Wired for Intelligence: Building Trusted Networks for the AI Era

What does 'AI-ready infrastructure' truly entail for enterprises? 🤔

‍

Discover the answers with host Patrick Moorhead, and Six Five Summit Intelligent Edge spotlight speaker, Justin Hotard, President and CEO of Nokia. This session highlights the foundational role of networks in scaling AI solutions.

‍

Key takeaways include:

‍

🔹The Indispensable Role of Trusted Connectivity: Why high-performing, secure connectivity isn't just an advantage, but a non-negotiable foundation for successful AI applications and enterprise scale.

‍

🔹Strategic Investments for AI-Ready Infrastructure: Understand the crucial investments enterprises must make in connectivity infrastructure to meet AI's unique demands, identifying key areas for strategic capital allocation.

‍

🔹AI's Impact on Telecom Networks & Nokia's Vision: AI is reshaping the evolution of telecommunications networks – Justin shares what this means for Nokia's customers, signaling a new era of network capabilities.

‍

🔹Building AI-Ready Data Centers & Ensuring Security: Define the essential characteristics that equip a data center for demanding AI workloads, and learn about Nokia's proactive approach to ensuring network security and reliability in the face of escalating AI adoption.

‍

Learn more at Nokia.

‍

Watch the video below at Six Five Media, and be sure to subscribe to our YouTube channel, so you never miss an episode.

‍

Or listen to the audio here:

‍

Patrick Moorhead: Welcome back everybody to this Six Five Summit. We're in our sixth year here and we are talking about making AI work for the businesses and even the end consumers. You probably know this already but you are in the intelligent edge track here. Here a track dedicated to networks, infrastructure, real time systems, powering AI at the edge. I am joined today by Justin Hotard, President CEO of Nokia from Finland by the way, for a timely discussion of what it really means to build secure and high performing AI ready networks. And you know we talked a lot on the show about, hey, there's so much talk about GPUs and compute and memory and storage but in the end if you can't connect all that together, data centers together, enterprises inside, it's pretty much for not or it's radically inefficient. Welcome Justin.

‍

Justin Hotard: Pat, it's great to be with you today and thanks for having me.

‍

Patrick Moorhead: Definitely, definitely. I love the background, I love that you're broadcasting from your headquarters here, but let's dive right in. You've talked about trusted high performance connectivity, a must have for AI, obviously I agree. But can you provide maybe break down on why it is such a big deal out there based on what you're seeing technologically or what your customers are talking about?

‍

Justin Hotard: Yeah. So I think first of all I'll start with the distribution of networks. If you think about where AI data centers are being built today, they're being built in new locations. We're seeing lots of news about large data centers being built in new markets, in some cases in new countries. There's discussions, the latest discussions coming out are around these concepts of AI embassies. And what this means is that our transport networks need to be modernized and scaled to support this demand. Because we're talking about incredible bandwidth and in some cases we're not actually keeping the data and the compute in the same locations, in some cases the data and the computer separate. And so even in those cases latency starts to matter as well. And so as we think about this, Metro networks, long range transport networks, all of these need to be upgraded. And that's where we're seeing a huge increase in optical demand and our front end routing demand. Great growth opportunities for Nokia. 

‍

Then when you get inside the data center density, we're talking about NVL72 if you're talking about an Nvidia system or 144 GPUs in a rack or two racks. This goes back to some of my experience in supercomputing in the past where we were putting optical Networks in to manage all that data and handle all the density that we're building inside the data center. Well, that means new technologies, pluggables, icd, new technologies for photonics inside the data center as well. And then the last one which is coming is going to be AI and mobile networks and that's going to drive new bandwidth requirements, new latency demands, and that will require innovation as well for our mobile networks. And that's where we see 5G advanced and ultimately 6G headed.

‍

Patrick Moorhead: Yeah, I'm glad you, you know, a lot of the discussion has been about inside of the data center and I'm glad you extended that first of all data center to data center, but also even between different countries, but also the industrial edge, which I believe we will see some tremendous growth there, at least in the next three years. It's hard to predict, but what I do know is that historically speaking the compute goes to where the data is generated and it's very efficient to be able to do that there as opposed to doing it all up in the cloud. But so let's spoil this down to where should companies be putting their money to get the connectivity AI really needs? You did break down the different areas, but are there any priority areas between all the way from the edge to the hyperscaler data center and everything in between?

‍

Justin Hotard: Well, I think they're all priorities, Pat. And I think you touched on an interesting one which is the industrial edge. And this is where we're through a lot of our partners. We're already seeing investment. One application is defense. Obviously there's some terrible conflicts and wars happening, but we're seeing that drive innovation in defense, autonomous AI, robotic driven drones that are providing defense capabilities. That means that connectivity now needs to be invested in, in the battlefield and that's driving innovation in terrestrial networks. And it's not just in defense. We're seeing it in public safety and we're also seeing it in certain industries where connectivity is really critical and it's driving autonomous or semi autonomous devices. And what's different about this is that if you think about connectivity in the past there's been some expectation of exceptions, right? We'd say five nine, six nines, you know, we might say three nines. In an enterprise, legacy enterprise environment, you can't have that now. You need something that's always connected, that's trusted, it's secure. And that means that companies need to upgrade those systems even if they have existing connectivity. And as I touched on earlier bandwidth latency, these are really important priorities.

