MWC 2025 Highlights & Insights (with Peter Jarich and Chetan Sharma)

Peter Jarich, Head of GSMA Intelligence at GSMA Intelligence, and Chetan Sharma of Chetan Sharma Consulting joins Leonard Lee of neXt Curve to recap the mobile industry event of the year, Mobile World Congress 2025.

This year, MWC drew over 109,000 attendees and almost 3,000 exhibitors who packed the Fira Gran Via for another post-pandemic year to match an attendance record set in 2019. So, what mattered at the otherwise overwhelming event?

Peter, Chetan, and Leonard touch on the following topics: 

• Key impressions from the event. What was different from last year?
• What were the key themes for MWC 2025?
• What technologies and applications stood out at MWC 2025?
• Peter, Chetan, and Leonard give their takes on:
  • Metaverse, 
  • Generative AI & Agentic AI
  • AI-RAN
  • Open Gateway & Aduna
  • 5G & 6G
  • Private 5G
  • Open RAN and Squid Game Season 1
  • Geopolitics of global telco
• Where does the trio think that we will land at MWC 2026?

Remember to like, subscribe to the neXt Curve reThink YouTube channel. Also, follow neXt Curve by subscribing at www.next-curve.com.

Check out Chetan Sharma, his research and thought leadership at www.chetansharmaconsulting.com. 

Check out Peter Jarich and GSMA Intelligence at www.gsmaintelligence.com.

Transcript

0:06

Next curve.

Leonard Lee: 0:09

Hey everybody. Welcome to this next curve rethink podcast episode, where we break down the latest tech and industry events and happenings into the insights that matter. And I'm Leonard Lee, executive analyst at next curve. And in this episode, we will be talking about the highlights and the hot takes from MWC 2025, which took place last week in Barcelona or Barcelona. And that I think that's correct, right? Barcelona, Spain, with over 109, 000 people. In attendance over four days of telco slash mobile industry madness. And I'm joined by two very, very, very, very special guests, the remarkable Peter George of GSMA intelligence and the legendary Chetan Sharma of Chetan Sharma consulting gentlemen. How's it going? I'm

Peter Jarich: 1:12

just trying to figure out, I think two and a half of those varies are for Chetan, and I get a half of that variable. I'll

Leonard Lee: 1:16

take my

Peter Jarich: 1:18

half,

Leonard Lee: 1:18

that's fine. I had to wake him up because he looks like he's about to fall asleep, so. It's early, it's early on your coast, folks. Yeah, it is. It is. And, but, you know, Chetan works hard. He gets up at, you know, like one of those guys that wakes up at 4 a. m. in the morning, right? Chetan. Come on. Oh, shit. Yes. Yeah. Yeah. So before we get started, please remember to like share and react and comment on this episode. Also subscribe here on YouTube on Buzzsprout. Listen to us on your favorite podcast platform. Opinions and statements by my guests are entirely their own and don't reflect mine or those of next curve. We're doing this to provide an open forum for discussion and debate. We hope you find the program informative. And fun. And with that, gentlemen, let's dive into this because there's a lot to cover. I'm really looking forward to your takes and your insights from this year's event, which, by the way, I thought was fabulous. why don't we start off with this, your impression in one word. Or maybe a pair of words, how would you describe this year's event?

Chetan Sharma: 2:29

You go, Peter.

Peter Jarich: 2:30

I'm going to say frank and honest. and what I mean is, you know, Frank, I'll just say, if I said Frank, people might think who's Frank, who is this Frank guy? Peter's talking about. so that's why I said, I mean, it's interesting because. If you looked at some of the comments right off the keynotes, right? I think some of them were like, Oh, it feels like groundhog day. We've been saying the same things over and over and over, especially around regulation, especially in the state of the industry. But the tone was different, right? The tone was much more of like, it was, it was much more direct. And I think no matter where you fit on the political spectrum, I think that we know that the U S administration in some ways has sort of enabled a more direct Frank. Sort of less polite tenor of conversation. And I think you saw that reflected, right? You saw that reflected in a lot of the conversations of, okay, folks, this is the way it is. Or even the conversations around geopolitics weren't so much like one side says, Oh, China's more interested in domestic issues. And there's no real issue with the, with other, the U. S. and the response not being, well, I think we have a different view. Another response was just very clear. No, you're wrong. And here's why. Right? So I think it is. It was refreshing in that, in some ways, in that there were a lot of conversations that were really important conversations, and they were still polite and genuine, but also a little bit more direct than I say sometimes.

Leonard Lee: 3:53

Interesting. Chetan?

Chetan Sharma: 3:55

Not to just add to what Peter said, but I think it was much more. pragmatic, more grounded in reality on a number of fronts. Geopolitics is one of them, but 6G, for example, is another one. AI was all over the place. You gotta have at least one thing that is on the hype scale. So AI was that thing this year. but I always love going to the show because it just gives you, you know, in four days, a good grounding of what's to come, and get a sense of, where the industry is at, and you can just, eek out a lot of pearls of wisdom from just talking to friends and colleagues that you cannot do any other way.

