Gabriel Brown, Omdia (00:00:11):
Hello and good morning, everybody. Welcome to the 2026 ORAN Alliance Summit here at MWC. It is fabulous to be here with you all to talk about the state of the art in open radio access networks. Thank you all for coming. And welcome everybody also on the live stream. My name is Gabriel Brown. I'm an analyst with Omdia. We're an independent tech industry research firm. I work in our telecom and networking division, and I'm delighted to be here as your moderator today. And also where needed, we won't need it a lot, but as an independent voice from the stage.

(00:00:51):
The event today is intended, of course, for mobile operators, but also for the industry technology and ecosystem players helping to build the world It's best open mobile networks. It's something like eight years since the RN Alliance was formed. And from my point of view, as an outsider, as an industry analyst, and we track this closely at Omdeer, I think Open RAN is moving into a new phase, one focused on productivity and bringing value into real world networks. Certainly that's going to be a lot of work. Not all of it glamorous, probably not much of it in fact. And walking in, I thought, well, the purple lights and the DJs as you come through is about as glamorous as this is going to get. But I do think as well, focused and diligent work in ORAN is really going to be the key to, first of all, driving the scale that we need, but also the staying power to see us through the next few years and into the next generation and all the exciting applications for advanced connectivity.

(00:01:56):
So I also noticed that we have, when I looked across the agenda, the public agenda for the conference sessions at MWC, this is probably the leading RAN discussion forum on the public agenda. Of course, there are lots of vendor booths, radios, antennas, and so forth, but I think that tells its own story as well, a little bit about how ORAN has become so influential. We have a terrific agenda for you today. It is compact, but also very full of insight and contribution and comment from industry leaders building and innovating in mobile networks. You can see here we have the agenda. We joined by my colleague, Ruth from UMDA, who's going to take us through the automation panel later this morning. So with that, let's get right to it. I'd like to introduce our first speaker of today. We're going to start with the state of Orang keynote from the chair of the board of the ORN Alliance, Thomas Lips.

(00:02:55):
Thomas, please join us.

Thomas Lips, Deutsche Telekom (00:03:01):
Thank you, Gabriel. A very warm good morning into the round. I hope you all had a good day one, inspiring day one. Now day two, let's kick off our Oran Alliance Summit here. Before I go into the state of nation, as I would like to call it on ORAN itself, let me start with the personal note. After quite some operational roles as CTO, I was joining this journey in Deutsche Telekom in 2022, mid of 2022. Ever since we are strategically moving forward, transforming ourselves means getting all the skills, understanding cloud and run, and really getting our hands dirty on integration, plus transforming our networks. Now standing in front of you here as chair of ORN Alliance, I must say I'm extremely honored, but I'm also extremely committed to drive this transformation of our industry and bring our Oran Alliance and its mission to the next level.

(00:04:29):
Maybe the first slide is disappointing you already a bit because our mission has not changed since 2018. What we are aiming for, all the interoperability, virtualization, intelligence, and the openness overall that we are driving is a mission statement that is super stable and that drives the RAN transformation for all of us. However, a mission is nothing. It's an intent without execution. And execution is the key. What you see, adoption across the world. Nobody doubted in the beginning that it will be easy for greenfield operators to go full blown on ORAM, but everybody was saying, wow, how those big networks, big customer bases, high quality demands, how they are transforming. And I can tell you, we all figured it out because it is not a one shot in all. Everybody is adapting step by step. Some have started with the cloudification journey. All cloud automation, commercial of the shelf hardware, employing this one, decoupling hardware from software.

(00:06:00):
Others have started to go open frontal, making understood how an integration works, how to bring different suppliers together. Really important one, but not everybody. Others again, focusing on the intelligence journey, intelligence automation. So all SMO, Rick, and all those elements. There is no one fits all, but we are all on a journey in implementing this across the world and adoption is really happening at scale, I must say.

(00:06:38):
Now let's look a bit what we have done in 2025 and what looks like number 147. I can tell you the number of specifications written, adopted is a clear fundament for adoption. And it was super important getting forward with more details on our specification, better fine-tuning and enriching what is there. But this is one element. The second element that is get adopted by regional organizations standardization bodies. We talked about Etsy last year. We see the same now from the American side with ADIS and initial small steps are happening now as well on the TTA side, on the Asian side.

(00:07:40):
Maybe one element more to mention on that slide, and that is our ORAN software community. We were starting it with reference implementations, giving it the base that people can work with that one and can enhance it. We finally have now decided to give it to get it into the next step into Linux Foundation. This important one, because this is the biggest open source community we have, and this is in itself a very big driven community that will help to develop this one further. Now, a quick one on this one, because we are very driven from our operator side, the needs from operator. But of course, implementation comes from all angles, comes from all contributors, suppliers. We are getting extreme good feedback on what is important to be fixed, what is important to be enriched. And normally what comes last, security, I want to mention first because this is an element that is important to us.

(00:08:54):
Security is not an afterthought. We have it embedded and we are driving it continuously.

(00:09:03):
Now, most of us have managed to integrate traditional radio units. I would say not a big challenge anymore. Once we have done it, once you have done two suppliers with each other, the next one is relatively easy. Important was now in 2025 to do more on the Massive Mimo part, especially now the ULPI stuff, really substantial enhancements on the test specs and on the profiles itself. Really important step because not many have done a massive MIMO integration yet, but I can tell you it's possible and we've seen it all in our industry now happen in real deployments. What I'm especially proud of is the SMO part.

