MWC Barcelona 2026 | Chris Lewis, Lewis Insight

Clips
More from MWC Barcelona 2026

Chris Lewis

Managing Director

Lewis Insight

play_circle_outline Uplink Surge and On‑Device AI: How Small Edge Nodes Challenge Data Center Traffic Dominance

play_circle_outline Outside-In Era: Hyperscalers Build the Backbone, Telcos Remain Essential Access Providers

play_circle_outline Point of control moving toward edge; sovereignty and jurisdictional issues.

play_circle_outline Control plane and meshed radios vital for managing distributed edge nodes.

play_circle_outline Agent marketplaces and app monetization create new telco revenue opportunities.

play_circle_outline Ubiquitous reliable connectivity is foundational; telcos must "nail" connectivity investment.

play_circle_outline Smaller telcos face consolidation, capital challenges, and greater government expectations.

play_circle_outline AI becoming central to telco strategy and culture shift.

Info
Transcript

Chris Lewis, Lewis Insight

Chris Lewis

Managing Director Lewis Insight

In this interview from MWC Barcelona 2026, Chris Lewis, managing director of Lewis Insight, joins theCUBE’s John Furrier and Dave Vellante to discuss the telecom industry's cultural and architectural shift in the age of AI. Lewis breaks down how the explosion of AI is pushing traffic and compute from centralized data centers to the edge, fundamentally changing network demands from a downlink model to an uplink-heavy, outside-in architecture. He highlights the critical question of where the point of control will sit in this new decentralized landscape, emphasi... Read more

explore Keep Exploring

How is AI and the rise of edge computing affecting network traffic patterns, capacity needs, and the role of telecommunications operators (including shifts between downlink and uplink)? add

What is the future role of telecommunications companies, and how will the rise of hyperscalers and AI at the edge reshape network investment and telco business models? add

Where is the point of control in the new network/edge/cloud infrastructure — will it remain with telcos, shift to edge devices, be controlled by applications, or reside with cloud/sovereign cloud providers? add

What will the control plane for managing edge computing look like, and who will control it — end users, cloud providers/hyperscalers, or new 'neocloud' operators? add

What impact will agent marketplaces (like those on the Apple/Google app stores) have on telecommunications operators, and how should telcos prepare to support and monetize AI agents given connectivity, interoperability, and infrastructure challenges? add

What should telecom operators prioritize to support emerging digital services (such as healthcare) and capture revenue opportunities? add

Can smaller telecommunications operators implement the measures and participate in the initiatives you were just discussing? add

What is the biggest challenge in implementing AI-driven (or overall) intelligence and new architecture in telecommunications organizations, and how do you see that unfolding? add

bolt Powered by CUBE AI

Chris Lewis, Lewis Insight

search

John Furrier

>> Welcome back, everyone, to theCUBE. I'm John Furrier, host of theCUBE with Dave Vellante, my cohost who's back from the floor assignment, going out and getting all the data. It's our three days of live coverage of MWC 26, formerly Mobile World Congress. I guess they go with MWC. Back here on theCUBE to break down the action. Chris Lewis, managing director of Lewis Insights, analyst roundtable with Dave Vellante and myself. Chris, great to see you again. Thanks for coming back on. Thank you.

Chris Lewis

>> Lovely to be back. Every year we come here, we try and digest what's going on, try and break it down, but it's always great to join you on theCUBE.

John Furrier

>> We're going to break it down. We're going to smash it down.

Chris Lewis

>> .

John Furrier

>> We're going to smash the coverage.

Chris Lewis

>> We don't want to break too much.

John Furrier

>> Unpack, break it down, smash it. Let's smash the news and some of the highlights. Just give us your take. Obviously, it's an AI show now. Obviously, the culture shift. The telcos can't ignore the distributed computing wave coming at them. Cloud native's gone hybrid. You got edge looking like it may hyper-converge. Are the telcos culturally ready to think differently around how they run their business? Will they blow it this year in this wave, because they have an opportunity, they got the data, they got the network. They can pull it all together with maybe intelligent node at the edge. What's your thoughts on-

Chris Lewis

>> You're going to start me on a simple question just to warm me up with you?

John Furrier

>> Oh, let's go. What'd you have for breakfast today? Let's start with that one. Did you have an omelet?