Leonard Lee: 4:30

Yeah, these are great ways of describing the event. This year. I agree with both of you. I guess I will add present. and I say that versus the future, because last year it was future 1st and I really I didn't I thought that was not a good focus or a good theme for the event. I was just glad to see that there was a lot of focus on the present when you talk to, a lot of the companies again, colleagues and friends. In the industry, there's this desire to get back to what is it that we need to do to move the ball. And I thought that was very comforting. I think it's a really good direction for GSMA as a whole. And so, hey, big thumbs up. Proof that I'm not like a Debbie Downer all the time. I just want the industry, like all of us wanted to go in a better direction. And I think GSMA this year and all the, partner members, they did the right thing. And you're absolutely right and you know that they had to anticipate this before. A lot of the geopolitical. turmoil, was stirred up. So, that was a good call. So, yeah, really great stuff. So, gentlemen, what were some of the themes that stuck out for you this year as you were roaming the halls and, you know, giving interviews or moderating panels? And Peter, what else do you do? You do a lot of this stuff, right? You're managing a friggin team of like, massive

Peter Jarich: 5:54

I go where I'm told. no, it's I had this conversation with some folks saying, in theory, the way I set up meetings and do what I do should be different now that I run a team versus when I was in But I always, I do the exact same thing I did probably 20 years ago. I schedule a bunch of meetings. Don't pay attention to where they are. So go from Hall 6 to Hall 8 to Hall 1 back to back. so I would say I think Chetan hit a good one. 6G, interesting 6G conversations, not as many as I expected, particularly around questions around geopolitical concerns and fragmentation of technologies. A lot of vertical conversations. I think if there was one sub theme that I thought was really interesting, it was, critical communications and defense as a standout growing, vertical interest, if you wandered the halls enough, you saw the folks from the DOD there, we all know that the U. S. Department of Defense is, looking at some of these commercial technologies in particularly interesting ways, a lot of discussion around network resilience to protect against either manmade or natural disasters. and really, We know those other verticals that we've been talking about, ports, mining, utilities, you've seen some of those, but I think the bubbling up of defense was a really interesting one that kind of captured the interest in pulling verticals in, the monetization of those networks, things like Open RAN, but then also, the geopolitical angle. So I think it was an interesting, encapsulation of all those things.

Chetan Sharma: 7:25

yeah, for me, the thing that really stood out, probably from day zero, just like Saturday, Sunday was the geopolitics. I think if you sit back and say, what are the ground truths that I didn't know, coming into the show? geopolitics has been an overhang for a long time, but the way it accelerated and permeated in every conversation, behind the scenes, was a bit striking. and for obvious reasons. beyond the tariff, discussion, I think there is, something brewing that could be a permanent reset in how alliances are formed. And we just don't know where it's going to all settle, but there is the questions being asked are very different than what they were 2345 weeks ago. So that was something new and different. NTN, as expected, is surfacing up as, kind of a must have feature. There is still a distance between, Having the feature and monetization, but NTN has jumped as something that every operator is looking at every country is looking at the lack of 60 or I guess, there's not a lot about talk about 60, but what will lead to 60? Like what's the 5G advanced path? What features are going to be there that lead to 60? so I think that's the right discussion. So we focus on here and now, and as you said, and then figure out what the road map is for 6G. And we, because I had this fear of, getting sucked into the 3GPP vortex, and not really stepping back and saying, let's learn from what happened with 5G. and focus more on the economics of these solutions and the infrastructure rollout versus just blindly following one another off the cliff.

Peter Jarich: 9:12

it is worth noting. I think maybe some of the, and again, I was surprised on the 6G front because, last year we had conversations around, let's make sure 6G remains a global standard. You would think a year later that seems even more in peril, given where we are in the world. I was surprised to think about more. We all know this week in Korea, it's the big 6G a thon, right? And there's a lot of conversations, a lot of discussions. So that may have stolen some of the thunder perhaps, and maybe siphoned off some of the discussion. But yeah, it was, refreshing. I don't think any of us missed the 6G conversation. I can't say I was actually missing it, but yeah, it wasn't quite there.

Leonard Lee: 9:50

I think there was a, like you mentioned practical before Chetan, a practical grounded focus on 5g, a lot of talk about essay, at least in the discussions that I've had, and, I spent a lot of time at their Ericsson booth. And so, you know, as we all know, and we've been to their events, 1 of the things that they're looking to do is really move the ball. On 5G, even ahead of all the 6G talk, I think that those are 2 separate conversations at the moment, because the dilemma right now is how do we continue to progress 5G where operators can benefit from. the, what I call this technological backlog that we have now with 5G as we're moving toward specification release 19. I mean, there's all these features you mentioned features, right? All these features that are untapped. And so. you know, 1 of the things that I think is important, and I think you're starting to see this urgency start to form this year is the need to tell the story about how our operators pioneering operators like T Mobile and others around the globe leveraging these new capabilities. technologies to create new services and what are some of these patterns and modalities of, monetization that can be had, right? Because, this whole monetization topic has been such a ridiculous one. Quite honestly, yeah. And and 1 of the things I noticed is there's this huge gap between technology and monetization piece. There's this big gap that hasn't been addressed and spoken to very well. I think we're starting to see people talk about it and, I think that's another hopeful outcome, of this more grounded agenda, whether it was intended or not. It's just it came off like that to me of, of mobile and Congress and what it seems like, Either subconsciously or consciously pursuing, this year and hopefully going forward. So anyways, I don't know what it is.

Peter Jarich: 11:57

I think it's interesting to your point, though. So I think it is tough, right? I think it's tough because you have different personas, which would show up to any show. So there's some people that are there want to think about that bigger transformation and think about how do we actually drive outcomes and monetize. There are other people who are there to talk technology. But you will always that they want to understand how we enable slicing. It's not their job to figure out how you make money from it. So it's always difficult when you've got that multi, that sort of different people there. But I'll say to your point, the number of times I heard people sort of reference things like the T priority service and what T-Mobile was doing as like here's a great example and we're actually seeing slicing and this is kind of cool. So I think people are, when they see those good examples. And that one I think plays to exactly what you're talking about. There's a lot of essay. There seemed like there was a Guarded optimism around SA, maybe, finally, sort of, potentially, taking off. and, you know, because we've been talking about it for a while, and we've seen the slow growth, and we know it's what holds back things like slicing, and we know it's what holds back a lot of things. I think it's also holding back 5G advanced discussion, because a lot of operators are going to want to have SA before they do. Even though it's not required for a lot of 5G advanced capabilities, a lot of operators will want to do it. So but yeah, I think some of those good examples, you're 100 percent right. We need to see them. And when they're good ones, people do get excited about them.