(00:09:55):
For a second talking SDT, because we are developing it ourself, but from an overall or an alliance, the framework, this service-based architecture, all the APIs that are there enhancing, and this is really important, cloud orchestration and all the observability that is needed with this one, all the R1 interfaces for third party apps, really substantial progress has been achieved and you will see later here and there some quite good demos on that one. So is the work done? No, not at all. I'm not sure when we can say the work is done because implementations means you get feedback and these feedback loops is what keeps us busy and what keeps us moving forward and doing more.

(00:10:56):
I guess most of you that are doing implementations, you will see statements. One is not yet complete, O2 is not yet complete. So quite a nice job for the colleagues to do, enhancing this further in 2026. But if you look on the right side, 6G is coming and it's faster coming than most of us would admit. I can tell you, you will see early deployments, probably even in 28 in some operators might not be complete standard. So it was important for us to kick off this journey and having our first, this was done in February now, and having done our first items work items on 6G now going forward, going hand in hand with 3GPP and with that one really being open from day one.

(00:12:05):
So in summary, what you see, what you saw us in 25, what you see us doing in 26, we are focusing on the implementations that's ongoing. We are closing the gaps that are still there. We are enriching the specifications, but, and this is important, we are already looking ahead what is happening in 6G because this will be super important going forward. None of the things that we are doing would be possible if we are not having all the operators, all the contributors, and honestly, all the more than 1,600 experts that are working with us. You see when we are having the meetings, extreme active participation, we see more smiles now in the face because things are happening, things are getting done and at least from my side and maybe on the behalf of the community, many, many things to all those experts that are helping us to make ORAN happen, to make ORAN richer and to make ORAN easier, implementable.

(00:13:22):
With that, I'm ending my speech and I wish you a very good ORN Alliance Summit and of course MWC as successful as always. Thank you very much.

Gabriel Brown, Omdia (00:13:39):
Fabulous. Thank you, Thomas. And Thomas will be back with us for a fireside chat after the next session. Thomas, terrific introduction. I guess very nice that you mentioned your commitment at the start and closed on that message. Give us a feel for the level of commitment among your peers, partners, ecosystem to the ORAM mission.

Thomas Lips, Deutsche Telekom (00:14:04):
We will see later a lot of us talking about their implementations. And typically what you see right now, tier one operators, we are leading the track, we are doing it, we are implementing it. I think that's the thing that is happening and there is no doubt.What I would say is now is even the time for tier twos to start, to start with a gradual introduction and getting used to the different challenges, but the opportunities that are coming with that one. So therefore, I would say on a good track.

Gabriel Brown, Omdia (00:14:40):
Good. Fabulous. Thank you, Thomas. Thank you. We're now going to move to our first panel session today, and if I could invite the speakers to come to the stage, we have a terrific lineup. Please, speakers here, join us. As everyone is coming to the stage here, I will indeed get my notes up, get everyone's job title to do the introductions. So the session here is driving operator success. One of the things the RN Alliance want to do is to show how operators that have got some scale can contribute back to the ecosystem, help you with your own projects, and of course the vendors supporting them. So please gentlemen, take a seat. Yeah. Perfect there, Rob, I think. And join in Yari, you're good there. So thank you all for joining. Let me introduce everybody. Let's start at the end with Ari kind of slurty, SVP of strategy and technology and CTO at Nokia Mobile Networks.

(00:16:01):
Welcome, Ari.

Ari Kynäslahti, Nokia (00:16:02):
Thank you, Gabriel.

Gabriel Brown, Omdia (00:16:04):
We have Henrik Janssen, VP and head of North America Business Development in Samsung Networks. Welcome, Henrik.

Henrik Jansson, Samsung Networks (00:16:11):
Thank you, very much.

Gabriel Brown, Omdia (00:16:12):
Sada Yuki Abeta, Chief Open RAN Strategist at NTT and head of Orex, CTO of Orex. Thank you. Rob Sony, VP, RAN Technology at AT&T. Welcome, Rob.

Rob Soni, AT&T (00:16:25):
Thank you.

Gabriel Brown, Omdia (00:16:26):
And closest to me, closest to me, we have Sudakar Panda, head of RAN at Rakuten Mobile.

Sudhakar Pandey, Rakuten Mobile (00:16:32):
Thank

Gabriel Brown, Omdia (00:16:33):
You. Sudakar, welcome.

Sudhakar Pandey, Rakuten Mobile (00:16:35):
Thank you.

Gabriel Brown, Omdia (00:16:37):
I'm going to start with you actually. You are at Rakuten Mobile, probably, well, probably not even the word, the operator with the biggest multi-vendor. If we can say open RAN network worldwide, you've done a tremendous amount in terms of interoperability. Please tell us a bit about how you're getting on and what you've achieved so far.