Chris Lewis

>> Oh, the omelet was great. So, what is fascinating is that with all the moving parts of the, let's call it the AI world, the reliability, almost the conservatism of the telecom industry in some ways plays to its advantage, because connecting things... Let's break it down to the way things connect. Between data centers, that growth has been massive. And of course, luckily we've got the optics to do that. So, the capacity is there. We're pumping more capacity in. At the edge, and the edge is finally, I mean, after what, 10 years of talking about the edge. The edge is beginning. There ere some announcements this week, Telefónica here in Spain with Nokia starting to build some little edge nodes around the country to help deliver. But what's interesting is that the real edge, out to you and I as users or individuals in the business, so on, AI's not really impacted that traffic yet. It's about 1% of access traffic. So, we've not seen it go right out to the edge. And what we're seeing at this show with the announcements from people like Qualcomm in those edge devices, in devices which will have their own models running in the near-term, that may well change it. And if you talk to Ericsson, they're talking about the uplink becoming much more important. So, we're moving away from a predominantly downlink model to an uplink model. So, I think we have to be very careful not to confuse the explosion of traffic going on in data centers, between data centers and what's actually happening out on the access network, because of course, on the access network is where the telcos actually make most of their money. So, they're selling to you and I, they're selling to businesses, they're selling into the ecosystem and partnerships. So, that's not changing. The demand will change in traffic flows in the future. What we connect at the edge will change. And interestingly, during the judging of some of the GLOMOs this year, I came across the first robo guide dog, which obviously I'm interested in that for future years, whether that will help me get around the show or not.

John Furrier

>> I hope it doesn't hallucinate.

Chris Lewis

>> But in some ways, as I say, I think it's really important to step back and say, "The role of the telco is not in question." We need telcos to deliver both access. The backbone networks have actually changed significantly in the last 20 years because the hyperscalers... If you look at who's investing in the massive submarine, massive global network, it's not just telcos. The Googles and the Amazons and the Metas of this world are also investing in it, so that model has changed significantly. Luckily, the capacity increases all the time with technology, with all the optics, as I said, and photonics. So, the edge is where I think the next wave will happen. And that's when we finally change the business model, because at the moment we've been building things where we look at it from the inside, we create nice new technology, and eventually, we expose it to the customer. What's going to happen in the next wave when AI goes out to the edge and there's those small models in the edge is that finally it'll be an outside-in model where what's been demanded from the outside will shape the way in which traffic will flow and will demand better service. So, telcos haven't had to react in many ways.

John Furrier

>> Yeah. I mean, that's a forcing function on the architecture. So, the traffic patterns is interesting because if you look at the sovereign trend, if you apply sovereign edge, the radio frequency is the boundary, okay? You have that, that's going to be an inbound opportunity for AI to say intelligently, "Where's the sovereignty? How do I shape that RF, maybe with quantum?" That's a big discussion we've been hearing. So, a lot of like futuristic markers have been laid down this MWC.

Chris Lewis

>> It doesn't change the fundamentals, does it? The fundamentals are we need to be connected, all of the sensors, all the things need to be connected and that, whether it's radio frequency... And by the way, we always need to put it in context that, yes, we use mobile for a lot, but we still use fixed connection for a hell of a lot and Wi-Fi extending from that. And whether it's 80% still connected through some form of fixed connection. So, yeah, sovereignty matters. I tell you what is really important, which nobody talks about, is where is the point of control in this new infrastructure, in this new architecture? Because traditionally, it's been controlled by the telcos. They've said, "This is the way we're going to move traffic around." And that point of control is moving ever closer towards the edge. And will it be ultimately in our devices at the edge? Which of course will say, "Yeah, we're in Spain, so it should be part of a sovereign activity, therefore don't take it anywhere else." And it's not just telco controlling that point of control. Is it the applications? Obviously, it's the cloud, sovereign cloud people. In many ways, we often dump it on the telco saying, "You're not doing enough," but actually, the telcos, a lot of the investment these days seems to be much more around consolidating around being the best possible infrastructure within country or within regions.

John Furrier

>> It's interesting because all the AI stuff around AI factories, you see the centralization, you see all the density around these, like say NVIDIA racks, but you think about telco it's totally disaggregated, spread out. These nodes are connected by the radio. So, there's a vision of, "Hey, if I can connect these disaggregated nodes and have them talk to each other, almost it's an asynchronous factory." So, the meshing of the radios could usher in a factory intelligent node to manage that control plane. So, that's the question you just brought up, what is the control plane for managing edge?

Chris Lewis

>> Exactly. Well, managing everything, right?

John Furrier

>> Yeah, managing everything.

Chris Lewis

>> Does it shift from the inside out to the outside in and managing who has that control? Do we, as users, have more control over it or do the cloud providers, the hyperscalers or do the neocloud... And obviously, as a business, you choose your partners, right? You choose who you think is going to help you run it. And of course, the complexity of the way everything being connected and as we push towards 6G, all the sensors that we might have in place, all of those things, there's got to be better control and better understanding.