Chetan Sharma: 13:25

some of these features need to be consumable just because they exist doesn't mean third parties can just take advantage of them. slicing is a great example where last year Erickson showed how slicing was integrated into the Sony cameras. that's a great example of The third party not worrying about the complexity of the technology is one to consume it and figure out the pricing and monetization on their end. And that's what you need from these features. They need to be exposed in a way that they are consumable. Just like an Amazon API, right? I just have my credit card. I want this feature. I put my credit card. I consume it and that's. That's the level of simplicity. It needs to be there, which was not there prior to that. I mean, slicing has been around for some time. I think we're getting to a point where these features, whether the operators used for their own purpose, like key priority or network management or congestion management, or you make it available to the enterprises who can consume it on demand as needed on a predictable pricing plan, then it becomes all of a sudden and you don't have to talk to anybody like you just go and figure it out on a website or through an A. P. I. I think that's when it scales. If there is too much hand holding or too much complexity, it holds it back.

Leonard Lee: 14:41

Well, yeah, regarding the network slicing, the way I look at it is more of an underlying capability. And if you, as you move toward five G. S. A. S. You have the ability to institute network slices in a very dynamic way. And what that means is that, from a service, perspective, you can design things that are event based or,, condition or context based. Right? So these can be deployed. Very quickly, they may be high value scenarios, but they're not permanent. they're ephemeral. They're there when you need it and then they disappear and then they're also balanced against whatever available capability and capacity in that particular location. And so, I think that's 1 of the things that's important for the industry to understand, because, like, you guys gentlemen have been saying, Network slicing has been a part of the telco discourse for a long time. A lot of people scratching their heads about, when's all the hypothetical stuff going to happen? oddly, the weird thing about technology is that, you know, it does realize itself over time. And I think we're at a different point now where the technology has matured and you can actually realize the capabilities are beyond what we had before. And so I think, a lot of us have been dishammered I don't know if that's the right term. Disillusioned.

Peter Jarich: 16:09

Yes, it's fine. It's fine. Disenamored is a good word. Disenamored. Okay, we'll get into the dictionary. Yeah,

Leonard Lee: 16:18

I was always really bad at the verbal section of the anyways. and so there is a bit of a overshooting. We disregard things like, ah, you're bringing up network slicing. We talked about that forever, but there's, there are these new possibilities. I think we just need to frame those in the proper way and, I think that that's the thing that we can do going forward is help people understand what is. Yeah. Now this nouveau role of, network slicing and all these other things that come together in compelling new ways. so anyway,

Peter Jarich: 16:49

I think as much as pragmatic though, and I don't want to be the negative Nelly or whatever. However, you referenced yourself, Leonard, but I, from a pragmatism standpoint, I saw a couple of times people talk about things like network slicing, right? Network slicing. And I think taking what you were saying chatting about making this stuff consumable, right? If I think about consumable and I think network slicing, it sort of leads to a conversation of network APIs, the open gateway initiative. We've seen all everything that's taking place on there and extension of things like quality and demand. Network APIs, big discussion, obviously this year, last year, we start seeing more and more operators getting involved. Quality on demand, a lot of the initial work has been done, around identity, sim swap, fraud, security, things that are, there's a very clear ROI. People are looking towards quality on demand going forward. Which kind of goes to sort of the broader slicing story, but I had some interesting conversations. People say yes That's really that's really exciting if we can get to the end of the year with one or two good use cases That's wonderful. We need to be patient and and I've heard people talking about and I think if If you were out there, and I think you were at the like Erickson pre brief They had the CEO of Aduna and really great job explaining sort of the business model It's a look some analysts asked him about numbers How much activity do we expect going forward? I said, look, we just need to be patient around this stuff. This takes time. Totally appreciate whether it's rolling out quality in a man, whether it's predicting the future on how much activity we'll see there. Appreciate the pragmatism of we need to be patient, but I worry because no one's patient. this is not a world where people are like, yeah, no, no, I take a damn. Heck, you know what? If you were doing badly financially, I'd be worried, but you're not, you're making tons of money. The industry's in great shape. Take all the time you need. That's not the case. we're not in that space, right? We're not in a space where telco is, pulling in tons of money. Right. We know that, right? you sit, it's the message you hear in the keynote, the first keynote of the year, you see that giant chart that looks like this and the number that looks like this shows costs versus whatever metric, right. And balancing that against, Oh, don't worry. We have time to figure this out. I do worry that. There sometimes, especially with some of these new technologies, there does need to be a little bit more. Um, uh, no, I'm at a loss for words there, Leonard, right? And I think, we need to recognize the continued urgency around making some of these things work because as much as. I think we're also in a better place where we know how the hyperscalers are partnering with operators. They're also, in many ways, on some of these things, very potential competitors, right? And unless we get things figured out, right? Unless the op how many times have we talked about a potential future in which Apple figures out how to make slicing a reality? We know it's in iOS, embeds it in, you know, the app store and sort of kind of like we saw with driving the, driving the narrative on eim, right? I think there's be a bit of an urgency there of, of making sure we execute on some of these capabilities and a lot of operators have it. Don't get me wrong, you saw so much activity from so many operators that they get it. I just think that needs to be a story that, they all understand.

