Network X
14 - 16 October 2025
Paris Expo Porte de VersaillesParis, France

Ruth Brown

Principal Analyst – Mobile Networks

Heavy Reading

AI for Mobile, 6G and Network APIs

Ruth Brown, Principal Analyst – Mobile Networks, Heavy Reading

Monica Paolini, Principal, Senza Fili

Emma Buckland, Senior Analyst, STL Partners

Lycett Christopher: Of summits and discussions really getting kicked. 6G started. Would you? Is there anything, I guess, in particular that stood out around that for you and anything concrete that seems like it's going to happen?

Ruth Brown: So I guess in terms of concrete, as Monica Paolini said earlier on in the discussion, it's still early days, but from the initial workshop in Korea there were some key points. I guess it really stuck out, I think. Firstly, there's a very different focus this time for 6G. It's a very different generation. So in the past, it's all been about new hardware. You know, refreshing the mobile core architecture, that kind of thing. The message is very different for 6G, so it's more about a gradual evolution of the 5G architecture and also sort of infrastructure or software. There's also a real push to be more realistic in terms of the business needs. They're trying to engage with more verticals and also not to overinflate the performance metrics, which I think we've seen in previous generations. So I'd say that was one of the big highlights from this and then you've got other messages backed by kind of collaborative groups of operators, things like making a much leaner specification for 6G. So not many options not sort of bloating it out too large. And then other sort of big aspects and strong pillars that we've seen across many other standards, things like the AI native message, stronger security, programmability, ubiquitous connectivity.

Lycett Christopher: Around less specifications, is that aimed at being more cost-efficient and more streamlined across the industry? Or is it something else that's driving that?

Ruth Brown: Yeah. So really, so in the past and in every generation, we've seen the widening of standards. So typically things like you've got lots of options for the same kind of capability and it makes it particularly difficult for devices. So if you imagine you've got quite a wide spect for devices, it makes it more difficult to get that into production. So by making a leaner standard, it's...

Lycett Christopher: Right.

Ruth Brown: Hopefully in the future going to be easier for people to implement in a sort of more standardized way.

Lycett Christopher: OK. And then you also mentioned overinflated metrics for previous generations. Is that around like overinflated metrics in terms of network performance or overinflated goals or overinflated, maybe marketing messaging?

Ruth Brown: Sure. So probably every kind of new generation, you get these kind of headline figures, what latency should be, what the sort of speed should be, that kind of thing. They always go up by at least tenfold, sometimes more. So I think when we sort of saw 5G coming out, first of all, it was like a one-millisecond latency figure for 6G. Some of the initial figures put out there are sort of much much lower. So really from the outside, and this was another submission backed by sort of global operators, was let's be realistic with the goals we're trying to achieve. You know, so we can build the use cases and sort of architecture and capabilities around it.

Lycett Christopher: OK, great. Monica Paolini, I know you didn't have loads, but is there anything you want to add on that?

Monica Paolini: Yeah. So all of these are really great changes that we should expect from 6G. So there is a lot of work ahead of us. What I noticed in the last months and maybe there is a little bit of less discussion, so there's still more than I think it's worth it about. Do we need 6G at all? And I actually, in a way, I don't even understand the question, because clearly you're going to have 6G. A new generation of sorts. We're not going to have 5G for an extra 100 years, so we're going to have something. The question is how we want to go about it. That is the better way of, you know, what is that we should do. And so it's not a question whether do we need it or not, it's how to get most out of it. And so as Ruth was saying, we need to be realistic. So we need to say, OK, this is what we want. What is that we can get in a cost-effective way? In a way that operators can to create a technology that operators can actually use in their commercial networks. So I think we need to distinguish the thing of what we expect in an ideal world, 6G to do, which is probably not going to happen, certainly not going to happen. And then to try to see what is more important for us. And I think I wish there would be more discussion about what is that is important for us, for instance, indoor coverage, which is, you know, and just an example, and then create priorities around it so that we avoid the kind of hype that we had with 5G, you know, in creating all these expectations that then are not met. Well, yes, because your expectations were too high.

Lycett Christopher: Great. And then I think a follow-up question is you both referenced maybe a bit around the industries and enterprises. That was talked about a lot, I think after 5G was launched. So is, I guess, how would telcos change to go and meet those needs? Is it really at this stage, more they're going to go and try and build 6G to go and meet some of those needs? What's different this time around that?

