What's Up with Tech?
Tech Transformation with Evan Kirstel: A podcast exploring the latest trends and innovations in the tech industry, and how businesses can leverage them for growth, diving into the world of B2B, discussing strategies, trends, and sharing insights from industry leaders!
With over three decades in telecom and IT, I've mastered the art of transforming social media into a dynamic platform for audience engagement, community building, and establishing thought leadership. My approach isn't about personal brand promotion but about delivering educational and informative content to cultivate a sustainable, long-term business presence. I am the leading content creator in areas like Enterprise AI, UCaaS, CPaaS, CCaaS, Telecom/5G and more!
What's Up with Tech?
Revolutionizing Network Services: Lumen's AI-Driven Fiber Connectivity, Strategic Partnerships, and Future Innovations
Interested in being a guest? Email us at admin@evankirstel.com
Prepare to unlock the secrets behind the AI-driven transformation of the network services industry with insights from Dave Ward, Lumen's CTO and Chief Product Officer. Discover how Lumen is laying the groundwork for the AI economy by developing advanced fiber connectivity solutions that meet the surging bandwidth demands of AI data centers and enterprises. Dave delves into the intricacies of AI workloads versus traditional network traffic and highlights their groundbreaking partnerships with companies like Corning to push the boundaries of fiber capacity and performance. Get ready to learn about the future of networking technology and the exciting advancements on the horizon.
In a compelling discussion, we explore Lumen's strategic collaboration with Microsoft to revolutionize AI infrastructure. Dave walks us through the deployment of extensive fiber networks for GPU-based data centers, strategically located near renewable energy sources to facilitate sustainable growth. Uncover the unique advantages Lumen holds with their extensive unused conduit and cutting-edge fiber technology, setting them apart as leaders in high-speed, low-latency connectivity for enterprise AI. This episode is a must-listen for anyone interested in the evolution of network technology, as we discuss Lumen's focus on enterprise services and their commitment to enabling cloud-native and multi-cloud architectures for businesses. Don't miss out on this deep dive into the future of AI and networking!
More at https://linktr.ee/EvanKirstel
Hey everybody Excited to dive into the world of AI and how it's reshaping the network services landscape. Really excited to have Lumen here and discuss their role in this quite transformative journey. Dave, how are you Doing?
Speaker 2:really well, levin, happy to be with you today and a great topic for us to talk about.
Speaker 1:It is indeed Thanks for being here, really excited for this conversation. Maybe start by introducing yourself as Lumen's CTO, a little bit about your background and the current vision mission at Lumen.
Speaker 2:You bet. So I'm Dave Ward. As Evan mentioned, I'm the CTO and Chief Product Officer here at Lumen, and that includes not only setting the vision for the portfolio, but also building it and delivering it across all the systems we have inside Lumen. Lumen's mission today, and most folks may know us as the largest service provider in the world with respect to the amount of peering. That goes on meaning we are the internet for the US. Everybody connects through us one way or another, but we're a facilities-based carrier, and so that means that we are really into fiber and connectivity and moving around big, big amounts of bandwidth around the country.
Speaker 1:So much bandwidth, so much data being moved these days, and hence AI is such a hot topic. Maybe give us your point of view, a unique perspective, on how the rise of AI has impacted demand for your network services and what's the future look like for Lumen and its customers over the upcoming years?
Speaker 2:Well, evan, as you said, AI is hotter than Hansel right now, and so what I mean by that and I'm taking a very specific perspective on what it means for networking, because we've all heard a lot of messages that AI is everything it's dessert topping and floor wax and everything in between but for a networking person, we feel we've kind of been left out of the conversation, and what I mean by that is lots of people have. For a networking person, we feel we've kind of been left out of the conversation, and what I mean by that is lots of people have talked about GPUs and AI data centers and how much data is required to train these models. But to have an AI strategy, not only do you have to have a data center strategy and a data strategy, you have to have a networking strategy to move that data to the data centers. And so we've really been focusing this year as a number of things have come together and we'll talk about that in a second on building out the backbone for the AI economy, and what that means is we are providing fiber and fiber connectivity to these new AI data centers and working with our enterprise customers to be able to take advantage of all those resources, and that's both with the big cloud service providers as well as private data center operators that are now building multi-hundred megawatt and gigawatt data center data center parks, all requiring lots and lots of bandwidth, and that bandwidth is lots of bandwidth and that bandwidth is coinciding with GPUs and the AI. Economy and algorithms have reached one point in 2024.