Leonard Lee: 20:13

Well, Chetan, any comment? I mean, he's making a great point, by the way. I think it's something that.

Chetan Sharma: 20:18

No, he's spot on. patience is absolutely warranted, but also you need a rapid, I think countries where all 3 operators or all 4 operators are on the same platform. It's relatively easier to offer that to the developer community or the enterprise. Versus if there's only one operator from a country who is signed up, I think you would expect that those operators in those countries will make good progress and the numbers will start piling up and at some point you have to discuss the numbers, because, you have to do trillions of transactions to make billions from all this stuff. And so we just need to see what path are we on. in terms of number of transactions taking place on a daily basis, once you have a sense of the trajectory, then you can do basic math, right? how many operators you add to this mix, how many more API's you add to the mix to see if there is a hockey car or it's a steady growth with both in both cases is fine, but you do need to show progress on a year or your basis that so it's meaningful. Otherwise, you know, As you mentioned, their competitors, which probably offer a more holistic offering, with storage and compute as well. And so if developer is thinking about a new way of driving some application, they just want a one stop shop. They don't want to go for certain applications for certain APIs to the operator in certain. Workloads to the hyperscalers and since so there is the tension as well as urgency to address that.

Leonard Lee: 21:50

my observation is a lot of the stuff is still pretty well misunderstood. I wrote a number of pieces in the run up of MWC about. Network APIs and Aduna in particular and it's role as an exchange. And so I think you guys are making a great point here. What is this Aduna thing? And what is an API exchange? What should we expect of it? And should it be the same as, let's say a platform. Right. But it's not entirely that I think 1 of the things that the Ericsson folks. And their partners have been trying to emphasize is that this is, this is kind of like a semi for profit. This is like profit to, you know, to profit enough to survive or sustain itself type of model. And I think that's where we have to recognize that from a business, entity perspective, right? And its role within the broader ecosystem. But Peter, you're absolutely right in the bigger picture. It needs to figure out. Okay. what is that large scale value that network APIs are going to, deliver and what is the role of the aggregators in these, network API exchanges in, scaling the markets for, exposed network functions or, capabilities and, I think trust is going to be a really huge thing. Guys, look, I don't know if any of you go to RSAC, but the fear around AI is palpable. It's growing because the threats that, AI, in particular, gen AI augmented. Attacks present is just absolutely frightening for the cyber security and cyber trust, community. And so in simple things like trust security, I think there's a huge opportunity for these network APIs where the operators are going to be able to, express their, network capabilities and supporting a lot of. Trust functions above the level and so this becomes less of an ROI thing for, companies, but more of a huge mismitigation because literally, I don't think like, you know, I look at my mailbox about 90 percent of the stuff I get now just spam or some kind of really compelling phishing, email or a prompt to get engaged in a conversational phishing. Chat stream. I mean, it's outrageous.

Peter Jarich: 24:23

Stop there before you get yourself in trouble. This doesn't sound like it's going. I don't know what these conversations are, but no, I mean, I think it's, it's, I think, I think it's your, I was about to say I was worried about AI before I saw the. Robotic vacuums that we decided to give AI and limbs, you know, not now that they're robot vacuums that can pick things up and they've got AI. I was. Oh, my gosh. No, but I think the difficult part is. Here's where we are. Two years ago, we had to explain to people why this was important, right? We had to explain that, oh, look, a common set of APIs is better than proprietary APIs because it can scale. I had to explain that. But now we're in the nitty gritty of how does it work, and how do exchanges work, and what are the pricing models, and what are the, and I think to some extent that's what folks like Aduna are there to help. Sort out, but it doesn't make it any right. So it's a much more complicated on top of that then building the demand, right? Going out building that, and I work with folks, here internally at the GSMN We're trying to figure out how do you build that demand? How do you go to potential? Consumers of this and say, here's why you want this, right? The social media companies that in some jurisdictions will get in trouble if they're serving up social media to someone who is under 16. So if there's a, if there's a simple way to go and prove this, but then, and here's the fun, like we have a round table on network APIs and developers, with a bunch of folks. Here was an interesting sort of thing that I had thought of. The industry has spent you talked about fishing and that's there's no joke there, right? I don't mean to make light of that fraud. That was pretty funny, though. I don't know about yours, but fraud is a horrible. I mean, there are all sorts of horror stories about people who lose not only their life savings, but in different economies are finding themselves basically part of modern slavery because they've they fell. The file of something so this is really horrible but we spent years telling people dual factor authentication and do all these things what's going to happen when now we use api's to make that more seamless we're going to have to reeducate people of well don't worry the operator has. Maybe the operator solved that by knowing where you are on your phone. And so I think it's more than just business level transformation. It's transformation the way we engage with our operators, which is really where they should be the whole time. They should be in the middle of that. So it's, yeah, I think ultimately it's a really good thing.

Leonard Lee: 26:57

any final thoughts on this big topic, Chetan?

Chetan Sharma: 27:01

I think it's to be continued.

Leonard Lee: 27:06

Okay. All right. So the next thing I want to do here is just get your real quick takes on a number of themes that, I thought, were pretty important going in. I mean, or at least, you know, they're, carry overs from last year, some of them, but then some that we anticipated would be, big things. this year. So the 1st thing that I want to ask you gentlemen about is the question that everybody wants to know. Okay? Metaverse. Okay. Thank you. Next topic.