John Furrier

>> Dave and I were riffing yesterday with some guests and we said, "Hey, Apple and Google took the monetization of the cloud wave with SaaS, the Apple Store and the marketplaces." If you think about agents, all these agent marketplaces, if I buy an agent in the Apple Store, it's got to run on the telco. So, if I'm a telco, I'm sitting there, I better start getting my act together because I'm going to monetize that. So, this is not even on the radar today. Today's like, "Oh, yeah. Get up to subscriber base. Here's a good plan for you." But ultimately, agents are going to be running wild. That's going to have to be monetized through-

Chris Lewis

>> Yeah. I think the swarm of agencies are real concerned because obviously consistency and the way they interact with each other, MCP and the like is fundamentally important. But one of the issues is that connectivity, fixed and mobile, is still often very invisible to the endpoint, to the customer, because you never often know what you're on whether you're on Wi-Fi or fixed connection or whatever. So, I think there are revenue opportunities. I think the issue today is that telcos have got to invest, and they are investing, to make sure their systems and their processes work most efficiently because we rely on those guys. They're not going to go away. We need that connectivity to be in place. In fact, we need it even better than it has been. Time and time again, I come back to the way in which mobile especially has been deployed in Scandinavia, for example, where you can get a connection when you're out on a mountain. You're skiing out on a mountain, you're still connected. And in London, you leave London, you lose a connection on the motorway going out of London, and the same is true in the US. So, if we're going to truly support all of these business cases, or society as a whole, let's say, we've got to have infrastructure investment from a connectivity point of view, which connects everybody and everything. And yesterday, I was talking in the healthcare event up in Hall 8 and I said, "I think there's a massive opportunity in terms of delivering not just the healthcare, but the patient experience, but there's got to be reliable, ubiquitous connectivity." So, I think if telcos focus on AI too much in the broader sense of AI, they're going to move the folks away from building that underlying essential infrastructure, which we need.

John Furrier

>> You got to nail the connectivity. They got to nail the connectivity.

Dave Vellante

>> We asked an interesting question in the Cisco roundtable that we were just at with Chuck Robbins, Jeetu and Oliver. You asked about the smaller telcos.

Chris Lewis

>> Yeah, I did. Yeah.

Dave Vellante

>> Because you said, "Okay, big guys, they can afford all the CapEx, the smaller guys can't." So, you asked Cisco, "Are you helping them finance that?" And they said, "Not really." And then, "Are you seeing any growth?" And they also said, "Kind of, not really." And so, what is the role for those smaller telcos? Can they fill in some of those discussion points that you guys were just talking about with the smaller telcos?

Chris Lewis

>> Oh sure, of course they can. And of course, as analysts, we look at globally, we paint this big global picture, but actually, not the irony, but the situation today with all the global geo stuff going on, it still goes down to local conditions. So, in the telecom world and in the data center world now, you've got to have resource locally, you've got to dig the fiber, you've got to put the radio mass up. Now, one of the questions is how many of these players do you have in every country? Because the European model was we're going to encourage competition to drive prices down. So, we pay a third or a quarter of what you pay in the US for your mobile service. And in India, they pay two bucks a month rather than 60 bucks a month in the US. So, the economics just don't stack up. So, the consolidation of, especially the European market of smaller players, those players are going to get closer and closer to their governments because governments need it to be in place to support the economy and society. So, you don't need four players in a small country. So, I think we will see a significant consolidation of the market and with consolidation, they've got more market power. And of course, if the government give them that more power, less competition, they will be expected to invest more and more. So, the cost of capital is an issue, and I think someone quoted 75% of the smaller telcos can't afford the cost of capital. Well, we can't be without this stuff, right? We need it to support public services, to support business. So, governments have got to find a way in every country. So, yes, there's a European Commission level approach, but every country's going to do its damnedest. And what of course is fascinating is that you look at the Middle East with all of its investments in AI and in connectivity, they could become an AI hub for the region. Obviously, we'll see how the whole geopolitical thing pans out, but their energy and stuff is like, because we don't have enough energy in most markets. So, connectivity, we need to keep building, investing for the long-term. And energy, we need to up the energy in every country. We do need data centers everywhere. We need that local bit. But the other thing which I touched upon in the Cisco meeting was that as we move the models into the devices, into laptops and into smartphones, that changes the dynamic again. So, where does that smartphone reach to when you're in Spain? Does it go back to your home market back in the US or does it pick it up locally? So, the architectural changes, as we move to this outside-in model, is massively different.

John Furrier

>> Do you see the biggest challenge of connectivity, AI focus or architectural reset? Of those three things, you mentioned connectivity, that's mission-critical, you got to have connectivity. What's the balance between those three in connectivity, focus on intelligent AI edge, and then what you just mentioned around-

Chris Lewis

>> Let's be clear with the AI angle is that AI plays a massive role in improving the way traffic flows. We're reducing trouble issues. We're looking at new traffic flows. AI will learn, we'll remodel, of course, as it goes along. So, AI becomes part of everything. So, that's why, in some ways, I'm hoping if in a few of these time we don't even talk about AI anymore, we just talk about overall intelligence and how we build the systems. Architecture is a massive issue because-

John Furrier

>> What's the biggest challenge there? What do you see happening?