Speaker 2:But also fiber and networking technology also has hit a new, higher plateau of capability. That wasn't possible just a few years ago, and let me give you a couple examples. One is that we partner deeply with Corning. Corning is able to fit 1,728 fibers in a cable. That at the start of my career we were excited about, eight fibers for the same size cable, and then it got to 144 and 288 blew our minds. But what's interesting about this right now is that it's obviously an order of magnitude higher amount of fiber per cable. Now here at Lumen we've got lots and lots of conduit in the ground and lots of space to be able to run this fiber, and this is what we're pulling to these data centers just oodles and oodles of fiber and wavelength capabilities, plus Ethernet and IP to connect the enterprises, to build a fabric for those data centers and again to be that backbone of the AI economy.
Speaker 1:Wow, that's so exciting. Backbone of the AI economy that's something I haven't heard before. And speaking of which, those AI workloads, how are they different from traditional network traffic that we know for decades, traditional network traffic that we know for decades, and what kinds of businesses are most impacted by these new appetite for AI?
Speaker 2:Well, what's interesting is that, as many have said, there's several phases to the AI journey and one is just training data, and training data means your data needs to be in these data centers running on those GPUs, and so really you can think of it, evan, as a huge step function in the amount of bandwidth necessary to move this data. Data we're talking about and it could be part of an IT transition to cloud is all of the documents ever produced and all the data ever produced by an enterprise documents ever produced and all the data ever produced by an enterprise, because what the purpose of AI, of course, is to unlock that data and train the models such that we know what's inside it. So the step function that's really interesting is not only in this first phase of getting data to those data centers, it's also that where the algorithms are located and where the data is stored so, for example, there's in the cloud. Of course, there are specific niches and segments that have arisen. So, for example, there's storage clouds out there and we know these all very well, and we talk about S3, buckets and Snowflake and other forms of storage clouds that are out there, just to name a couple that data now needs to get to those GPU clouds. So the step function for us is coming at us either in whole dark fiber or multiple fibers, or also customers are demanding tens and hundreds of 400 gig waves at a time. That is just a fundamentally different proposition than we've ever heard before, and I'm going to take this one step forward as the data needs to get off the enterprise. This probably is going to lead to a step function in access bandwidth for those enterprises. We're all really happy One gig, 10 gig and now state of the art might be 100 gig, but what's capable? Or 400 gig delivery and very soon 800 gig and 1.6 terabit. Things start getting very, very interesting at those units of bandwidth because the second phase we'll call it that of AI being inference, where you have trained models and now you want to run data against it. You need to have extremely low latency and extremely high throughput to be able to work in that second phase. And then, third, when we're talking about reinforcement, learning and human interaction with that data. You can think about this with all of the existing systems that are in an enterprise's IT systems, from the sales forces to workdays, to autodesks and on and on. These are all humans interacting with that data and therefore reinforcing the use of that data and constantly retraining those models. So the units of bandwidth that we're talking about moving forward from 2024 for, let's say, the next five or seven years, are just fundamentally different units of bandwidth than we've ever seen before for enterprises or building backbone services.
Speaker 2:The last point I wanted to make is in the networking world there's been a big push for things like software-defined WAN or SD-WAN, and those often have been used to replace things like MPLS VPN circuits, which has been around for 15 plus years. And what's interesting about SD-WAN is those running over consumer broadband circuits just are not at the unit of bandwidth that we're talking about. And MPLS VPNs are also a hub and spoke architecture between campus prem and not really working directly with or building the network to private data center operators, gpu data centers or to CSPs, and so what's interesting is that we've got an architectural transition away from MPLS VPNs. We have another architectural transition of SD-WAN that's not going to meet it, and we have the rise of GPUs and suddenly new fiber technology. So, evan, it's a brand new world out there needing a new architecture.
Speaker 1:Wow, really exciting time. And gosh, I hear conversations support these AI demands on the network side, specifically on reliability and speed and other network strategies.
Speaker 2:Really, what we've done is we've created, as I mentioned, a new architecture which we're calling private connectivity fabric, which is a mesh-based architecture, not hub and spoke, not built by gateways or pinch points or hourglass designs in our network, but second to enable that. That's the architecture. Private connectivity fabric, the delivery mechanism is really a platform, api-driven, portal-driven on top of our services for being able to monitor dark fiber, of our services for being able to monitor dark fiber, enable waves on demand, ethernet, ip, private IP as a service and on demand, and that digital platform is really part of the transformation that Lumen has taken on over the last couple of years and, into the future, matched up with these bandwidth capabilities. So we have a new architecture, getting away from hub and spoke VPNs and getting away from overlay pasta salad and getting to one which is basically is delivered via this platform. That can not only work with the workloads that are moving the data.