Ruth Brown: So I think it...

Monica Paolini: Nothing really.

Ruth Brown: Yeah, I, yeah. I was going to say, so like Monica Paolini says, it's a, it's a hard...

Monica Paolini: Sorry, go ahead.

Ruth Brown: Challenge in many ways, trying to kind of incorporate industry from the start. When you're building a new mobile generation, but one of the major differences I've observed looking at the early workshop for 6G versus some of the previous generations is from the outset, from the sort of kick-off workshop, I suppose there is more industries involved, so you've seen several of the especially like automotive industries, some of the aeronautical type industries that and I think some of the 5G organizations are really trying to include these. How successful this is going to be is another kind of opinion. I mean, it's very difficult, I think, until people see tangible use cases to really get involved and start pushing requirements directly into the standards. So maybe it's a case of more collaborative working, as these organizations are trying to build.

Monica Paolini: Yeah. What I would have to, I would have to add something that, you know, so 6G is is going to be a 3GPP kind of thing, right? And traditionally, for obvious reasons, 3GPP is operators and vendors. So the enterprise is not really represented and it's really hard to do it, not because they don't want to, but because the enterprise is, so there's so many of them. It's so fragmented. It's everybody. And for, for enterprise to have a presence at 3GPP is is actually expensive. So it's really like an ecosystem and collaboration issue at some point and and so there is a question of how to deal with that. The other thing is that Wi-Fi is actually doing a lot of the work that 5G, 6G can do in an enterprise. And it's more more complex. So I think that we should think about of having 6G or 5G today even to be more sort of integrated with Wi-Fi. So it's sort of an expansion of Wi-Fi. So for an enterprise, it'd be much easier to sort of conceptualize and deploy a technologies as an extension of what they already have, rather than another network which is completely separate, so I think that that would be that might, might help.

Lycett Christopher: OK. And I think a final follow-up question that is so you spoke about Wi-Fi, Monica Paolini, we've spoken about satellites or LEOS or NTNS before. Is it an universally accepted view from all? I guess access technology stakeholders that 6G can be something which incorporates Wi-Fi? Ran satellite and maybe some other access technologies where relevant? Or are there practical objections to that? Or this is just individual companies, individual industries maybe pushing to be a part of 6G. What's your take on all of that?

Monica Paolini: Well, I think that, OK, Wi-Fi is like, it's always, well actually you want to keep Wi-Fi. Separate from 6G to seven because they are different. They they just they both provide their own advantages, but they're different. So there is an advantage of having multiple technologies, so you want to be integrated but different. Now satellite that's driving a lot of work going on and I expect to continue. One thing that I'm really...

Lycett Christopher: OK.

Monica Paolini: Excited about is sensing and that's going to be more 6G I think than 5G because sensing basically uses the signal to that is already there and it's available to really understand what's going on. So instead of having a camera, you can use sensing to figure out where things are and what they do. And the thing that's that's really, personally I'm, I'm, I find it very interesting that there could be a lot of use cases that do not. Basically because you use the existing infrastructure. They are low cost. Well, they're cost-effective to deploy.

Lycett Christopher: Great. Evan, was there anything else you wanted to add to that, Ruth?

Ruth Brown: Yeah. So I'd say the whole convergence debate really, which is what we're talking about. I guess we've been trying it for a number of generations. What I think is probably different in 6G is there's very much a focus from the start. So from previous generations, a lot of the convergence piece didn't really work because it was quite a complex these complex architectures to achieve it.

Lycett Christopher: Mm hmm.

Ruth Brown: You know, you had to buy or pay for additional functions, device support, etcetera. I think as we get into 5G, you know, things like satellite technology is really evolved. So we've got these kind of much more sophisticated satellite NTN systems. I got higher throughput, you know, lower latency, Wi-Fi. Obviously Monica Paolini's mentioned is kind of slightly different, but there's a massive indoor coverage already. A lot of it is kind of within the enterprise space, but I think bringing those together is really going to achieve and help us get around some of these coverage gaps. I mean how you implement it? I think it's got to be a kind of a lower, more simpler architecture than we've tried in the past. You know, we've got some schemes already that you know with Wi-Fi, like open roaming, which are trying to achieve this kind of roaming from like wireless networks to mobile networks. But it's really trying to bring these two or three ecosystems together. Is is probably more challenging than just saying yes. We're going to have, you know, ubiquitous coverage or we're going to be a converged network in 6G.