Sudhakar Pandey, Rakuten Mobile (00:16:56):
Okay. So first of all, thank you, Gabrielle, and thanks to our esteam colleagues here. My name is Sadakar. I'm heading radio access network in Rakuten Mobile. In 2018, when we plan to launch as MNO, there was a choice whether we need to go with traditional RAN or we need to go with the open RAN. And I don't know, at that time it was not known whether it's good decision or bad, but we have chosen the other path. We have challenged the way the network was built traditionally like plan, build, and operate. And 2020, 25, five years, fast forward, I can proudly say the decision was right. We have built largest open RAN, fully virtualized cloud native network. In network, now we are running around 350,000 cells with different suppliers in ecosystem. So what we can say that we have not only built the ORN, but also build the ecosystem.

(00:17:58):
And in terms of if you're following Rakuten, last month, Rakuten has a group announced for 2025 financial results. And I'm proud to share that Rakuten Mobile has achieved annual EBITDA profitability within five years of launch. That means what it converts masses that it's transformative. We have taken a bold decision, which was not easy. It was tough journey, but finally it was rewarded. So it's transformative as well as sustainable. In December, what we have announced that we have got 10 million plus subscriber, which we targeted. And while we speak, we are somewhere around 10.2 million subscriber. So thanks to all the customers who believes on Rakuten mobile network and to the team that they've taken a path which transformation and innovation, and today we are able to set the example and lead the path in ORN.

Gabriel Brown, Omdia (00:18:59):
Thank you, Sudhkar. And 10 million subscribers is obviously a big achievement. 350,000 cells, you say. Tell us a bit about how much of that is traditional RRH versus do you have massive MIMO? Do you have multi-vendor, massive MIMO deployed?

Sudhakar Pandey, Rakuten Mobile (00:19:13):
Okay. So we started at 2018 or 2019. At that time, we were not on time, but before time. So standard was still getting up, it was getting defined. So we started with VREN journey and Nokia was the biggest partner and we are having 170,000 radios from Nokia, which has been deployed in 4G. But gradually we moved to ORN compliance radios. In terms of ecosystem, if you see, we are having more than 15 vendors which has been onboarded. You name it and we have in our ecosystem. In terms of deployment for, it's like start from 4G, 5G, indoor, outdoor, Femto sales, all type of deployments it's covered. You specifically talk about massive MIMO, and that was something people were very acceptable that I don't know whether it will work or not, but 80% of our deployment were the massive MIMO and it's deployed in the dense market.

(00:20:11):
And we offer people that we won't show in PPT, come to the Japan and experience by your own. So it's working on ground.

Gabriel Brown, Omdia (00:20:18):
Fabulous. Thank you, Sudika. Rob Sony at AC&T, give us an update on your journey. I mean, you spoke here last year. Please tell us how you're getting on.

Rob Soni, AT&T (00:20:28):
So first, thank you for inviting me to participate, Gabrielle, and thank you to the RAN Alliance also for encouraging our participation in these panels and these discussions. I think it's good to share best practices and share understanding of what you're doing and how far you're going. And we heard a lot from Thomas this morning about the challenges of Brownfield versus Greenfield. And we have Greenfield here and then we have Brownfield here. So we have a different perspective overall on how we are approaching ORAN and what we are doing overall, but we have significant progress to report compared to where we were say a year ago when I spoke about where we were. A lot of things were in planning and definition phase. We've continued to move forward beyond the planning and definition phase into execution. This is a big year for us in terms of execution for ORAN.

(00:21:16):
And you heard Thomas talk about really three different areas where operators could play and gain experience on ORAN and ORAN platforms. We are playing in all three. So that includes Cloud RAN, that includes SMO, and that includes mixing and matching radios and baseband. We're very excited today that we have in our network and are on track to deliver radios. And it's announced publicly with Onefinity and Erickson with an open front hall, ORAN compliant. It is an ON managed radio. So with hybrid mode, this allows us the greatest flexibility of bringing in new radio partners as we go forward. We also have a fully compliant SMO, which is now scaling actually this week, dramatically increasing the number of sites that we'll support and that SMO will support what we need specifically for data management for FCAPs, as it's typically called, but also specifically for the RAN intelligent controller.

(00:22:13):
So that RAN intelligent controller is also brought in, as well as finally cloud RAN orchestration. The intelligent controller, RAN intelligent controller gives us a new layer of programmability. When we envisioned that two or three years ago, people were just trying to figure out what AI was, but now we see that this is going to be the linchpin for essentially for us to enable and make the network AI capable, AI ready. In fact, we're already exploiting that today where our network is already leveraging and moving forward on a variety of different types of algorithms that will help us with optimization, with energy efficiency, with provisioning and management. And we're looking forward to see how we're going to fuse Agentic on top of that and what we'll do with that. But right now, we have traditional AI machine learning based algorithms that we have from our legacy platforms that are now porting, moving directly to SMO compliance.

(00:23:06):
And we're seeing those algorithms become more rich, more capable as we add AI capability. So this is a fantastic thing and uplift for us. We're also, as I mentioned, doing CloudRAN and our CloudRAN is truly ORAN compliant. So it does support, again, full ON management. It also has a fully open fronthaul supporting both massive MIMO as well as traditional radio. So this gives us an opportunity to scale and we are looking forward to moving more aggressively when Granite Rapids is available to us in the lab. We're doing some initial testing, trialing, and we will actually have commercial service. We do have commercial service running today on a number of Cloud RAN sites in the Southeast in the US. So a variety of sites are now up and running. Very excited about that as well. We also have third party applications running on the SMO. The SMO has become, again, a platform for us for innovation, a real opportunity.