Chris Lewis

>> Well, the biggest challenge is people at the moment, because breaking down the silos and talking about how, within a telco organization, whether it's your big US ones, the Verizons, AT&Ts and the like, or whether it's a small one in Croatia or somewhere, they've built themselves up, focused on building the network and eventually they offer a service to the customer. The relationship between the engineering, the IT part of that, and then the lines of business who make the products and sell the products and the channel, it's a massive change. So, there needs to be a change in empowerment and the power of the people delivering the services. As an industry, we spend way too much time focusing on the inner workings and not enough time on what the product is. What Cisco used to call intent-based networking, which is way ahead of its time. So, what is the intent of being connected on your mobile device? It's your personal life, it's your business life, it's you being part of society. We need to-

John Furrier

>> It's interesting. You mentioned earlier connectivity, but now you have multi-access connectivity, seamless meshing, Wi-Fi, 4G, 5G, 6G and fiber. Now, you're going to have that and you mentioned diversity of locations. If you have virtualization, you can virtualize everything now, radios, user plane functions. I mean, all these things are now potentially virtualized. So, you have virtualization coming in, so you got some tech going there. And then, the question's going to be, how does the AI take advantage of that? Intelligence, how do you bring intelligence to these-

Chris Lewis

>> But John, there's two sides of it. So, how does AI take advantage of it? How does AI help that happen?

John Furrier

>> Yes.

Chris Lewis

>> So, AI is working behind the scenes to make sure that the right virtualization is in place, the right connectivity goes in place. And then, the AI that we use to do our daily jobs, that sits on top of it. And of course, that's why you need that ubiquitous reliable connectivity because when you send a request to whichever your preferred AI engine is, you want to get the response. I've showed in the past, I use a thing called Be My Eyes, which takes a picture from my camera or my Meta Glasses and describes to me what's around me, phenomenal. But of course for that, you need to be able to get to access to GPUs and whatever.

Dave Vellante

>> Yeah. I was saying to Chris, that's like the movie we saw with all the light you cannot see, when we were closing our eyes and it was describing everything in great detail. So, Chris, we saw in the internet wave where the over the top vendors won all the monetization, the telcos put forth all the CapEx and all the infrastructure and then the cost per bit kept coming down and down and down. And the guys like Netflix won, we saw a couple years ago, the battle between, I think it was the CEO of Orange calling out Netflix and they retorted, "Well, we're delivering so much value over the network," et cetera, et cetera. Okay. Now, when we enter this AI wave, we've got another opportunity here. Telcos, again, winning in connectivity. You've got the satellite folks, you've got the cloud vendors, you've got private 5G, private 6G. Who is going to win in this next era?

Chris Lewis

>> So, the Netflix example is a great one because Netflix have the end product. And bear in mind, they've massively disrupted the content world already, so they have the end product that people want to pay money for. And so, what I see whatever sits at the end of the value chain is where the money will be created. And telcos fundamentally aren't at the end of the value chain. They've never been great dealing with customers. In some ways, I think the wholesale market for telecoms is a much more interesting market. We'll leave the retail people to deal with the end customer, both business and consumer. So, I think the people providing the value, whether that's obviously social media guys have got massive value, because there's still an advertising model in there. So, the Metas, the Amazons, the Googles and Microsoft have a massive role to play. So, what I think is happening is that the AI democratizes many formerly separate silos of connectivity, compute, storage, security, and we may well start to see all of those elements become part of what we pay Netflix for because you need to back through the system to make sure it's delivered securely. So, in many ways, we spend far too much time as an industry, and the IT industry does as well, talking about the inner workings and how clever we are, that's what people expect from us. They don't want to understand how the radio network works, how the fixed network works, what's going on in the data center. They expect that film to be delivered, or that parcel to be delivered, or that communication between your mother and your kids to be delivered. So, I think it really, A, the economics are changing, of course, because as you say, the cost per bit comes down significantly. But B, the devices we use, and we'll see a lot more sensors coming through in the next 5, 10 years as we move towards 6G. So, it's a wraparound us. It becomes a service wraparound us as individuals, as a household, as business and as society. So, you've got to look at where the value is created. And the value isn't just created by the technology components. It's created by the combination of all of those components.

Dave Vellante

>> Okay. So, it's likely that there'll be new players emerging that take advantage up the stack, but how about the existing players, the app vendors, the cloud players? Do you see any of those as exceedingly well-positioned to win in this area?

Chris Lewis

>> Honestly, I mean, app providers, yes, as long as they stay absolutely relevant. And we've seen a lot of discussion about how they're going to be destroyed by automatic code-generation within AI. People pay for expertise, people pay for trust. And so, obviously, as more and more AI generates more and more product, that's likely to end up being very similar. So, having something which actually distinguishes you from the competition is going to be really important and the customer relationship becomes important. So, Salesforce's relationship with its customers, they will be using all the AI tools to try and improve what they deliver, as will anyone else delivering any value to a business or a consumer. So, yes, some new players will emerge, perhaps some aggregation players will emerge, but I think the pressure will come from the fact that the model is going from being one driven by technology to one driven by activity and consumption.