27:45

I'm sure in a couple of booths there were. Interesting to go there. I'm sure there

Leonard Lee: 27:51

were a few. I, did I not see the off side? Yeah, I know we have to be kind. All right. So, the elephant in the room, if not the el, just the elephant. generative AI and agent ai. What did you guys think?

Chetan Sharma: 28:07

I mean, gen AI transformed into agent AI this year, and last year it was all about gen ai. This year it was all about agent ai. Yeah. so I mean, AI has been so. Clearly there's the hype part, but then if you look beyond the like beneath the surface, especially in,, there are three areas that have surfaced over the last year where AI is making, an impact that it was not doing before, one is, clearly customer service, and that's the low hanging fruit, Second is knowledge management within the enterprise. And so that's not unique to telco. It's across, all the enterprises. But the one that where it's really starting to make some dent and it's a very complicated, deployment is that of network management. How do you automate functions of an enormously complex, network and network nodes in a way that, manages congestion? It does fraud detection, detects anomalies, solves real time, issues automatically, energy management and so on and so forth. So I think there is definitely, progress on that front, because there are a lot of software solutions to generally hardware ecosystem and here, both, the traditional OEMs, but also a number of startups are taking a crack at it. And, you know, companies like Openga Networks or Aira and others, are making substantial progress in how you ingest network data. And then turn into a some sort of actionable intelligence in a closed group manner, which is enormously hard to do. Yeah, and so there is definitely good progress in terms of on all these fronts from the startups from the from the OEMs. And that's what I'm most excited about.

Peter Jarich: 29:54

Yeah, yeah, Peter, I say, I mean. You can't spell agentic AI without Jenny. I just, uh, yeah, I mean, honestly, Jenny, I know. So I think, I think what's interesting is we always need something new, right? You don't want to go another year and say what? So yeah, agentic AI was the new hotness this year. Everyone wanted to talk about it. I read a blog post, which I think was really spot on where it said, remember a couple of years ago when Gen AI was the big thing. And then everyone you talk to, every supplier had gen AI integrated into their solutions. They're like, look for the same thing for magentic AI. And you sure, same thing, right? You saw the number of booths come talk to us. We'll develop an agent with you. We'll develop an agent for you. And it really did take over, but I think ultimately I think it took over because the question is, well, what do we want? AI to do it. We want to actually deliver an action, right? So I think it is pointing to that interest there of actually having a human in the loop in some cases, but actually taking action where it becomes really not whether or not all the agentic AI we saw was really agentic AI or not. I don't know. But what I think was more from my standpoint, I think if we were looking for something new, Agentic AI was clearly there, but what I thought was new on the AI front, which I thought was more interesting for me, was this focus on what we can do with networks to support AI, right? Because we know that we've all had this conversation around AI for networks versus networks for AI. And Chen called out like the perfect example of AI for networks, right? We're seeing AI being used, network automation, network optimization. We're seeing progress there. That's where a lot of operators are focused, but where they haven't been focused is how do we need to Invest in and develop our networks to support future AI workloads. We, that's been a discussion point that's been out there for a while. And of course, all the infrastructure vendors are going to want to talk that up. Hey, you know, make sure you're investing in your network to support AI. But I think we saw some interesting stuff this year from whether it was like AWS working with operators to get their data. Just to get their data in place, which is critical or whether it was the Verizon launch right before MWC focused on edge and edge workloads and edge inferencing, whether it's just people like building out backbones, I think you're beginning to see some of this focus on as a way to monetize their networks, recognizing that AI is going to require certain capabilities. And that I thought was particularly interesting. Agentic AI was the buzz, but this bubbling up of operators who seem to be investing in places that will support AI. With the potential to monetize was more encouraging.

Leonard Lee: 32:29

Yeah. Yeah. Yeah. My observation was, that we went from last year, there was this notion that generative AI would be able to do things that are very granular and deep in the network, code generation spit out, network config files on the fly type of thing. I think we've. Moved away from that. And now, there's a realization that you really have to start from the top down and so doing things in a longer time scale, where the general AI tools can have the opportunity to. Assist a person in the middle, in making better decisions. And so it's more of an assistive, copilot type of model I didn't quite see anybody with the courage this year to suggest that you're going to take this stuff to, level 4, even 3. Autonomous networks. and if anybody can actually show me something in production, give me a call. give all of us a call, or maybe Chetan knows somebody that did it, but I didn't see anything like that this year or anyone courageous enough to say that they've deployed it in production. I think generative AI is starting to, again, get grounded, the conversation get grounded whether people know it or not. And then the agentic stuff I think is really just becoming augmented or slightly enhanced RPA. You know, you look at the framework, you look at the tooling, you look at how narrow the deployments are in terms of the actual, application of generative AI. it's starting to get grounded as well. And that's what we can expect in every hype cycle. Right? These technologies land where they are actually usable and these are some of the patterns I started to see. I think we still have the rest of the year for the industry to catch up and, realize that, these are, the valuable, implementation, patterns, for using this technology.

Peter Jarich: 34:38

Be careful, Leonard. Be careful. so I think you're right. And it was, I was at CES, I was in a session around AI and someone said, Oh, agentic AI is the big new thing. Does everyone out there know what it means? And there's a bunch of people like it's RPA on steroids. And I was like, Oh, okay. But what's interesting is, and I think we all need to be careful of this is, is I've been in so many group chats and emails where someone talks about some new agentic AI innovation And then you get like, a dozen responses going. Oh, I that's just ml. I've been doing like I was doing that 20 years ago with cobalt and I think you do see this, particularly with folks who've been around the industry for a while to sort of, I know we're all a bit jaded, but we look at these things and go, oh, that's not new. that's machine learning and I was doing that in college back in the yeah maybe not so I think and I see that sort of a well agentic is just RPA it's a great way to describe it. But we still need to be open to at least some of this being innovative or else we're going to seem old. We're going to seem old and cranky.