(00:24:02):
What we also see though is as we mix and match hardware and software, we see the partners and the ecosystem evolving very rapidly to provide different types of capability for orchestration and for automation, for applications like dynamic sensing of spectrum and spectrum utilization. This was part of our bigger project. The bigger project for us was really about modernizing and refreshing our network and allowing us to exercise all the spectrum that we had purchased, and we just bought more. And this has been a fantastic impact overall for our subscribers, seeing significant improvements in what their actual user experience is. So it's not just about introducing programmability or operational costs, but it's also about touching user experience for us, and that's really the core of it.

Gabriel Brown, Omdia (00:24:48):
I love how you mentioned impact on subscribers there, because obviously ultimately that's the point of all this. Rob, would it be fair to say with some years now of experience with the ORAN architecture that there's no going back for you or in the balance?

Rob Soni, AT&T (00:25:03):
I don't know what going back looks like. So yes, we're very much in the space of going forward. I think that now that we have put in place those foundational pillars that allow us to bring in things, we see different opportunities in how we source and how we manage the network. I love the word brand transformation. This is a transformation and definitely appreciated Thomas mentioning this as a transformation in the way that you do things, affecting not just how you operate, how you architect, but finally even how you build and who you choose to work with.

Gabriel Brown, Omdia (00:25:38):
And if I may, Rob, one more question. What would you like to see more focus on, more attention is needed, I guess in your case or in the ecosystem more generally?

Rob Soni, AT&T (00:25:48):
I think we've been pretty clear with Urban Alliance. First of all, we're grateful for the specifications we have that have allowed us to build product that we think now can be supporting inner operation, but I think there's a continuous need to stay focused. We know that ORAN was originally organized as a project. I've heard that before. People didn't imagine an outcome that actually was almost like an SDO or standards driven. It's important to align working groups and their activities so that if you're actually operating across all of the stratums I mentioned, SMO, Cloud RAN and mix and match or open fronthaul, that these specifications are fully aligned and remain fully aligned and that release packages are actually disciplined and focused. I think there's probably a strong need to graduate groups so that in the end, the groups that we need to move forward in those areas where ORAN is having a direct impact, have the resources they need.

(00:26:45):
We all know that standards organizations are typically resource constrained. It is an expense for all of us, and if you can't easily trace it back to revenue, it can be challenging to source. So having the people in the right place to support and make those decisions, I know these are hard decisions for the board to make, hard decisions to execute, but I think it's time to stay very, very focused and not miss the pin. So if there's opportunities to continue to focus and stay focused, it's easy to get lost in the swirl. This conference, there's a lot of noise about 6G and about AI. I think 6G will come, but what we're really looking forward to is that what has ORAN has given us the opportunity, and it's an opportunity. It really depends on whether that opportunity is actually exercised. It's given us an opportunity to regularly roll in new spectrum with innovative partners.

(00:27:40):
It's given us an opportunity to densify the network with new partners. It's given us an opportunity to exercise the spectrum we have. It's given us an opportunity to bring AI today, not wait for 6G. So truly moving the network forward towards AI native, stay focused on those goals. Yes, 6G will come. And yes, ORAN should play a role in that. But finally, I think it's very, very important that we don't lose sight of the fact that what we are here for. What pays finally the bills? It's 5G, and we think Open is an important tool in the evolution specifically of 5G.

Gabriel Brown, Omdia (00:28:16):
Okay, fabulous. Thank you, Rob. We're going to move now to a vendor view with Henrik, then Ari, and then we're going to move back to Sadiuki, who's kind of combines the two a little bit. So Henrik at Samsung, at Omdeer, we obviously track the RAM market very closely. You're clearly, when you talk about open VRAN and particularly on VRAN, the largest market share. Tell us a bit about some of the learnings you've had from your deployments to date.

Henrik Jansson, Samsung Networks (00:28:47):
Right. Thank you, Gabriel. And also thank you to Dakar and Rob for sharing your experiences. So at Samsung, we've had a privilege of working with tier one operators globally on some of the most demanding networks. Verizon, obviously, Telus, Vodafone and as well as KDDI. And this journey really from early trial deployments to deployments of tens of thousands of VDUs has been successful. We've learned a lot, and I'd like to share three key learnings from deploying ORAN at scale. Now, the first one is that when you virtualize RON at meaningful scale, you don't get credit for openness if your reliability and performance is even slightly unpredictable or inferior to the traditional RON. So we have engineered our solution to be cloud native from design and to continue to take advantage of new compute platforms as they become available. And feature parity for 5D, and I would say as well as for LT is not the nice to have or something we put years later in the roadmap, it's mandatory to deploy at scale.

(00:30:11):
The second learning is that automation is no longer optional. So when you deploy a hundred sites, you can manage complexity manually, but when you deploy 10,000, you really can't. So the deployment, the lifecycle management, the optimization, the energy efficiency, it all needs to be software driven. Zero touch provisioning is key. We have also proven our AI-based energy management system in Verizon's network now, and we're working with tellers to deploy our AI-based streak in their network. The third learning is about ecosystem economics and open interfaces, and we've heard about this today, they are ascension They provide a lot of value, but they also risk increasing the costs. To handle that and to drive down cost, repeatable integration is needed throughout the life cycle. And many operators today are considering to deploy Open RAN with a single vendor or with a very limited set of vendors as a start.