Lycett Christopher: OK, great. I think that's it then for 6G. So we'll move on to AI and I'll just message Emma to let her know you can hear. OK. OK, right.

Emma Buckland: I'm here. I'm here.

Lycett Christopher: So moving on to AI then, I mean, back when we spoke previously and not it was not long after MWC, there's a lot of talk about AI ran agentic AI. Maybe. Maybe it was kind of pushed by NVIDIA a little bit. Has that topic moved on in the last few months. Has there been anything notable or is it as we were?


Emma Buckland: So specificity for AI ran, you mean?

Lycett Christopher: Yeah. Or, to be honest, AI from mobile networks, augmented AI, I guess AI for the mobile network in whatever capacity.

Emma Buckland: So the discussion of AI, so I think it's AI for the RAN. So where AI solution is bringing improvement to the performance of the run, which is encapsulated by the RAM concept. It's stereotopical. So it's still a topical discussion, but it's been around for years, so that has not receded and it remains very relevant and perhaps it has a new Monica Paolini.

Lycett Christopher: OK.

Emma Buckland: Which is AI ran, but it's nothing new. I think the bit which is AI enran where the infrastructure would be shared at the same site. Between the ran workloads and the customer workloads, which on paper sounds like a great idea. Still baffles a lot of people in the industry, I think, including an analyst. There's a number of questions as to why would anyone want to do that. So why would an operator buy GPU, which are really expensive? Put them on their cell sites to run internal workloads, which don't necessarily need GPUS really. And whether with an enterprise customer want to put their workloads in a sense side that just the simple question of of physical location, you know with enterprise be near that sense site for it to work. So. Apart from. An operator. In Japan, whose name is escaping me. I think it's either sobank or entity that has a specific offering called a trust. I think that is about AI ran. I haven't really heard an awful lot of take up certain in your life deployment of it. And I remember the CTO of a Tier 1 American operator saying that there was no need for GPU at the same site as of now. So it's still something that we are, you know looking at with interest.

Monica Paolini: Yep, Yep.

Emma Buckland: But commercially and technically, it's not really clear how it would work.

Monica Paolini: I think that you're absolutely right. You know, if you look at what what's going on right now, I think that the whole AI thing, we need to really think about it in, in the long term, because it's AI, it's automation. So you're basically giving up control and there is an opportunity, but it is also a risk that you screw up things and then things don't work very well. So you need to be very careful. So I think that we need it is important, but it's more of a longer, a longer term. Play and you know, as Emma said, it's like we don't know yet exactly how is it going to work. I think it's going to be very important, but it's very hard to say. It's going to be this thing. This thing is going to be that, you know, it's have that impact and I'll quickly. You can do it, but said that you know, it's like, you know, whatever you do right now, you can. You can see operators starting to use a little bit of AI, and then it's going to be more, but it's going to be, I think, more gradual. One thing where I'm actually less optimistic about is when operator. Start sort of have this plan of use offering AI services for your subscribers and not for customer service. For customer service fine. But to just become like a an AI portal or whatever and that you know, it's kind of like when you know they used to way back. When they saw Skype coming up and they just did the same thing, thinking that it could be better or an alternative to Skype and it didn't work. And I'm worried that we're going into the same thing. There are so many AI tools that, you know, in fact, it's hard to get to get rid of them. You know, everybody offers you the AI agent, the AII helper or, you know, the AI app. And and so I don't see operators really making money out of it or being really successful. Not that I'm so I think that they should concentrate on using AI for their network, for customer service, for their own thing, rather than to offer. Services that are not tied to to their networks.