Leonard Lee: 35:38

No, I think that's where the grounding happens. You know, actually, the odd thing is, as you get older and more experience, you do see that, the newer generations don't know about the old stuff and then things that were esoteric to the old generation, all of a sudden become a new thing. And it's. That's why I think analysts play an important role, technologists who have legacy play a role in reminding folks about the history. And there's nothing wrong about saying that it's RPA. AI is an incremental, technology as much as everyone says that it's revolutionary. And it will make things that are already on the continuum of, AI evolution, to make, you know, application patterns like RPA even better and there's nothing wrong with that. That's great. But I think the height, narratives suggest something far more dramatic and outrageous than what the reality is. And so, you know, I guess, I, I will dare to be dangerous.

Peter Jarich: 36:46

How's that, Peter Wow. You were, were you doing a Nick Cage, impression there? It sounded like it sounded No, I think I was Oh my God. I love you guys. That was good. No, I think it is. I think for me, I think just on the adjunct conversation, right, I think you're right, having a bit of a history there, because, you know, what was the one thing that we didn't see this year? We did see AI devices. I mean, don't get me wrong, I think DT. Had their example yet, but last year, right? Oh, that's true. The big last year. There was the humane pen that's right. That was sort of the big news and those kind of all went away, but the reality is Those were just hardware forms of agentic ai. Yeah. You were supposed to tell your rabbit, go and make a buy flowers for my wife. And it was supposed to go do it, right? But we weren't quite there yet. And so I think it is important to have that sort of history and go, okay, well maybe next year's the year for those devices where we get agentic AI figured out. I don't know. Yeah.

Leonard Lee: 37:43

We'll eventually maybe get there, if it makes sense. I think is really what it boils down to, now thoughts on private 5g, because last few years, private 5g was supposed to be sort of this, you know, stimulant for, for the 5g movement, any thoughts there? What'd you guys see? In that regard,

Chetan Sharma: 38:04

you know, kind of steady progress. from last year, I think they're in private 5g. There are two worlds, China and the rest of the world. And so in China, it continues to be on a hockey curve. the rest of the world is making steady progress. And so the question is, what are those forms of progress? you know, operator spectrum, CBRS, and so on and so forth. And so it's still not at a point where it's at the knee of the curve. It's just, you know, you go from 10 to 20 deployments, but still fairly minor stuff in rest of the world from a private 5G point of view, where the real progress is happening in China on private 5G front.

Peter Jarich: 38:40

It is, I think, yes, and we saw definitely picking up. It is incremental. You need the solutions in place. You need a broad ecosystem. and I think it was always going to take a while for this to ramp up. You need the enterprises. I think this goes back to, what you were saying, it means those good examples, right? So as you see those examples. it becomes much easier for some other enterprise to say, I get it. Here's how it fits into what I'm doing. And then understand it fits into every vertical is neat because every vertical is different, right? Manufacturing has a different requirement than mining, then a porch, then public safety, then whatever.

Leonard Lee: 39:15

And you know, Chetan, I read your, event summary, by the way, awesome. Anybody out there who wants To read a really nice recap of, MWC, you know, go to Chetan's LinkedIn profile. he has that posted as a newsletter. I'm sure you're going to have it on your website as well. But you mentioned something about the reluctance or I don't recall how you described it, but it was along the lines of enterprises are reluctant or cautious about approaching.

Chetan Sharma: 39:46

I think what I was, talking about is the operators being reluctant to go after the enterprise market,

Leonard Lee: 39:51

the

Chetan Sharma: 39:54

way they should. I think the big part of money we have been saying since, the start of 5G is the enterprise market. And I think only China has cracked that nut. and other operators are just, not quite there yet as to how they want to approach that marketplace. should they do it themselves? Should they go through the channels, through SIs and so on and so forth? So I see a reluctance, which is born out of, should we build this? First, this capability and then go to the market or organically grow it. And so that is, taking time. whereas in China, all three operators are just hitting the road.

Leonard Lee: 40:33

the other thing, that I'd like to get your takes on, AI ran, you kind of hinted a little bit about it, Peter, but what were your observations about AI ran MWC? I know it's a rather new topic because AI ran alliance was introduced just last year, or at least it formed last year. But where do you guys think we are with AI RAN, how's that conversation, shaping up?