John Furrier

>> I want to get your thoughts on two things. We'll talk about 5G first. I want to get the space angle. First is the 5G. Obviously, it didn't deliver on the promise on ROI, but the technical specs were great, low latency, high speed, but it didn't get the carriers the monetization because there wasn't a lot of applications. I mean, I'm paying less now than I did for 4G. So, it's got higher performance. So, it's good for me.

Chris Lewis

>> For consumers, it's great.

John Furrier

>> Not good for the carriers. So, the question is, what applications are you hearing in the conversation or seeing or have an opinion on that could unlock more aperture for monetization with the leverage of 5G and 6G coming?

Chris Lewis

>> But I think it goes back to my previous point, that the way we're connected, whether we're on 5G, 6G or Wi-Fi or on a fixed connection, that becomes invisible to us. And because the channel of delivery coming out from whoever's delivering it, whether it's content or an application or a commercial activity, because it's invisible, the revenue doesn't accrue to the telco. So, in MWC here, with the keynotes yesterday and every year we come and talk about what was mentioned in the keynotes and it's nearly always a moaning about the lack of return on investment. So, I think we should rename it Moaning World Congress. But the issue is, "Guys, get on with it. You've got a fantastic business." It's changed-

John Furrier

>> BMC, Bitch & Moan Conference.

Chris Lewis

>> Yeah, but it's changing massively, right? So, yet you're having to radically change because the telco industry didn't change quickly enough, let's face it, whether that's to do with the employment issues, to do with country issues, to do with competition issues. So, the industry is changing. The workforce is shrinking. We know that. Let's face it in AI, we see major job removals from the engineering focus jobs like telco. But in terms of the product that we've got to sell, it works. And it's fit for purpose. In my good old software days back in Logica, when you're designing better software, you could design the best ever bit of software, but if it took you 10 years to do it, you'd miss your .

John Furrier

>> Well, I think that's going to be interesting to watch because I think if they get the foundation right for connectivity, get the architecture and culture, they can see how it plays out. So, I think we'll watch that very closely. I'm bullish. I think they're going to get this right.

Chris Lewis

>> I mean, it's not going to go away, John, is it? We need the connectivity. In my darker moments, I think we might remonopolize the telco market and have it delivered as a state service.

John Furrier

>> I worry about that. That's another level.

Chris Lewis

>> Because it was never efficient, was it?

Dave Vellante

>> We always made the point that the telcos have been great at connectivity. They're not going to lose that battle. Start from there and go from your strength.

Chris Lewis

>> And sorry, to go back to your satellite comment, so space and satellite, it's a relative niche filler-inner for things. Yes, it will deliver in some places, but it doesn't change it. But one thing I say yesterday, I spoke at a health-tech event yesterday. I think for industries, telecoms to deliver the underlying service which connects an industry like healthcare, and healthcare is the perfect test case, right? Because everybody, whether it's me, as a blind person, my elderly mother, whoever, you've got to deliver a service, which includes everybody. And that's all the way from your local GP through to surgery and that whole healthcare journey. So, we keep talking about this year-after-year that the telcos have got to... They're not going to take over the verticals, they're not going to take over healthcare, but they can certainly deliver a much more beneficial service. And I was talking yesterday to someone saying, "That one of the things we may well do in the future is that an elderly person may well not have the technology knowledge to set up a video link, but if you just send them a button on a screen, they can tap on it and that creates a one way, so the doctor can see the patient."

John Furrier

>> I mean, I'm bullish on 6G inside premises. I think the telcos have a huge opportunity. Again, I brought up the space comment because I just posted last night on LinkedIn. I had a couple beers in me, so I got a little salacious. Brian Modoff, friend, who's former financial analyst covering telco, he and I were chatting. He wrote on his LinkedIn, "Invasion from space for the mobile operators," and had a link to Starlink. Okay, so a little play on Elon. I wrote, "Great perspective. SpaceX, AWS are all waiting at the gates. If telco doesn't evolve, they will be eaten by the barbarians." And so, the premise there with the little over the top haymaker was if space evolves, I mean, Starlink is pretty damn good and it's going to get better, so that's always on. So, assume there's a space overlay that could put risk-

Chris Lewis

>> It can put pressure on them.

John Furrier

>> Is there a pressure point like, "Hey guys, this is the last time you got this last shot at monetizing with all that data"?

Chris Lewis

>> I don't think it's the last shot.

John Furrier

>> Okay. Yeah. Yeah.

