros and every market is gonna be different, right? Gil is seeing that in his business where it might be different doing what he's trying to do on the East Coast of New Jersey or East Coast of the US versus what's happening say on the West Coast of Africa. Because the infrastructure is different, the market is different, dynamics are different. So the same is true on how you're interconnecting and what you're building out, right? The physics are the same. It still takes cement as long as it takes to dry. But you know, your right of ways, your building out infrastructure, your construction resources, all of these things vary from market to market, how people are getting to you, what they're doing there, what they're capable of. So, you know, I think we talked about this last year. If there was a one model fits all, then we'd already be drinking Bellini's at this point. But there isn't. So I guess that's a good thing for all of us because it's gonna keep us busy for years to come. I agree. I'll take it a little bit on the other side in terms of, you know, we're just scratching the surface in terms of what we're seeing for on the wireless infrastructure. And that's a lot of what we're dealing with at our company. You know, the densification projects that are going on, CRAN 5G, the amount of fiber that is being built or about to be built over the next, call it 24 months is like nothing we will ever see. Well, I have seen, I should say nothing we won't see. We haven't seen. Because it's the amount of fiber that all of these wireless players are gonna be taking down in metros and again as they extend their footprint. And which is gonna be great for all everybody sitting at this table because it means more bandwidth is coming down, you know, the pipe and needs to go somewhere, needs to get distributed. So, you know, obviously we're gearing up on that on the first light side. We have the network infrastructure, you know, Richard touched on it. It was one of the reasons why we decided to go on a path of M&A was because again, you needed that scalability but you also need the engineering firepower. And as like Najam said earlier, you know, you can't look at this payback as being 12, 24 months. You have to take it out and be a partner with some of these wireless folks. They are building out your infrastructure or helping you build it out. But I think Frank said this to me a year ago. I don't wanna pay, I don't wanna be your infrastructure bank, you know, where I'm paying for everything. You know, this has to be a partnership as you're doing some of these projects. So, you know, that's the way we look at it. But, you know, for everybody in this room, I think the exciting thing is what's coming down, again, the pipe here is large fiber deployments, lots of bandwidth, lots of opportunity for everybody in this room. I agree with you Pat, it's dumb. Thank God. It's embarrassing. We're all good, we're all good. Everything's working out. But thanks to Facebook, Microsoft, the rest of the OTTs, the world has changed forever. It's only been the last 15 years that we're addicted to these phones in our pocket. But more importantly, as we become an aging population, our cars will get us to places that are actually driving, which I would really appreciate 20 years from now. The Ubers of the world. The applications are endless. But at the end of the day, where do the apps sit? That's up to the smart guys to figure out how to do it cost effectively and they need optionality. And to have that optionality to say where my app is gonna sit, where it's gonna reside, how I'm gonna manage it, means that they've bought through the network architecture and all the diversity needed to do it. So, those are the secrets that keep us all safe. They need us in terms of infrastructure. You know, we're fortunate enough to be in a great location with great partners like Tata and Aquacoms, across the ocean in two different ways. Because at the end of the day, customers need to know how do things work. Let's make sure we design things that are gonna last for the future. And let's make sure that the lights are always on, but more importantly, the internet's always working. That's what we care about today. Absolutely. I might turn it to the floor for a question or two. One in the back. Just a quick technical question. Regarding integrated objects, the router technology curve seems to move slower than the transport technology curve. Here is what you see in the market. Do you see a point where technology and optics will be just in the routers and the forklift out of transport here? I'm just curious on your thoughts, a little bit of problem-ascaping there for you. I'll take that. So, yeah, I think we've seen a move recently that people trying to integrate the long-haul optics and the routers, but it's not actually that new, right? It's an idea that I've seen Cisco trying to do for 15, 20 years now, and it's failed every time. And I think there's a reason for that. Optical technology develops at a vastly different rate than router technology, switch technology. And the problem, the penalty that you pay in interconnecting these two things is you've now, you've coupled them together in a way that is very difficult, right? You've made now a low-density line card that isn't as reusable, isn't as scalable, isn't as cheap, and probably came to the market 12 to 18 months slower than it should have if you had just been able to decouple the two. And the cost of interconnection of these guys is not that difficult, right? That's what's driving a lot of the QSP28 SR4 market today, right? Millions and millions of these units are shipping because for the cost of a $200 optic, you can replace the need to lock yourself into a dedicated piece of hardware that's not a good fit for your network for years to come. So I personally see no life in that. I think it's a really bad design and something that's going to fail again. The only thing I'd add is, I agree with Raz on his points, in certain scenarios, like a data center inside the data center model, you have a ton more interconnects inside the data center than outside the data center. And if you want to go really high, like 400 gigs, one of the options may be onboard optics. And that's say, can we make this work in this time frame? I think it will eventually still come out of the box and be a pluggable again, but initially might be an option to look at onboard type stuff. So more of a technology reason for doing it, because we need capacities faster than more of a business guys. Any other questions? I have a question. I've reached out on the bill and venial market place, but now we're making, basically, adjusted by society, what is your medication. So it's how limbo we are, and that's where these interconnections come into play. So have you started to see any trend where those interconnections are faster, where you're not debating five to 10 days for a cross-connect to go through a process to get from one floor to another? Or are you still starting to feel some of these pain? That's yours. Yeah, that's all you're going to say. Obviously, there's a pain point in the speed of interconnection, whether it's one day or five days, but it's more than that. It's the pain point in the number of people that were needed to place that order in the first place, to run it through sales, to run it through provisioning, to get an LOA, to get a correct LOA, to project manage the plugging in of the patch cable to the router. That's the thing that takes weeks or months, in most cases. So any large company has hundreds or thousands of people that are project managing this type of stuff. That's what's slowing down the evolution of the real just-in-time stuff. So there's nothing a data center can do. If they made the cross-connect happen in six hours instead of 12 or 24, that doesn't really change things. What changes things