Ruth Brown: I think probably just to add to Emma and Monica Paolini's points and you know, brilliant, brilliant points. You both brought up that, I suppose the two main sides of AI aren't there. Are can you make money? Which, you know, Monica Paolini's kind of spoken quite a lot about. And how can you be efficient or save money in your network? I suppose I've slightly biased from the sort of. How can you save money? Because I've come from a sort of an engineering service provider background and I can see lots of advantages. Options for things like this kind of copilot generation of code activities, and we're seeing use cases around that within Pox. You know, using it for recourse analysis and some of the digital twin activity. Like we said, there's the operators are still fearful or, you know, they're apprehensive about, you know, this control thing, you know, partly because of, you know, you know, giving AI the keys to the whole of the network. And sort of keeping that, you know, a solid foundation for use and and in terms of security, but also there's that fear factor, I think a little bit of the more you increase with AI and it does the thinking is that going to start eroding the skill set? Within your business. So I think that's also a little bit of the sort of thinking. But there's, I mean there's certainly a lot of opportunities, but I think there's some way to go in terms of achieving those? It's kind of a long-term plan.

Lycett Christopher: Right. I think there wasn't actually really much else I was going to ask around AOI unless there's any other topics or points that you think we should address or you'd like to share around AI.

Emma Buckland: I think one comment I have is that often operators talk a lot about using AI as a revenue generation and it can in a small extent in a certain extent help with revenue generation in the sense that it will improve the way in which the network perform. And it will help the operators to monetize their existing resources more quickly. So there's an efficiency thing which has benefits in terms of quality of network and then for customer retention. But so I can see why the operators want to see AI as a revenue generation thing. But by and large, the moment it remains, the cost reduction and optimization. Of the use of the network and the resources exercise. And and it's fine because at the end of the day, your profit is your revenue minus your cost. So if you bring down your cost, it's still good for profit. It might not have an effect on revenue, but it still is still good for profit, and that's what that's profit that keeps company companies, you know in the in the in the black so. I think the difference between AI for cost reduction and AI for revenue generation is cool to make.

Lycett Christopher: OK, great.

Monica Paolini: Yeah, I agree. It's like we should think about AI as being making the networks more efficient and therefore increased profitability rather than increasing revenues because AI is like, yeah, it's like, you know, it's like 5G in a way. You don't sell 5G, nobody, nobody's bought 5G, nobody...

Lycett Christopher: Yeah.

Monica Paolini: 'S gonna buy AI. They're gonna. You know you're gonna make your network more efficient and therefore you can serve more customers you can. Do better things. But you know, it's not something you sell on its own. Or I haven't seen any convincing case that they can do that.

Lycett Christopher: Yep. OK, great. Well, I think we can move on to APIs then. So my my first question really around APs is. Probably what's the biggest challenge that telcos are facing right now with APs and with I guess where they are in that process, what's what's the biggest obstacle that's in their way that they probably need to resolve?

Emma Buckland: So again start if you want.

Lycett Christopher: Yep.

Emma Buckland: So obviously it depends a little bit at which operators you're looking at. So the largest ones. Will usually have a fairly advanced program on implementing. What we call the Open API, so the APIs that follow the same standards whether that API is implemented in operator A or operator B.

Lycett Christopher: You know.

Emma Buckland: So typically it's the API that are pushed by the camera or the GSM Open gateway. Initiatives. So those operators will have implemented a number of APIs. I mean usually it's a handful on the open basis because obviously operators have had this directly. Dozens of APIs in the catalog. That's nothing new, but they are property those ones. So I'm not talking about the property one, I'm talking about the open ones.

Lycett Christopher: Yeah.

Emma Buckland: So they will have implemented the Tier 1 operators. A handful of those APIs. Usually it's to do with security. So it's SIM swap or number verify or that sort of thing. So the the challenge for them is perhaps not so much technical, but now it's making money out of it.

Lycett Christopher: All right.