Speaker 2:Let's say you want to move 30 petabytes in a month. That'll take you a number of hundred gigs to be able to do that. But now let's say you want to bring this down to two weeks. We're going to bring this down to a day and you're moving petabytes of data. To be able to train those models, you need a consumption model as a service that's going to match that as well, and so we built a new architecture. We have a new digital platform. We have new consumption model. New architecture we have a new digital platform, we have new consumption model, and now we have this new bandwidth that we've constructed into our network, and that matchup is the fundamental strategy that Lumen has going forward.
Speaker 1:Wow, phenomenal. So in terms of your private connectivity fabric solution, you and I know all networks aren't created equally, but a lot of CEOs aren't networking gurus per se, so how do you think about customizing that for different business needs, different sizes, different industries, healthcare, finance versus manufacturing? Maybe describe how that looks in the real world.
Speaker 2:So what this really means is that, instead of building out an enterprise network again, as I mentioned, hubbin spoke to all the campus or prems or branches that when an enterprise is constructing a network, they can build any of the links that are necessary in that mesh topology to not only enable their own enterprise but also enable ecosystems of partners around it. We have several payment customers that have built private connectivity fabrics across the nation and they enable their ecosystem of payers that are part of their system to join the ecosystem simplistically through our API-driven platform, as I mentioned. That's one example. Also, as folks are trying to get to these new multi-hundred-megawatt, gigawatt data center parks and they want dark fiber or they want waves, we know that right now that the key to business post-pandemic is how fast can I get that connectivity?
Speaker 2:That legacy telco manner of taking months and months to deliver a circuit is really what we're going after with this digital platform and enabling also to simplify the number of suppliers and telco partners that are necessary to be able to deliver this. So we have a very large construction effort underway in the nation to be able to connect data centers, clouds, new on-ramps and follow the new data center corridors that are being built, evan, as you're aware, there's over 2,000 data centers being built in the US, as we're speaking right now, and those all need connectivity to be able to get there, and that is Lumen's target and being able to deliver it via that digital platform.
Speaker 1:Fantastic Another blockbuster announcement with Microsoft that they've chosen Lumen to build out their infrastructure to support their AI efforts. So industries like retail and logistics, education, so many others, are going to benefit from a kind of AI makeover. Tell us about that collaboration and some of the benefits.
Speaker 2:The work we're doing with Microsoft is not only to build new fiber rings for them and connectivity into their major new GPU-based data centers, but really to also build a backbone for that company itself and towards that end, it's not just a Microsoft announcement, karina's soft announcement. We have a number of customers who are constructing with Lumen and Evan. To put it into perspective, the amount of fiber that we're talking about, just with the announcement that you described, end-to-end that fiber would go halfway to Venus from Earth. So we are talking about a lot of fiber miles being put in the ground and a lot of route miles. Now Lumen's taking advantage of some of these by putting our own fiber in the ground to be able to give access to enterprises to this AI fabric or AI backbone, which those who, again, who are in the networking industry really understand that these AI data centers are following certain corridors in the country.
Speaker 2:Those corridors are also related to new power construction going on in the US and, frequently, the rise of renewable energy, and so many of these data centers are not in the traditional places that we would have thought about them before, near major urban centers. Instead, they are where renewable resources and renewable energy can be constructed, so they're often in deserts or in far-flung areas throughout the country. So what's really fascinating about this, evan, is that not only is bandwidth important and latency being important, but, of course, the ability for us to create for enterprises a network architecture, that is, that private connectivity fabric to the GPU data centers. They need to go private data centers, cloud on ramps, all their prems and enabling us, with the Microsoft announcement, is to begin to build these new routes and put this major amount of fiber in the ground.
Speaker 1:Incredible. Can't wait to see how this progresses. So the beauty of this industry is customers have lots of choice for connectivity and when they're looking to leverage AI, they have multiple options. Why, lumen, you have talked about many of your sort of unique strengths. I mean scalability, integration, security, compliance. What are you seeing? Why are our customers choosing you for their sort of unique AI deployments?
Speaker 2:It actually goes back to the history of the company and what makes Lumen unique in this position.