(00:31:27):
We think that's a quite pragmatic approach. It makes it possible for you to build an open foundation while still limiting the integration complexity. And while you're there, you can start to add multi-vendor, but of course, only as long as you can make the integration predictable and efficient. So that's why we work very closely with, we have over 30 ecosystem partners where we do IoT, we do pre-validation and regression testing before anything goes out in the field. So in terms of how these experiences have changed our R&D and how we view that, I would say that for us, cloud native or on compliance automation hooks is baseline. It's not add-ons anymore. And we take a very industrial approach for how we work with ecosystem partners for integration.

Gabriel Brown, Omdia (00:32:24):
Thank you, Henrik. And obviously having a baseline is a fabulous place to be. It's an achievement, I think, for the ORAN mission in many ways. Ariet, Nokia, let's hear from you. Obviously Nokia has been involved since day zero or minus zero, I think. Give us a little bit of perspective from your side.

Ari Kynäslahti, Nokia (00:32:44):
Right. Indeed. We have been there since the very beginning, Rob. You were with us those days and I better say a different role. I remember you from early days very well. We have really built it into our product so that it's in every product. For us, ORN is not only CloudRun, it's also in our purpose-built product. It's very much about the open front hall, but thanks Thomas for reminding us with your keynote speeds. It's also very much about the Rick SMO. Rob, you also talked about it. So fantastic new network element introduced by Oren Alliance. Let's not forget that one either. So we have proven it. We have a network like mentioned, Rakuten, Tocomo, T-Mobile, Germany, or the telecom. So those are very well known public ones. We have many others. Typically, our customers approach this as like a basic requirement. Many of them don't actually want to even mix and match vendors A and B over the front hole, but it's the insurance policy that you need to be all on compliant.

(00:34:02):
Almost everybody asks for it. So that's why for us, it's like every product, there is no separate product line for something else and then the ORN. It's all ORAN. And then we have been able to also mix with our product third party providers in a commercial network. We have many cases like this and fewer public sort of combined Georgia Telecom. And there we are open to embrace the ORN ecosystem and always look at new partners to join. We often work like Pujito is one of the public ones, but then we also use it for our benefit in another way. So sometimes there might be some RF variant which is very specific for a certain use case, for a certain market. And then ORN Alliance helps us a lot. So we can hand over the paper, the ORN Alliance specification for the front hall. And before us even doing it, they probably have read it and worked against it.

(00:35:21):
So the ecosystem, I really can see it helping the industry. And we have these kind of examples as well, which are typically not public, but for us it's like fantastic opportunity to pronoun the portfolio with the partner.

Gabriel Brown, Omdia (00:35:37):
Sure. Your SVP of R&D, if I've got that right, and CTO of the mobile network division there at Nokia. Tell us a bit about how the ORAN architecture as a baseline helps you in your future development and what comes next in product and technology.

Ari Kynäslahti, Nokia (00:35:55):
Right. So indeed this kind of a opportunity to get companies to help us with the portfolio, as I mentioned, is one thing. Something we recently announced is openness on our RIC side. So RIC for us is part of the SMO, Montreal SMO product, and we introduced the open interfaces on that one. And then the R apps towards the third party community, getting through the RAPS interface deep into our radio access network kind of data, us exposing it to the third party community, I believe will be the big next step. So there we will be seeing like an ecosystem around something completely new we don't even know about when we expose all that data for the third party, even smaller companies. So this is for me the big next step.

Gabriel Brown, Omdia (00:36:57):
Sure. Okay. And we had the rare thing I think on Sunday with a joint statement from Nokia and Erickson to interoperate on SMO and Rick and so forth. So we're going to study Yuki better now. Just before we hear from him, just we're going to have some time for audience questions right after that. So if you do have a question, we're going to have microphones coming around. Raise your hand, say your name and affiliation and so forth, and we'll get to some of that as well. But Sadiuki, your reflections.

Sadayuki Abeta, NTT DOCOMO (00:37:27):
Yes. First of all, thank you very much me. And I'm now sitting between that. So one phase is actually NT documents operators. The other phase is that Orex is at the vendors. So the reason I guess I sit in between in between. So as for the operator side first, the NT document already deploys much vendor network from the beginning of the 5G. Even the 4G area, we introduced much vendor network together with the partners. And 5G from the 2010, or even the 2019, we use the open frontal interface or alliance and developed the ORM network nationwidely. So then what actually NT document did, we tried to expand the oral network, not only in Japan, but with outside Japan. To realize that we created our ecosystem working together with the partners because as I already saw the Samsung and also the Nokia mentioned ecosystem is very important to reduce cost, to this time.

(00:38:48):
And so then we should have some kind of reference model if then we can easily move to that next step according to the requirement of the different operators. So we first created the open ecosystem together with the partners and those interpreter tests with the partners and also making some difference model. So the solution, pretty integrated solutions, so including that radio server cars and also the software. So then we bring that solution to not only the Japanese operator, but also the outside Japan. So the big operator like AT&T, they have big power and also that they can do the integration with DKRs integrate to their partners. But as Thomas mentioned, the tier two or tier three operator do not have the capability to do it by themselves for them and pre-integrated solutions. So that case, they don't need to do the integration by themselves. So pre-integration solution is very easy to introduce the opener in their network.