Chris Lewis

>> So, if you look at what Starlink has done, it's sold out in many places. I know in London it's sold out, there's no more capacity because of course you've got spectrum limitations, you've got bandwidth limitations. It delivers a great service where you get it and if you haven't got fiber to the home already. You've got to remember that the longer-term investment of fiberization of cities and countries is rolling out. So, if you're going to connect to that, you more likely to use that, it's a much better mechanism for connecting. So, will space take over? I don't think so. People are talking about putting data centers in space. It's going to come up against physics at some point in terms of capacity, round trip and latency and so on. So, yes, we should consider it and it provides an alternative if it wakes the telcos up and makes them invest more quickly.

John Furrier

>> What's your thoughts on one scenario that's... It's a scenario. I won't say it's most likely, but you could look at the telcos, and you were saying this earlier about they're in a good position with the connectivity with going into healthcare and all these areas. They could literally run clouds in metro areas. And you could literally say, "Run a private cloud," or, "Hybrid multi-tenant for homeland security and start thinking of how to slice and dice generatively shape the traffic based upon real time situations, emergency..."

Chris Lewis

>> So, let me give you a great sovereignty example. So, there's a great healthcare example where, and let's take a European example. So, Europeans, if they use a certain hyperscaler for their data, the US demands access to that data. So, if you're going to be sovereignty, you're not going to put your data in their data center. So, for a healthcare environment, especially a National Healthcare Service, why wouldn't you, in major hospitals, put a data center in there, put a nuclear power plant in there to power it because your hospital needs it anyway. You may well end up with some extra power, which you can push back into the national grid. You've got a physical security, local connectivity, because you don't want to be shipping all of your massive scans all over the world. The doctors need it in that facility. So, actually, in some ways, all of the pieces that are moving around now may well move back to be providing more localized service. Geopolitics suggests we want to have security, reliability within country, certainly within the healthcare. Defense is a totally different topic where we're seeing a lot more activity, of course. But yeah, there's-

John Furrier

>> Sovereignty is real. I mean, it's a real thing.

Chris Lewis

>> It absolutely is. Yeah. Yeah. And of course, sovereignty is brilliant because A, the connectivity players are national by definition. Governments are interested in it because there's a political leverage in it to make them look better from a political point of view. And then, customers go, "Hang on a minute. Am I getting a better service by doing this?" Yeah, you feel more... There's always been this issue. I've said many times that when we talk about our telcos, we have the old two... Remember the two Venetian play masks? You have a love mask and a hate mask, right?

John Furrier

>> Yeah.

Chris Lewis

>> And we love them when we hate them. We love them because they're our local champions. We hate them because they've given us terrible service or they've ripped us off, but that's changing. So, they need more intimacy. Do they have to do everything? No, they have to facilitate all those other pieces.

John Furrier

>> Well, it also brings in obviously peer-to-peer, business-to-business.

Chris Lewis

>> Exactly.

John Furrier

>> So, now you have connectivity, not just one way, it's also marketplace-driven.

Dave Vellante

>> Okay. There's another side of that though that I want to get your opinion on because the cloud repatriots are now talking sovereignty, and I get it. And there's, clearly, to the points you mentioned, Chris, concern about privacy and national security. At the same time, they're the best tooling, it lives in the public clouds in the US. And so, what are you giving up for that sovereignty?

Chris Lewis

>> Well, do you know what? I mean, if you follow the AI logic, then those tools that are in the public cloud today, you can take AI and development locally for yourself, can't you? So, if you follow what Jeetu was just saying at Cisco, if 100% the code for those products is being written by AI, then you can downscale that quite easily locally.

Dave Vellante

>> You also heard what Chuck said. He goes, "Well, the guys who developed OpenClaw, they don't have any clue around customer experience." It's so difficult to actually do. Now, is your bet that that changes, that it's going to be dramatically simplified to deploy AI for the mainstream enterprises?

Chris Lewis

>> I believe so. I mean, it makes absolute sense that the hindrance to that development within telco in the past was obviously, as I said, culture and people and business organizational structure. A lot of that is changing as they slim down to be right sized and right purpose for the future. Then, from a software point of view, if we have this more open architecture, then yeah, the tools will come, right? Because if it's not developed by the hyperscalers, the NeoCloud guys will do it or to your point, some new players will come in and they'll use their AI knowledge to build things which are more... So, it's going to ever bring things closer to the customer.

Dave Vellante

>> So, how does Jassy win at this game? Because I think you talk to the cloud guys, and particularly Amazon, they feel like they have a design to sovereign. They can go in with VPCs and space zones-

John Furrier

>> Well, I mean, they're doing a lot of space stuff. They got a lot of space activity going on. They got satellites. They're looking at satellite, but they don't have AI. So, this is why I like NVIDIA's AI factory.

Dave Vellante

>> Who doesn't have AI?

John Furrier

>> Well, Amazon's on their comm side. They have AI and Bedrock in the cloud, but if you look at all what they're doing in telecom, they're building out good, significant-

Dave Vellante

>> But specifically as it relates to sovereign, they can put in local zones, they can put in-

Chris Lewis

>> Oh, sorry.