‍

Patrick Moorhead: Yeah, I love the Edge. And you know, a lot of questions that I get is well, wait a second, wasn't the edge hot like seven or eight years ago? Industry 4.0. But we've done a lot of research on that, a lot of thinking and the biggest difference is that AI performance per watt at the edge is about 100 times where it was seven or eight years ago. So what you could do on the edge was actually limited. And in the seven to eight year period as well, the management of that data, the management of those applications has become a lot more seamless than it was. So we are bullish on this, let's call IT Industry 5.0. That's terrible. But hey, I'd like to shift. You know, Nokia has been the backbone for multiple telecoms for, for decades. You've expanded the business to enterprise. But I did want to auger in on, on telcos here. How is AI going to impact the telco networks and what does it mean to your customers?

‍

Justin Hotard: Yeah, I mean, I think first of all, you know, look at the announcements that are just, they've come out recently. You know, Google announced a new set of smart glasses, right? You've got rumors that Apple's going to follow with something much sleeker that's going to create different bandwidth. If you think about AR and VR or even just the current gen AI activity that's happening, the uplink becomes much more critical because I actually need to process the data and the video and the uplink as well as the downlink. So that means bandwidth changes, that means upgrades and enhancements building on the 5G investments that many of our customers have made in mobile. And what's interesting about us is we're one of only two major players in the western world that deliver the complete solution in this space. But it also means modernization to how they operate these networks. Because if you think about what availability and trust, meaning it also means enhancements in operations, we need to move to more autonomous operations, we need to have secure pipes. I mean if I'm running a public safety application, for example, I need things like network slicing because I need a secure slice so that the video I might get as a first responder can be connected in real time back and you can start to see how these applications cascade. And so it is going to drive a wave of investment. 

‍

So if you think also about the bandwidth demand, these new services are going to create more bandwidth than backhaul. So that's driving investments in optical network upgrades. And one of the things, one of the things you're Seeing in the US in particular, which I think is always an early adopter in this space, is a lot of investment in broadband and you're seeing obviously consolidation. There's been a lot of news in that, whether it's Verizon and Frontier, charter and Cox AT&T buying some of Lumen's assets, all of those point towards a trend of enhanced broadband investment. Well, that broadband investment also creates new opportunities for services and applications because now that broadband network is getting extended and we're seeing higher and higher bandwidths for fiber to the edge, which creates, you know, and obviously signals the demand that we're going to see for these applications and services.

‍

Patrick Moorhead: At Mobile World Congress this year, I was actually, I was very optimistic about some of these, these use cases and I know some of the advanced services maybe took a little bit longer than people had expected, but seeing real world companies doing it, I think the best example are the first responders where they do need that incremental slice. But I saw that taken across the industrial edge with robotics, manufacturing and even overall healthcare. So it was really exciting. That was one of my key takeaways from Mobile World. Mobile World Congress.

‍

Justin Hotard: Yeah, I was going to say I'll jump in there on a couple things. I mean, you know, we're seeing it with ports. So logistics is a great example where you think about the complexity of moving all these cargo containers around and starting to automate some of that, providing additional intelligence, you know, allowing remote control, remote management of some of these capabilities. You know, we're seeing that in that space. You touched on hospitals. I think there's early trends around, you know, about network slicing for hospitals and healthcare potentially as an alternative. And then of course you get out into any environment where it's indoor, outdoor, where you've got a field service, where WI fi just doesn't really fit and 5G, 5G advanced, some of these services are really important. And I think the shift for us in a telco industry, if you think about in the past, we've missed a lot of these waves, Pat. We were putting in 3G when the Internet was happening. We were trying to build a better voice network. Then we were chasing it in 4G. And I think what's interesting with AI and what we're seeing with the early signals in 5G advanced, but what's coming in 6G is now we have an opportunity to jump ahead, anticipate where the technology is going and build AI ready networks to be able to support the applications and services that will come from AR and VR robotics, autonomous solutions and others.

‍

Patrick Moorhead: What does AI ready data centers mean to you? You just use that term. I mean the way that we segment the AI ready data center. First of all, AI ready is typically GPU or an accelerator. It could be doing training, but it 100% is going to be doing inference. And it's spread across hyperscaler, tier 2 CSP, telco, data center, sovereign cloud, and. And even the edge where you have raised tile flooring.