(00:40:13):
So we bring that pre-integrated solution to that country. And also we have already discussed some of the SMO. SMO very important to reduce that lifecycle management because we need to upgrade the software, hardware, cash rayer. So then if we do have some automatic operation, that is very costly. So SMO is very important and our SMO, of course we supported our alliance, but we are also targeting to support not only the OM vendor, but also that targeting that incumbent vendor in the future. At this moment, not yet, but if we can do that, more easy to manage that ORM vendor and also the other vendor that is easy to deploy even though the brand field operators. Sure.

Gabriel Brown, Omdia (00:41:11):
Okay. Thank you, Saruki. So do we have any audience questions? Just raise your hand. Anyone taking their glasses off or shrugging off their jacket or having a little stretch is going to get called up at this point. Oh, we have a gentleman here with his hand raised. Just three rows back on this side, mic person. Mic person has gone missing. Okay. Oh no, here we come. Veronica, we have a gentleman here, third row from the ... We have a roaming mic.

Rob Soni, AT&T (00:41:48):
You could just shout. Yeah. If you stand up, we can hear you. If you stand up, we can hear you.

Juan Vertis Corredondo - Audience member (00:42:01):
Hello. Good morning all. I'm Juan Vertis Corredondo. I'm now working as independent advisor and I have a question for the panel in relation to the first point, Henrik, that you made in your observations about the critical points for open round deployment, which was feature parity and performance position. I'm interested in understanding whether any of you have visibility of any performance or capacity benchmarking results, particularly for massive MIMO open run deployments compared with traditional run. We have all the spectrum TDD where it's essential to have the best kind of performance there. And to enable the scale is very important to have visibility of those results and it's difficult to get that information. So maybe any of you could help us understand where the position is in relation to that or what the roadmap to have visibility is.

Gabriel Brown, Omdia (00:42:54):
And if I'm recognizing you correctly which may be wrong, you've got background in UK networks.

Juan Vertis Corredondo - Audience member (00:42:59):
Yeah. I have been working for three, for 12, 13 years, recently working on the merger with BobaFon and now I'm not

Gabriel Brown, Omdia (00:43:06):
With

Juan Vertis Corredondo - Audience member (00:43:06):
Them anymore working independently.

Gabriel Brown, Omdia (00:43:08):
Understood. So performance benchmark, performance data on massive MIMO, anyone. Henrik, I'm going to land on you first, but anyone come in.

Henrik Jansson, Samsung Networks (00:43:17):
So was your question about performance on an assembly deployment that is over and compliant or specifically mixing vendors?

Juan Vertis Corredondo - Audience member (00:43:29):
It's both interesting. Obviously it's more interesting vendor view because you could expect that it's potentially more complicated to achieve the results that you have in a traditional run when you are mixing two vendors over the open fronthaul. But even though without that, it'll be interesting on the standard results. Nokia is here or Samsung, you have experience in your traditional run and you're moving to open run. What are the differences that you are seeing in performance and capacity results, particularly in massive MIMO? Multi-vendor deployment, even better. So any information is always helpful to understand where we are heading to.

Henrik Jansson, Samsung Networks (00:44:08):
Yeah. So as I said for us, our baseline is open round compliance. So our masimum products are open round compliant and we don't ... So for us, the performance is the same, so to say, right? Our top performance is open run compliant performance. And I think obviously there are some open benchmarks that you can get. You can benchmark data for various markets where you have our deployment and where you have other vendors deployments. So that is available. Then yes, of course there are cases where there's been benchmarks between, for example, our own product and other vendor's product, but they are not open as far as I know. Some customers have done that, but it's not something I can share openly.

Sudhakar Pandey, Rakuten Mobile (00:45:01):
Sorry, if I can add, in Rakuten mobile as beginning of the panel discussion, what I told that 80% of our deployment in Japan is massive HIMO. So there are more than 20,000 plus sites which are running, which is ORN compliance, and it's fully ORN compliance. And there is crowdsource data which is available, and that is available in public, not to private domain. You can go and look into it. We are comparable compared to incumbent in Japan and certain KPIs, we are better than that as well. So in terms of performance, I believe that it's comparable. It's not a major concern in terms of KPIs.

Rob Soni, AT&T (00:45:44):
And just from us, I think we've been doing multi-vendor for traditional radio, and that's fully ORAN compliant, even validated with the radios themselves being able to do compliance testing specifically for that capability without the benefit of the base ban. We also established together with partnership through the NTIA and 22 ecosystem partners, including our direct competitor, Verizon a lab in Dallas, to allow for interoperability to move forward and to do the type of testing. That lab itself is right now primarily testing traditional radio and pairing traditional radio. This next year, we moved to massive MIMO and massive MIMO testing across a variety of vendors. For us today, we actually have massive MIMO running with an ORAN compliant interface. Granted, the base band and the radio are from the same vendor, but we'll be moving that too also as new bands and new frequencies open up. We had a safe base of what we needed for the bands that we had, but as new bands become available in the US through the big beautiful bill, we expect that there'll be lots of opportunity for us to look at MassiveMiMo as a new insertion opportunity for third parties.