Speaker 2:Originally, when companies like Level 3 were built, let's say 25 years ago, they put an incredible amount of conduit into the ground and we have lots and lots of that being unused, to be honest and so we were uniquely positioned as a service provider who had conduit. With now, this brand new fiber that I talked about a few minutes ago, we can have an order of magnitude more bandwidth, therefore more efficiency in utilization of our infrastructure, and then, by having the metro reach of being such a large network that we have the lumen topology as it exists, we've been transforming that now to be 100 gig capable, 400 gig capable, and you're going to soon see those experiments and the release of 800 gig and 1.6T, as I mentioned. Now, why Lumen? We have the infrastructure in the ground, we have the fiber partnerships, we have a longstanding tradition of providing transport, ethernet and IP services to our customer, and so, therefore, we're uniquely positioned as the only one with these capabilities and this infrastructure right now to take advantage and deliver this AI backbone.
Speaker 1:Wow, that's really intriguing, and this market space is evolving so rapidly. You're doing so much work to prepare the network for future growth. Maybe talk about some of the investments in new technology infrastructure. You're doing so much work to prepare the network for future growth. Maybe talk about some of the investments in new technology infrastructure you're making. How are you going to stay ahead of the curve here? Because the curve is accelerating, to say the least.
Speaker 2:Well, what is interesting is I mentioned a number of things coming together. We talked about the GPU data centers being built. We talked about the fiber We've got the conduit but also in transport, the rise of ZR and ZR plus pluggables into our routing and switching equipment. The density of the routing devices are now 2RU, with 52 terabit chips available in them, two terabit chips available in them. It just is.
Speaker 2:It's been a long time coming and, having spent a lot of my career building a lot of those packet flinging devices and photonic devices, my career is now running the network and being able to take advantage of this incredible density that is available in the market fiber transport and packet switching and packet routing that hasn't existed before and it's been about 10 or 15 years since such a fundamental transition, and really that transition was getting to 100 gig ports.
Speaker 2:Then it was a quick jump to 400 gig ports and you can see, with the equipment and the transport equipment that's available, that jump to 1.6T is within our reach. And so really preparing our network has been constantly transforming it from an accumulation of acquisitions to one in which we have seamless integration across our network, across our network, and now we're focusing on being able to provide that extremely high-speed backbone between our metro areas and use our entire topology to be able to service our enterprise customers. Interestingly, evan Lumen is the only ILEC in the US, one that doesn't have a mobile phone play, so we're not chasing the SIM. Instead, we're focused on enterprise business services and fiber-attached enterprise business services and really are the only company out there focusing on that part of the market. And so, given we have a singular focus and we have a singular vision of getting enterprises to be enabled with AI, to get them cloud-native, to get them to build hybrid and multi-cloud architectures, that's driven with our portfolio direction and the construction of our network.
Speaker 1:Wow, fantastic opportunity, and you've been in this industry a long time, no disrespect, but it sounds like you're a kid in the candy store when it comes to a lot of these emerging technologies. What are you personally most excited about that will help support AI in the future?
Speaker 2:Evan, I have been in the industry a while. I'm experienced. We're going to go with that. But look, I'm one of these networking geeks that has every one of my red and white blood cells, has an IPv4 and v6 address. And look, if you're not, if it can't be done on the internet, I'm not sure why we're doing it. And if you're not doing it in the cloud, you're probably not doing it the right way.
Speaker 2:So for me and this sounds a little bit grandiose I fundamentally want to fix some of the internet architectural flaws and for me, cloud needs to be a first-class citizen. Data centers need to be a first-class citizen. I also would like to be able to fundamentally change, in this country at least, how fiber hotels are constructed. I don't believe that we need to have these hourglass designs in our internet architecture, or the huge cost and rise of meet-me-rooms doesn't make me very happy, evan. So I'm here at Lumen, not only to be a part of the largest expansion of the internet in my lifetime and I was here in the late 90s and I was here in the 2000s and I've been around for a bit but just by the amount of bandwidth that we're talking about fiber and packet switching, packet routing. This is truly the largest expansion of the Internet and, evan, I got to be a part of it. This has been what I've been waiting for for a long time in my career, and also the ability, with a network, to solve some of these fundamental Internet architectural problems.
Speaker 1:Wow, wonderful. Well, your enthusiasm is infectious and we haven't even dived into areas like quantum or 6G or massive IoT or so many other topics. We'll save those for another episode, but thanks so much for sharing your vision and mission, and it's an exciting one. Thanks, dave. Hey, thanks, evan. Appreciate the time. Likewise, thanks and take care. Everyone Follow Lumen on its various social channels. They put out some amazing educational content worth a read. Take care.