‍

Justin Hotard: Yeah, I think it's very similar. I think what it means to me is we're building networks in a different way. And the way I think about it really simply. And we saw this again, going back to my supercomputing roots, we saw this in supercomputing, right? The cloud was all about virtualization. How do I put as many workloads onto one computer? And in AI it tends to be the opposite. How do I leverage the compute resources across one workload? And so it's really a complete inversion of what we thought about in the cloud era. And that drives different networking demands, different security protocols. And so I see us supporting a lot of those early investments across all the markets that you described. Because the customers that are at the front end of this recognize that it's great to have all this compute infrastructure, but if you don't have connectivity that actually performs, you're not actually going to deliver the applications and services with the performance, the latency, the security that customers need to drive the kind of scale adoption that the market's anticipating.

‍

Patrick Moorhead: It's interesting, Justin, if enterprises of all kinds, including telcos, they become AI ready, they're also cloud ready. Right? If you look at the growth of the hybrid multi cloud, you had talked about separating the computer from the data. There's that. But also the reality is that most companies have three or four hyperscaler contracts with different companies and they're really trying to get better at organizing that data and the applications across all of that, plus their legacy infrastructure. And this is where networking counts as well. I do see a downstream improvement. If people improve their networks for AI, they're going to get a bonus. We call it data center modernization. That's the moniker that we're using. So in other words, they get a twofer. How do you view that? Does that fall into what you and Nokia are looking at?

‍

Justin Hotard: Yeah, I think it does, Pat. And I think there's. You touched on something. There's been this tradition of, of data being moved to the compute. The reality is that we're going through a cycle because of the pace the scale and the pace of demand for compute, where power availability, density is going to mandate compute has to be in a certain location. That's driving a whole modernization cycle on how we bring the data to that compute again. Right now that means connectivity. But to your point, over time we could see those workloads bifurcate back into the enterprise or the edge. And the beauty of what we're saying now is with the investments in technology that we see again with optical technology, the enhancements to front end routing, the next generations of 400 gig, 800 gig moving to 1.6 terabit connections, that's going to create flexibility so that your network is future ready and irrespective of how the, you know, how the technology evolves.

‍

Patrick Moorhead: So it's been a great conversation so far. Thank you, Appreciate that. Our viewers thank you too. But I do, I do want to wrap here, Justin, with what I consider two areas that are just fundamental to AI and that's security and reliability. You know, we've seen this historically, I don't know, the last 50 years in it, which is the more things you can do and even the more disaggregated things are, the higher the risk, higher the security risks and the reliability could take a hit as well. How are you making sure that networks stay secure and reliable in this new age of AI?

‍

Justin Hotard: Yeah, this is one of our core vectors of investment. Because if you think about Nokia, what's unique about us is we're the only company in the western world that plays across mobile, fixed infrastructure and optical. And the reality with AI, and you said it brilliantly, Pat, every new application or new technology innovation is a new attack surface for bad actors. And let's just take the example of smart glasses. Now I've got something that's delivering AR&VR. I need to not just think about threats in the core application, but I need to think about threats in intercepting the actual device itself and potentially delivering poor information to that first responder. Or it could be an autonomous drone. So there's a whole new surface of potential attack points and what we're seeing is the end to end view and the cross network view. To be able to look at all of that and say, look, it's not just that we're having a denial of service, but how do we ensure that the model and the inferencing engine that's delivering the insight is trusted, is authenticated, is the one that we expect. 

‍

And so we have to think about all of that in our network design and our teams are doing a lot of work around that to make sure that it's all trusted, it's verified and, and of course I think for the telecom industry, which sometimes has fallen behind in it, we have an opportunity to really leap ahead with more extensibility. It's why we invested in a small acquisition and we came out with an open source strategy around APIs in our network operations stack because we see the value of providing that extensibility through core networks for provisioning for new services. And I think this is going to be a place where it's going to take the entire technology ecosystem and new partnerships to be able to defend and anticipate and protect against some of these new challenges. So it's an area we're very focused on and also one that I think has a lot of potential for new innovation vis a vis what we've seen in the past in telco networks.

‍

Patrick Moorhead: I appreciate you turning up the contrast ratio on that. I don't think enough people are familiar with the only capabilities that Nokia provides. And by the way, as a side note, I've been, you know, watching from afar some of the things you've been doing inside of the company, expanding the footprint, you know, making some tuck in acquisitions and it's pretty exciting. It's a long way from hpc, but maybe not.

‍

Justin Hotard: Well, it's an exciting time to be in connectivity because as we used to say in hpc, compute without connectivity is pretty lonely. So we're excited to help bridge this new era of AI and support all of the great innovation happening from model developers to GPU and accelerator companies. And I think it's an exciting time for Nokia and so been a great first few months for me and look forward to continuing the conversation as we continue to innovate and enable this future.

‍

Patrick Moorhead: Yeah, I agree. Thanks again. Thanks for tuning into the Intelligent Edge Spotlight here at the Six Five Summit in its sixth year. Be sure to check out more sessions in this track as we explore the tech and the reality of IT and enabling AI in the edge. From telco to data center to enterprise use cases and everything in between. You can see the full agenda at sixfivemedia.com/summit and we will be back with more insights and conversations shortly.

‍

‍Disclaimer: Six Five Summit: Intelligent Edge Spotlight is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.