Dave Vellante

>> And that's their play.

Chris Lewis

>> That's a totally valid point.

Dave Vellante

>> Okay. So, are-

Chris Lewis

>> As long as you can ring-fence it. I mean, I don't know whether you ever use any of these e-bikes when you come to London, for example, one of the boroughs in London just decided they don't like Lime Bikes. So, you're cycling along an electric bike and you get into the borough next to me and they stop.

Dave Vellante

>> Really?

John Furrier

>> Like the dog zone.

Chris Lewis

>> Exactly. You can't come in there.

John Furrier

>> Electronic fence collar.

Dave Vellante

>> That's dangerous.

Chris Lewis

>> But you've got to demonstrate that It's secure and it's local.

Dave Vellante

>> And do you feel like governments will trust that, the US hyperscalers doing things like local zones and doing their own sovereignty play?

Chris Lewis

>> Well, I think it goes back to that scale question you asked earlier about the smaller telcos. Can they do it all themselves? No, they can't. They need help. And will they get it from hyperscalers? In many cases, yes. Let's face it, there's also the Chinese around. So, if you look at different parts of the world, you look in Africa, the Chinese players are very well-embedded in Africa helping. They help build roads and airports and Lord knows what.

Dave Vellante

>> Yeah, very much so.

Chris Lewis

>> So, yes, there are plenty of significantly-powerful players willing to go down and work on that local infrastructure.

Dave Vellante

>> What's the sentiment in the continent? Well, you're not in the continent, but what's the sentiment in Europe around US these days with the current administration and all the bravado and the we're going to take over Greenland and Canada's the 51st state. What's the sentiment?

Chris Lewis

>> Well, I mean, if we treat it as entertainment and we think it's not real, obviously, we're going to miss a trick. But it's very difficult to have a rational discussion about it because the position declared one day then changes the next day and then... So, the UK prime minister just stood up to Trump for the recent invasions and recent bombings and said, "No, you can't use UK facilities." But now, that the Iranians are bombing the borrowing UK-

John Furrier

>> They said, "Go ahead."

Chris Lewis

>> Well, because he's a lawyer, right? He's a lawyer and he said, "I don't feel comfortable from a legal point of view." So, let's face it, we have to work... The US is a massive economic power. I think what Trump's doing potentially reduces that power. I know he thinks he's making the US stronger. I think in some ways it's going to lose friends.

Dave Vellante

>> Well, it comes down to innovation. So, right now the mainspring of innovation coming out of the four foundation model companies, OpenAI, Google, Anthropic and xAI, are driving a lot of that AI innovation. Obviously, you have China with open source models. What's your prediction in terms of how that plays out? I mean, how vital are those four US entities? Can the rest of the world's whatever straight ?

Chris Lewis

>> The bigger question is do we need four, right? Do we need four going forward?

Dave Vellante

>> Well, let's say it goes down to... Well, I actually think... Let's say it goes down to three, maybe-

Chris Lewis

>> It could easily go down to three. But the point is that the power available in other countries or the intelligence, let's say the artificial and the real intelligence available in other countries is significant. AI you might develop in China or you might develop in Europe. We'll learn from those models because you can ingest all that information. But ultimately, it comes down to the fundamentals of physics. Have you got the physical infrastructure to do that construct, the ingestion? Have you got the energy, the power to drive it, to keep it driving? Does it create value? What value does it create? And those large language models have got a big issue. I said Google's come back from way behind to be very relevant, but OpenAI I think has some challenges. How does it make money? It's raised another, what, $200 billion recently? There's got to be some product at the end of that value chain.

Dave Vellante

>> Well, I think that they have to move to the enterprise .

Chris Lewis

>> It's too much free. No, my point being, of all the users of the AI services, what percentage are paying for a service? It's small, right?

Dave Vellante

>> Right. No, their consumer business model is not sustainable. They have to get into the enterprise and they have to dominate up the stack and actually become the... At my premises, when we talk about how difficult it is to deploy AI, the LLM vendors are going to have to make it simple and they're going to get paid for doing that. And you're going to pay them through APIs to get tokens that are going to manufacture intelligence. That's their game, at massive volume.

Chris Lewis

>> Or you do small language models, which take a lot of the learning and do it locally.

Dave Vellante

>> Yeah. I don't know. I think theCUBE power law suggested that we're going to have those small language models. The more I think about it, the more I think that the large language models are going to either subsume them-

Chris Lewis

>> Oh, well, they'll try and ....

Dave Vellante

>> or they're going to become the small language model because they're just better at it.

Chris Lewis

>> But it then goes to your point about working with customers. So, if you go and talk to Orange about their Orange Live that they've deployed, the layer they've deployed takes a bunch of different models, including I'm pretty sure AI, Google and Anthropic and Mistrial as well, and actually delivers a layer. And they've trained all of their employees. They're now beginning to offer it as a service to their business customers. So, they see themselves as a trusted party in this. So, they're taking the raw power that comes in and so they become a channel to market for... So, there are some opportunities, by the way, which I think is a great angle for it.