(00:46:54):
Do we worry about performance? Absolutely. We're performance obsessed. We want to win. We care about our customers. Is there an opportunity perhaps to think about ... Is there a way to think differently about MassiveMiMo? I think the reality is right now is MassiveMiMo, even though the hardware is in the third or fourth generation, maybe even fifth generation, I think for the Nokia folks. What I've seen is that the software is still a space even for where it's aligned between Baseball and Radio where there's still opportunity to pick up performance. There's still opportunity. And I think that's been well written about, that there are still opportunities out there. And I think when you expose that and you go to open interfaces and you expose that as a programmable layer and you can manage the radios independently, this is where more eyes will look at this and potentially find yet more opportunity to gain from massive MIMO.

Ari Kynäslahti, Nokia (00:47:44):
Yeah. I'd like to see this is a journey. So I mean, our products are out and compliant, and to be honest, I'm disappointed of the kind of slowline as how slow the ORN market has been taking up. If I look at something like Delore analysis years back, they were seeing a lot bigger numbers for today than we actually see. So we have all the time being prepared and want to go forward, but it has been slow to take up. And on that journey, products are compliant. Like I told, we have the actual public networks we can talk about, which are today using the regular RRH. And then the next phase, I talked about SMO, but also the massive MIMO. So we need to see this as a journey. Okay,

Gabriel Brown, Omdia (00:48:38):
Good stuff. We have a gentleman in front here, please.

Drew Clark - Broadband Breakfast (00:48:41):
Yeah. Drew Clark with Broadband Breakfast in Washington, DC. My question is about the increasing fragmentation of the global marketplace and the US, of course. There's been this rip and replace funding and guidance. And a month and a half ago, the European Cybersecurity Initiative, I guess, seems to propose or put forward a similar idea in European nation states. In your opinion, does this fragmentation of the telecommunications equipment marketplace help or hinder the ORAN initiative?

Gabriel Brown, Omdia (00:49:22):
Okay. Anyone fancy a go at that? I mean, it's a bit of a rabbit hole, so we can talk a little bit about it. But I suppose the other thing is, does anyone want to challenge the premise a little bit? Is it fragmenting in the way that it's stated that Sudako, you have a go at it since ...

Sudhakar Pandey, Rakuten Mobile (00:49:40):
I don't think so. It's fragmenting. It's helping, and that's needed as well because any new technology will need consistent investment. It's not coming free of cost. So that government support is very much needed. So my answer is helping rather than fragmenting. So we want more and more governments to come and support these initiatives.

Rob Soni, AT&T (00:50:08):
I mean, you heard yesterday, I think, in the keynotes and probably at the GSMA opening and you heard the chair talk about the risk or the fear of fragmentation. And several of the CEOs of the operator partners just mentioned about it or discussed that there's concerns, legitimate concerns about fragmentation. We are focused on what we are trying to build today, and that's an open path. And we think that if there is ultimately a need for us to have different solutions between different parts of the world, openness is probably a key to ensure a diverse supply chain, but I don't know if we can really comment specifically on if fragmentation does one thing or another to open. I think fragmentation will have a bigger impact on the industry than open.

Ari Kynäslahti, Nokia (00:50:53):
I think two years ago, I was maybe a little bit worried about the fragmentation and bifurcation, but if I look at the actual work happening right now in 3GP, everybody have their hands dirty working on the release 21 of 60 standard together. I don't see bifurcation there. It's like we did the work for the 5G in the same community in 3G PP.

Sadayuki Abeta, NTT DOCOMO (00:51:26):
So I bring this region to not only the Europe battles, other countries, and talking was that as those operators are talking was that the European operator and the US operators, other vendors, we bring the solution to that Europe and also South CEG or other countries, but I don't see that big fragmentation, especially from that specification point of view, it's all same. And just which frequency actually we need to support the frequency fragmentation is that not only that this airbag, you can see itself is different from country, that kind of, but the equipment we can be use it and the interface point of view, no difference.

Gabriel Brown, Omdia (00:52:12):
Henrik, you're a European working for a South Korean company who's VP of North America. You've got three geographic regions all in one there. Just give us a very short thought from your side.

Henrik Jansson, Samsung Networks (00:52:27):
So I think in general, fragmentation doesn't help, but I think it's a little bit different topic than ... Anyway, all right, apparently I'm muted. I think in general, fragmentation doesn't help. So it's not good for an industry to be fragmented. I think it's different in a way from Oran as such, which is a specification about openness. And I think ultimately ORAN could help avoid situations because you're not so dependent necessarily on one single vendor doing this way or the other, but you have optional selecting components from various vendors.

Gabriel Brown, Omdia (00:53:06):
Okay, good. Any other questions here at the moment? Okay, Sudhakar, very quick one for you to start, and then a longer one maybe. Your massive MIMO front hall profile, what are you using there, just so people are-

Sudhakar Pandey, Rakuten Mobile (00:53:19):
It's profile six and above. So it's started with profile six and it's compliance after that as well with the latest profile.

Gabriel Brown, Omdia (00:53:27):
Sorry, second profile?

Sudhakar Pandey, Rakuten Mobile (00:53:28):
Six and beyond.

Gabriel Brown, Omdia (00:53:29):
Yeah. Okay. We

Sudhakar Pandey, Rakuten Mobile (00:53:30):
Started profile six

Gabriel Brown, Omdia (00:53:32):
And

Sudhakar Pandey, Rakuten Mobile (00:53:32):
That's

Gabriel Brown, Omdia (00:53:32):
About it. My longer question as a gentleman in the middle here with ... Okay. While the microphone is coming to him and here you go.