Peter Jarich: 41:01

Yeah, I think there's a lot of questions about what it means, right? So I think when it formed last year, to some extent, there was some, and they did a really good job. I think if you pay attention to what the AI RAN Alliance was doing, and they've grown their numbers for a reason, because, people see that connection, but, I think there was some suspicion. Right, because of the various different use cases they called out, one of them is, run your AI workloads on the same compute that you'd run as you're doing RAN and AI the same. And when you're not doing your RAN, you could do AI stuff. I think there was a bit of a suspicion of, is this just a way to sell me more chipsets and sell GPUs? Wait, is this just NVIDIA saying, you need to buy more GPUs, and I'm going to give you more, I'm going to explain to you other reasons why you need to buy GPUs and a way to grow the market. And so I think there was a bit of a suspicion there. And also, are we ready, prime time, for running these RAN workloads on GPUs? What I think is interesting is, a year later, I think people have sort of understood more of everything that this means, right, which is part of what Chetan was saying, using AI to improve our RAN operations, right, embedding AI into what we're doing in the RAN to improve that. And that's obviously going forward towards 6G. some of those use cases of how would I run RAN on GPUs. but also recognizing that's possibly a bit all of the above as I evolved my network. And here was an interesting example that just talking with someone, this is for anyone who hasn't been done to see part of. I think that the value of being there is analysts that you have conversations that things make sense. But I was talking with a big operator who is trying to think about their enterprise strategy. Right. And the first part of most of these enterprise strategies involve things like, putting a server on prem. Right. And then you do some of that server. And they're a big proponent of virtualized RAN. So, hey, so I'm going to put the server in the enterprise, run some RAN workloads on it. Now, in the enterprise, they may not need to be running that RAN with the same level of redundancy or the same level of capacity as you would in the outdoors. So if you do have a server that you've deployed in the enterprise as part of your enterprise 5G strategy, and you're running RAN on it, and you have spare capacity, Well, then maybe it makes sense to use that same capacity to do some AI things, right? And when you think about it like that, and you connect those dots, then people go, Oh, okay. So yeah, I could see how this could all fit together. So I think as people understand what AI RAN means, everything from optimizing the RAN to doing multiple different workloads. AI for ran, ran for AI, AI on ran, whatever, again, same conversation before, it takes a while for people to get, but it makes sense in some instances, and now we just need to see where those instances are and how they play out. Does any of that make sense? Yeah,

Leonard Lee: 43:44

yeah. So yes, it did make sense. So congratulations, Peter.

Chetan Sharma: 43:52

So, it's, quite interesting at the macro level, when Jensen was talking about a sovereign AI, I think, that pretty much defines how the two camps are progressing. So countries who have believed in sovereign AI, given what's happening with geopolitics, I think in those countries, They're also the hyperscalers and not that deeply present. operators play a role in data centers. I think those places the air and is getting traction first because there are multiple ways to justify the investment, building your own data center using that I. Capacity for your own developers in the country, keeping your data in house, keeping your models in house and so on and so forth. And so Indonesia, Japan, Korea, India, some of the Nordic countries. So those are the first movers in the space where, the hyperscalers are deeply present, like Western Europe or North America. there is a hesitation and a wait and see approach in terms of figuring out. Well, if I build this fabric and I don't really have a data center business to justify it, then will I be able to create demand or not? Let's keep the discussion around whether running ran workloads on GPU makes sense or not. Just from a pure business point of view, can you create enough demand for the process? Other workloads that's supposed to be run on the GPUs. And so there is some hesitation from operators. They want to wait and see before they decide one way or another.

Leonard Lee: 45:27

Yeah, the way I looked at it, as this whole, maturity or evolutionary curve that you describe Peter in your earlier comments I think that 1st phase of A. I. N. ran is something that. Is relevant today. Everything else I think is too hypothetical, especially the thesis on, the monetization of, access GPU or available GPU capacity at any given time. There's a lot of technical gaps there that need to be bridged. I think operators and the industry eventually, and this is probably going to happen next year, are going to realize, look, there are too many bits and pieces that are not there to get us to, the fourth state of AI, or the fifth. but that 1st 1 is really interesting. It has very little to do with GPU. It's more about how do we apply various techniques from the radio all the way to, let's say the Rick or the S. M. O. in, driving some optimizations more, let's say. Optimized sleep functions and, power management, like those simple, boring stuff that actually has huge impact. There was another thing that, oh, geez. I wish I recall. yeah, there's like fleet management concepts as well. Right where you can use advanced, AI techniques to manage a fleet of, cell sites. So that, you are using. Your backup power to, run your cell sites during peak rate hours, and then recharging at night at lower charging time, windows in the day or the evening, I thought, something simple like that was awesome. And you're looking holistically at the whole. system and you're applying AI in a very, very ingenious way, actually, that has outsized impact in terms of cost savings, just beyond the efficiency of, of, you know, the ran itself. Right. It's just literally a out of the box thinking, really cool solution. And it's something that I saw at the Ericsson booth. so those are my thoughts on that. I kind of think that, AI RAN is going to go the way of Open RAN. So given that I am the inventor of Open RAN Squid Game. Let's talk about that for a moment. What happened to Open RAN?

Peter Jarich: 48:09

Oh, sorry, I froze. My screen froze. I didn't hear the question. Uh, Chetan, you answer, because I didn't hear the question.

Chetan Sharma: 48:21

Open RAN, the, It's like, whatever it's not a defined thing. It can be whatever you want it to be. And that has been the problem from the start. I always felt that Open RAN missed the 5G cycle because it came in probably a couple of years too late. And by that time, a lot of the prominent operators had already Figured out the architectures and roadmaps and so trying to fit that in into that discussion has always been challenging and operational part that Open RAN players have minimized Over the years is the toughest part to overcome You know when you go talk to people who run these networks on a day to day basis, they don't want to get anything touched by A plethora of vendors in the name of efficiency, like if you are gaining 10 percent efficiency, 20 percent efficiency, but operationally you're going to go down 30%, like it all nets out. So the challenge for Open RAN is, you need, one or two players who take on the burden of putting it all together and be the interface to the operator to run the networks in these various markets. Unless you have that system in place, Open RAN kind of falls apart. Like, if you have 20 vendors And maybe two lead vendors, how invariably something will go wrong. How do you go about debugging the problem, A, and B, who do you point to, to fix it? Who, who gets penalized for that outage and so forth? So the operational things have never been like streamlined or well understood. And I think that has been the undoing of the Open RAN principle, as it was stated early on, right? Now Open RAN has morphed into something else, like either, one player Open RAN or whatever it might be, it become into 6G. So the notion of, open interfaces is a valid one and a useful one, for the ecosystem. But how you operationalize Open RAN is where it gets tricky. that has been a challenge with Open RAN.