Emma Buckland: So it's commercializing it, so it's channel to market. Choosing the right pricing model and so on. Then there are less advanced telcos, which are smaller, which obviously don't necessarily have the deep pockets to make investment into implementing the open APIs into their networks because the business case isn't necessarily clear. And they might be looking at what the. The area movers are doing without necessarily doing anything themselves. That's why really the large initiative that you see like the open gateway, if you see actually who has actually launched those open Gateway APIs it tends to be the bigger the bigger operators and because in some count. For those APIs it makes sense that all operators in the country. Have them valuable and often you have a cluster of operators who in one country all have all launched SIM swap for instance. So for the love operators, I think the challenge is more is less technical now and more on the commercial. Nature, and obviously for the smaller telcos that haven't really started there still remains a bit of a technical. Challenge in actually implementing the APIs in their networks now. You don't implement APIs unless you have a network of a certain HS, so there are certain APIs that can run on a 4G network. You don't need 5G and don't even need 5G standalone. But for certain APIs, in particular the ones that we call network performance APIs, so they don't really allow the 10 core, they don't only allow the 10 core to exchange information about the network, they also allow the developers not just get that information, but. Have action on the network. So decide what sort of of. Capacity they might get on bandwidth, that sort of thing. And for that, Union networks that are of a certain calibre. So typically in mobile we need to find GSA. And you will need to look back on the correct question that we we had. We will discuss before you need a certain level of autonomy because you're not going to adapt the network performance of your network when an application requires by hand. It needs to be something that is done in an autonomous manner. So for the high level APIs the one. That we call network performance APIs for instance. You need. There is a technical challenge even for the Tier 1 operators to have. In mobile, a network which is five GSA for instance.

Monica Paolini: May I say something that is on top of what Emma said that you know, different operators have different, you know, different face, different challenges at different. Phases, and I think that that kind of creates a meta type of challenge, because if I'm an operator and I have all the APIs out there perfect it, it doesn't really work until all the others have it too. So for instance, in Brazil APIs are used a lot because all operators have found a way before the even the the Open Gateway initiative to work together. And obviously that is hard because every operator has its own pace and but it's really because the value of an A. For an app developer is tied to the fact that you can use it across the network, so for instance. A lot of APIs. Are available from, say Google and Apple. And those are clearly not as they don't have the same functionality as the one that an operator does, but they're really cheap. It's very well understood how to pay for them, how to use them, and it's only two providers because pretty much all the phones are either Android or Apple. So if I'm an app developer, especially a small one, I just have a deal with both of them and I'm done. So you can see that and those are widely used. Everybody uses them so we can see how we know how that works. But how? Why does it work? Because Google and Apple APIs are different. But it's only 2 1/2, so I think that we need to kind of start think more about in those terms because the APIs are going to be are mostly directed at. App developers so that that's a challenge. But also an opportunity for operators to to finally learn to talk and work with app developers in a more. Sort of in in a way that is more valuable to to the app developers so that they are willing to pay.

Lycett Christopher: And all this. I'm assuming there's steps to make that happen and and are. Are they working towards that? And is it something that you see as inevitably happening? How do you see that? That, I guess evolving over going forwards.

Monica Paolini: Well, it might happen, or it might not. I mean that's that. I guess that's a challenge because so the the GSMA and Kamara are doing excellent work in trying to provide the foundation that sort of the plumbing for everybody to do the same thing.

Lycett Christopher: Yeah.

Monica Paolini: So it's like, you know, you have the plumbing, so everybody can do the same thing and well, they can create APIs that have the same functionality so that they can work together. So instead of having the situation where you have Google and Apple. With their own API structure which is different, they would have the same. Everybody will be used in the same API structure, so it is a huge value there. The question is will operator I mean because operators are moving at a different pace or they might not go there at all. That that is kind of slowing down the process because. You need to have everybody to do the same. So like now for instance you can have. In any country you might have one operator doing one API and the other one doing another API. Well, then we have a problem then. Then it's also across countries because let's say that I am an Uber, OK? And I want to use the APIs. I want to use the same APIs ideally through every country. Maybe not every country, but most countries, or at least the countries where I have more business. Because so if I'm the app developer. Uber I I need to spend time to integrate the APIs into whatever I do into my app and that's expensive. So I do not want to keep doing it for each operator.

Lycett Christopher: Mm hmm.

Monica Paolini: For each country and I would like to have all every all the users, you know, covered. So that's kind of a chicken and egg situation, but we we need to get there. There is a lot of operators are committed to it. But clearly as Emma was saying, everybody's moving at a different pace.

Lycett Christopher: Right. And you mentioned Kamara and Open Gateway, are they, are they complementary standards working on on slightly different things and then ultimately having both is a good thing for the industry. Or are they competing?

Monica Paolini: No, they are. They're working together. They're just like you didn't think of 1 without the other. OK, OK.

Lycett Christopher: Oh, OK.