Dave Vellante

>> Well, the question is, John, are those small language model developers the next rack space competing with the cloud, or do they actually have a viable path themselves?

John Furrier

>> I think the large language models will not subsume domain proprietary models. Small models don't have to be public information. I think the large language models mop up all the public domain stuff at scale, but have gaps. And I think specialty models will play well in telecom because edge... I see the hyperconverged edge as an existential threat if people don't keep the dumb pipe strategy going because I see models moving around to compute at the edge, and that's going to be control plane. We talked about fabric. You mentioned control plane earlier. I think once that solves, the models can move around. So, for instance, I could have a small language model for myself on what my shopping habits are, what my preferences are, and I could have it on demand. I mean, I already see people skiing with a drone that flies over their head and films them while they're skiing. Why can't I have a model virtually following me around the network? Well, I could opt in for that and say, "Hey..." So, that has nothing to do with OpenAI or Gemini. But now my model can tap Gemini and OpenAI. I can have an agent go do stuff around that.

Dave Vellante

>> I agree with that scenario.

John Furrier

>> So, again, you're assuming that the small language models are in competition with the large language models.

Chris Lewis

>> They're not competing with them. They're .

John Furrier

>> They're differentiated.

Dave Vellante

>> And I'm assuming the reverse, that the large language models are going to try to gobble up every piece of real estate that they possibly can.

Chris Lewis

>> They may try.

John Furrier

>> This is going to be a security challenge because most people reject OpenAI, from what I'm hearing in the hallways here in the telecom side, because they don't want to put that data in their cloud because they don't trust them.

Chris Lewis

>> But John, can I just demystify something you said? So, a dumb pipe, as you describe it, is not a dumb pipe. It's an incredibly intelligent pipe, which is adapting to what's going on around it. The telcos have failed to move up the stack for all sorts of reasons we could go into for hours and hours and hours. But the fact is, if you build that connectivity model, which adapts around whatever the scenario is above it, they're still going to make money. We're still going to pay them for that connectivity. We may not pay them in increments and growth multiples, but it's still there. It's a very good, reliable... It's a utility business. Your financial clients have looked at telcos as a utility stock for the last 25, 30 years.

John Furrier

>> Well, when I say dumb pipe, I mean the old adages of, it's just networking stupid, just move packet from point A to point B. But in a disaggregated edge in a hyper-converge, there's too many devices to manage, so the AI has to manage putting packets... It's not static rule-based. You can't say, "Oh, this packet needs to go here. That's today's networking." I consider that a dumb pipe. However, I think generatively, to your traffic shaping comment earlier, I think it's going to get smart enough to know that this packet could go anywhere. Where does it go? Decision is made in real time. That's different.

Chris Lewis

>> And that's my thing about the point of control. Where's the point of control? Who's making that decision? And increasingly, the agents will make that decision, obviously as part of a broader set of agents working within-

John Furrier

>> And the dumb pipe strategy, I would argue that the telcos operators have been operating under that mindset of selling connectivity. They talk about subscribers, revenue per circuit, but that's changing and when you get-

Chris Lewis

>> It has changed.

John Furrier

>> It's changed. Yeah.

Chris Lewis

>> Yeah. So, we used to pay end-to-end, don't we? Now we pay to be connected to broadband... For your good old internet, we called it onramping. How do you get onto it, right? So, we pay for that. And will we pay more for that? If it's more appropriate to do, we will. We want fatter pipes because we want to do more video. We want lower latency if we're doing more computer vision type stuff. Yeah, we'll pay for it.

John Furrier

>> Well, now they're going to go to the dumb AI strategy.

Chris Lewis

>> You know, that's not such a bad sound, is that, right? Because AI is so-

John Furrier

>> Dumb pipes go to dumb AI.

Chris Lewis

>> It's so much part of everything. It's just doing what it's told to do, right? So, agents do what they're told to do. Yes, they learn, but they're basically-

John Furrier

>> Chris, it's a great pleasure to have you on. And we got a break for lunch now, but I want to say always awesome to chat with you. You're a veteran, seasoned vet, you've seen it all, you've seen the movie many times here in the industry and always fun-

Chris Lewis

>> It's always fun putting it in context, isn't it?...

John Furrier

>> to smash the content here, breaking it down, breaking it up. Dave, as always, great to have you.

Chris Lewis

>> Exactly. We've got to break it down and roll it back up again. Yeah.

John Furrier

>> Yeah. We put it all back together. Broke it down, put it back together. All right. I'm John Furrier with Dave Vellante. Chris Lewis here, breaking down the analyst perspective, MWC. Again, structural change is coming. People are adopting. We think it's going to be a good tailwind for the industry. Stay tuned for more coverage after this short break.