Peter Jarich: 50:26

Now I thought you were going to take us in a different direction, Leonard, but I would agree that AI RAN is going to go the way of Open RAN in that it's just going to become a background thing that exists, and maybe the terms will morph and it will mean something different, but it's just going to be there and we won't talk about it. I had a conversation with a small software supplier that develops RAN stacks. they sell into a lot of different vertical industries. And they said, we don't talk Open RAN. We interoperate with everyone. the concept of the basic idea of what's going on with Open RAN. Now, would there be a market for us if there hadn't been a big push on Open RAN? Maybe not. And so the concept that Open RAN was going for is sort of alive and strong, whether or not we use the term. Another comment I was talking about. I'm talking to the bd guy was saying i'd never talk open rent, They deliver silicon, right? You know ran silicon small small players supporting the particularly on the radio side of things and said look i'd never talk about open ran And the marketing guys like I do, it's like, well, yeah, that's because you're trying to highlight what we do and people get it. It may not always be exactly open, ran compliant, but what they want to know is, will you work with different baseband providers? Will you work with different, you know, radio, like, will you fit into an open solution? And I think AI ran will probably, I think you're right. It may go the same way where. Maybe it means different things, but the concept of integrated AI into the way we operate the radio access network and the way we run things, I think will be there, whether or not it fits those exact terminologies. I think same thing with Open RAN, right? I think, yeah, it's going to be single vendor. In some cases, it might be ecosystem. In some cases, it may not actually be Open RAN. Right. You don't need to do, I think a good example is you're talking to SMO and RIC, right? You don't, I mean the SMO and RIC, I mean the RIC is an Open RAN construct. Yeah. It exists, but you don't need Open RAN to have a RIC. you can institute a RIC without an Open RAN network. Those two things are separate. At the same time, It was born out of Open RAN, and I think it's a good example of like, we may not talk about Open RAN in the same capital O, it might be small O Open RAN, not capital O, but the basic concepts I think are, you know, that's what's going to have the impact on the market.

Leonard Lee: 52:36

Yeah, I totally agree with you on that last bit. I'm a big fan of the SMO, and the RIC, especially the non real time. I think it's more about selectively interoperable ran versus open. Right? It's like, what kind of tools and do you want to use what level of interoperability? What set of. Of, open interfaces and where at what level, that's basically what it's going to boil down to. And it's going to be specific to each of the operators because each of

Peter Jarich: 53:09

them. that's just a really bad marketing term, selectively interoperable ran coalition. Right. Selectively. I mean, that has marketing all over. But you're right. I mean, It'll work in some but you're right. I mean, that's the thing. it's not sexy, but it's right.

Leonard Lee: 53:24

We can start it next year. Yes. So, okay. To wrap up gentlemen really quickly, where do you think we're going to land next time or next year? where are we going to land at M. W. C. 2, 026. Quick thoughts for our audience. Say more.

Peter Jarich: 53:44

I think we'll still more.

Leonard Lee: 53:45

All right. That was easy.

Chetan Sharma: 53:47

I think I will continue to be the dominant theme next year as well. Probably more NPN, discussions, more, alignment, hopefully less geopolitics next year, but

Leonard Lee: 53:57

Okay. So here's mine. I think next year, instead grappling with global first, because of the geopolitical challenges. How does the mobile industry stick together and not become so fragmented that we lose all the benefits of all these years of trying to come together and create common standards, interoperability, all that stuff. I think that's going to be a real. big topic just because, like what we got, we've been saying, actually, collectively, I think, through this session, which, by the way, gentlemen, I think is friggin awesome. I love you guys. I want you to know that. I love you guys, uh, we love that. Hey, that that's what we're going to have to deal with. And I think through the course of the year, the 3 of us and the analyst community are going to be really challenged with questions. And, a lot of the industry is going to be looking for guidance. And so hopefully we'll have some further conversations and debates about that. Gentlemen, thank you so much. this is really great. Peter is great to, you know, have you on again.

Peter Jarich: 55:09

Bug me, bug me anytime.

Leonard Lee: 55:10

Yeah. And Chet, no, of course, always. I mean, this guy's like frigging legendary, right? He's like, I'm going to see you at GTC. Yeah, are we going to hang out or are we going to avoid each other like we did?

Chetan Sharma: 55:23

I don't know how it happened. Maybe it's a smaller, smaller set up. So we'll run it through. Yeah.

Leonard Lee: 55:28

Yeah. That, that, that was really tragic that we didn't bump into each other.

Peter Jarich: 55:33

I'll send my digital twin. I'll send my digital twin to GTC. So you can, you can meet him. You can meet up with them.

Leonard Lee: 55:38

Cool. Yeah. And you can hang out with Jensen's, you know, 1 billion digital twins. but, yes, everyone. Hey, thanks for listening. We appreciate you. but watching this very special podcast recap of MWC 2025, make sure to contact and connect with my guests, Peter Jaric and, Chetan Sharma. I suggest that you reach out to them on LinkedIn. That's probably like the safest way these days and find out more about Peter's work at GSMA intelligence at www. gsmaintelligence. com. And of course, the illustrious Chetan Sharma and his wonderful work at Chetan Sharma Consulting. Just go and check out all his great content at www. chetansharma. com. Consulting. com. And then also please subscribe to our podcast and it will be featured in the next curve YouTube channel. Check out the audio version on buzzsprout and also subscribe to the next curve research portal at www dot. Next dash curve. com for the tech and industry insights that matter. And we will see you next year to recap MWC 2026.