Monica Paolini: GSMA does other things out of APIs obviously, but you know the whole initiative is is really like it's it's a tight thing because so GSMA it works for the operators to what what are the?

Lycett Christopher: Yeah.

Monica Paolini: APIs what should they do? How to, you know, sort of like from a standardization point of view, but then Kamara does the actual, you know?

Lycett Christopher: Add.

Monica Paolini: Hosting them, helping everything to come together. So you know that's it's it's really complementary.

Lycett Christopher: Correct. Emma, was there anything else you wanted to add around? APIs or any pointers you want to put forward?

Emma Buckland: Well, I think I might have mentioned before I think. Question we we are asked by clients or we hear is? Web price through an API hub. You know the price point is actually quite key because this is a new service and like every new service is difficult to gauge. What is the appetite for the markets to actually pay for it? And the risk is obviously if you price it too high, nobody will take it any further. Price it to lower your leading value on the table as the as the telcos. That's not, it's a. It's it's a. It's a nice question to ask themselves, I suppose, because it means that technically they're able to to offer them, but it's. It's not easy to solve. And I think. One thing that Somet telcos are probably aware that it's something that maybe we should discuss a little bit more as an industries that one API. Will in some cases be critical to the use case and in some other cases it would just be nice to have. So for instance, if you think of quality on demand and you're using it probably with other APIs to fly in a drone, but quality on demand. Needs to be bang on for the drone to actually fly properly between the two points. If you're using quality on demand for a conference call like the one we're having, it would be nice to have a good quality of call, but nobody's going to die if we. Don't have a good quality of call so. You would see a rationale for in one case. Setting the same API, the quality on demand more expensively for the drone use case, as you would for. For the enterprise case, which would be a conference call? That's a nice idea on paper, but of course practically it means that you can actually segment the SEM API in your BSS for the different type of user. Different type of use case, and similarly you might want to charge an API more or less depending on how busy. Your network is and how in demand that particular API is at that particular point in time, and if you were able to do that as a telco, you would maximize the value. But obviously whether Tedco can actually implement that in their network, mainly because of supporting system, not just because of the capability of their network.

Lycett Christopher: And it would be a could be a potentially customized pricing model as well. So whether it's like premium or whether it's cost straight up, are they? Is that how you see it?

Emma Buckland: But you don't really want the freemium that is free to talk much other way, that there will be no value creation. That's not sure, but you can imagine a context where the API is free for very small volume, so that the developers can play with it. And after that, they obviously they will want you will want them to pay for it. So we don't see a premium model where in that context where you get the API free just to in the same pate sort of thing.

Monica Paolini: There might be cases where you might have a a not. Not really a freemium, but basically free APIs because there might be. There are APIs where it's the operator that benefits from it. So for instance, let's say that you use an API to manage traffic the traffic. So sorry, video compression for instance. This would allow the operator, well, the app developer to get a better service, but also. So the operator to be more efficient in the use of the infrastructure to manage congestion and things like that. So you can see a situation where you use the API to basically. Have a benefit. So again, I guess it's like you don't get revenue from the API directly, but you increase profitability because your network is is more efficient. So there is this but but that's kind of a separate question. Now obviously as I say like it's very. Hard to to understand how to price them so you know if if you look at Google Now they just price is like a flat fee basically. Same same with Apple. Well, not flat fee, but it's like a volume based, but even the volume is might not really work because the pans are really what kind of how important it is to you how valuable. So I think that we still need to think about it and. There are a new new players that are emerging. To basically help the app developers to customize the APIs to their needs. And so and also to deal with diversity. So if have operator A and operator B that have the same API, but they're not quite exactly the same, you know they they can, they can help. So there is a whole different set of players because you know you can go and buy. You can buy now T-Mobile. AP is on Azure's marketplace. You can go there and buy them, but. How can you really use it in a suspected? Some will still need some help.

Lycett Christopher: OK, great. I think that was it then from my side, Ruth, I know you have want to contribute much to this. I know if there's anything you did want to add or not.

Ruth Brown: No, I I mean I know them or I kind of read them at a high level, but I'm not in the same depth that Emma and Monica Paolini are. That's interesting to listen to. Thank you.

Lycett Christopher: OK, great. So unless anyone has anything else, and I think we can, we can leave it there.

Emma Buckland: OK, Chris. Thank you.