is the business drivers around that. How do you get the order placed and get it provisioned and get it configured and get it out there? That's where you need to look to the model that's worked great in the cloud space, which is real infrastructure around automation, around treating it like a, it's got to all be done in software. It's got to all be orchestrated. It's got to be measured. It's got to be, you really have to think about it in that kind of scale, and then you open up a whole different kind of business, right? You open up people who are able to reconfigure networks in very different ways. They're able to really move stuff into the cloud in much more pervasive ways than they were before because of the six-month process to get one interconnection for one thing through one carrier. That's the real value. I'd like to add, one of the things that we did is since we don't charge our current piece of cross-connects, we said, connect to everybody. When you come in, order 48 and connect to everybody. You're all set. Congratulations. You don't call me at this point. Just go ahead and light up your circuits and do what you have to do. Get your NNI's rolling. So by getting rid of the cost, in that sense, now you can just connect to everyone at the same time when you first walk in. It's like obviating the value of interconnection in some ways, and bringing it back to, this is the location where you ought to be. So I think there will be some battles along those approaches, and I've learned something from Richard today, which is stack the audience with people who don't love, use some easy questions. One more question. Thank you. I just want to ask a couple of content just to hear a lot about current demand perspective. How do we cross the point at which it's about the consumerization of the network and how to go from a demand perspective? Really just about high definition video being consumed, download, and upload with AR and VR. How do you see actually driving in far more than enterprise or clad? Are you driving that infrastructure to the point where we can find our own government? I can give you a sort of a specific answer, but there's a generic answer as well. We always, sort of every year, we do these five-year projections what our growths are going to be, and we look at these five-year chart saying, whoa, hang on. Boom. OK, first two, three years is fine. This level of growth, the last two years, it can't be that big, and we'll just cut it down, and we'll bring it down artificially because we don't know any better. We've been wrong every time. Every time it's higher than we thought. So, yeah, traffic is growing. The trouble is that, and I think I've said that, I guess I'm using this sort of example again, there's so much compute in this thing in your hand, and then there's an army of developers out there writing code coming up with ideas to use this compute that you have in your hand. Guess what? Traffic's not going down. And that's been my experience every time we've done projections, we've been wrong. You can come up with different scenarios of what it takes to do. Our patterns are changing for sure. We used to do a lot of messaging. We started doing a lot of photos, and then we started now a lot of videos. We're starting to do a lot of 360 videos now. We're starting to see that. A lot of VR is going to happen. So experiences are going to change over time, and things are going to move. But one thing that's been constant, that the growth has been faster than anybody has projected. Now, you do have to then split the growth in two ways, at least the way we think about this is one is machine to user traffic, which grows at a different pace, because that's about the user and where they are and what type of services they can consume, what device they have in their hands, and things like that. So it's a very different sort of growth curve, and we worry a lot about projecting that side model. The other one is the more difficult one, which is machine to machine traffic. And machine to machine traffic is several times more than machine to user traffic. And Frank would tell you the same, Google will tell you the same. Machine to machine traffic tends to be a lot more. And it's more unpredictable, because we at the same time we have hundreds of thousands of compute machines behind it, and lots of software developers coming up with new things to do that they want to do, and they just unpredictably add more traffic to it. So traffic just grows on that side as well. Plus those pesky machines never seem to go to sleep. They're just always working and always buzzing away, you know? I'm just asking, at the moment, if you can see we're driven in space, there seems to be an awful lot about generation of in-consumption of video and generation of video. Before we get into the machine to machine world, which looks to me like it's two or three years down the line before that, it's really going to start to actually try to be tough. I don't know if you can say that. No. No. No, it's happening. I mean, if you think about simple use cases, all right, too much information conversation going to happen. You know, you go to get a massage. You used to go find a place. Now there's an app. Don't match, you do a masseuse near you, right? So people will keep coming up with new ways to solve problems, because you do have compute on both ends, the server side and your CPE side, right? And you've got to build a network to connect the two together. My apps usually just tell me where to go get pizza, but. Not a massage. I said too much information. You know, whatever folks are both. No, but if you look at the number of companies and enterprises out there that still have in transition to the cloud, they're going to move to the cloud and they're going to do what they're doing today, but when they get to the cloud and they move their workloads around, then they're going to start to figure out how they also evolve and innovate their own business, which is then going to change the dynamics again, right? So, you know, we're just at the beginning of where we stand. How that translates into traffic? I mean, yeah, if we could answer that, then we would probably, we would already be drinking bullies right now with Gil at his club. So. So one thing I would add to that is, you know, there's still whole industries that are built around mailing hard drives today, because they're not able to get the connectivity that they want when they want it, how they want it. There's still a ton of room for growth in that space. And when you talk about things like self-driving cars, now you've got to start thinking about massive amounts of telemetry between massive amounts of devices that are out there that need to talk to each other. So that machine-to-machine traffic is entirely the growth. That's, there's more machines than there are humans, and that's really what's driving the bandwidth needs. Yeah. They'll be building more machines. And I was going to actually just piggyback on that last comment and say that we're also going to see some of the same machines doing much more and actually communicating much more, as well as so many, many more devices. Obviously, cars was featured here today, but all kinds of other devices that are going to start to actually communicate as well. And I think that's just the beginning of all kinds of commercial applications and the like. So we are in many ways in a technology-driven as well as a behavioral modification dynamic that is all seemingly compounding the proliferation of network utilization. So with that, I'm going to bring this to a close. I have to admit, I did not think we would actually be able to cover the global networking and communications infrastructure and to the depth of pizza and massages, but I think we covered a lot of ground today and I appreciate everyone's attention. See you all next year.