Bruce Peterson - Audience member (00:53:47):
Hi, Bruce Peterson with Aglocell. I'd be interested in your comments as far as the value you see from the front hall as compared to the value you see you'd receive from the SMO. So from the standpoint of from the radio side versus the SMO and the Rick and the additional applications.

Ari Kynäslahti, Nokia (00:54:13):
I can start. So maybe I go back to my previous comment about the journey. So ton of work on fronthaul, almost like a trivial nowadays. So I see more work now and more bus happening around the SMO and creating the ecosystem through the R apps. Coming back to the recent announcement with Ericsson around it, good example. So for me, there is more interest around SMO nowadays.

Gabriel Brown, Omdia (00:54:48):
Yeah. Okay. Anyone else?

Rob Soni, AT&T (00:54:49):
Maybe just to comment and follow on that. I agree with Ari. I think we've had Frontaul in our network for a long time and whether it was exposed or open or not, it is an unlocking mechanism for us to bring other vendors in, and that gives us opportunities in some cases to bring innovative solutions. And we do see that, that the radio partners are trying to differentiate themselves within the box itself, and they can, especially with respect to size and power consumption, and then finally what they can achieve in what timeframe for our targets on frequency and frequency allocation and radiated output power. We see opportunities there, but admittedly, we didn't really have an open, programmable, multi-vendor layer in the network before. So the SMO is new and it has a lot of undiscovered, unrecovered value, and we're going to find out what that is over the next few years as those who are actively deploying it start to use it and see the benefits.

Gabriel Brown, Omdia (00:55:50):
Okay, good stuff. We have a gentleman here. We've got four minutes left, so this may be the last one. Let's see how you go.

Ramesh, HCLTech (00:55:56):
Yeah. Hi, Ramesh from HCLTech to Rob. There is this universal network management platform as SMO that you envision sometime back, which folds in not only RAN, but core transport and other domains as well, right? So how do you see the SMO and RAPs kind of extending to all these different domains and how collapsing everything into a fundamental cloud is the necessity for that? So where do you see that going or when do you see that happening?

Rob Soni, AT&T (00:56:26):
It's a great idea in principle, but I think reality is for a lot of operators, those domains are owned by different vendors and they fundamentally have different strategies with respect to automation and the evolution of automation. So while ORAN could lean forward and say, okay, I want to define an SMO for transport, or I want to define an SMO for core and link the two and provide end-to-end service management and orchestration. I think it's a nice vision, but it's likely pretty hard to realize. So I think what it really makes more sense, of course, is to start to think about as these schemas come together, as these three different domains come together, is to really look at Agentic as the opportunity to stitch them together. So an Agentic layer that sits over the top of the three of them with some form of orchestration automation is really the right vision.

(00:57:12):
It's a really open question. Is it something that the ORAN Alliance would want to lean forward on and set a blueprint architecture specifically in the ran domain for how Agentic would be inserted? It is an opportunity that we would support, but I don't yet see a lot of attention to it yet.

Ari Kynäslahti, Nokia (00:57:32):
I agree with you, Roman. If I think it from the use case perspective, let's think like slicing. So slicing needs to touch all parts of the network, radio, transport and core. And when building a solution for dynamic slicing and making it very easy for you to use, this is exactly the question in our mind that could we extend the SMO to cover everything or should we rather make sure that the slicing management on top uses what there already is? And this is now the approach. So if one day we see that SMO expanding, let's see, but today in practice, we need to work with what we have in a use case like slicing.

Sudhakar Pandey, Rakuten Mobile (00:58:18):
Yeah. Just to add a different perspective to it, that SMO at least bigger front, it has to orchestrate all cloud transport as well as RAN, and there is a reason to attend. At least we started to some extent in that within mobile and there are tangible benefit to it because in isolation, there is always automation available on each individual layer, but how network operator can get 360 view of the network and including one of the classic example is anomaly detection. There is a layer which is monitoring all other layer down the line. So it's very easy for an operator to find out where the normally is. And then we all operator know the problem. Detection is more time taking. Once you know where the problem is, it's easy to fix. So I think ORN can play a very big role here in terms of strategizing, especially SMO and all there on the RIN, it has to manage through

Gabriel Brown, Omdia (00:59:17):
That. Okay, good stuff. And so we do have an SMO specific panel coming up. There's obviously a lot more to say here. On that note, thank you all for your questions. Thanks to all the panelists for sharing their views today. Please join me in giving them a round of applause. Exit stage left, I think Shakespeare said. Cheers. Thanks a lot. Thank you, Ari. Henrik. So we're going to come to our next session, which is a fireside chat. We have Igal Elbaz from AT&T, Thomas Lips from Deutsche Telecom. If each of those very distinguished gentlemen could come to the stage, please. Format's going to be a little similar. It's a smaller setting. I'll take a seat, but again, we'll take some audience questions as well. So if you do have a burning issue you want to get out there, please, gentlemen. Yeah, that's good. Yeah, you're good.

(01:00:26):
Okay. So Igal, good morning.

Yigal Elbaz, AT&T (01:00:30):
Hello.

Gabriel Brown, Omdia (01:00:30):
Thomas, welcome back. Igal, you are SVP and network CTO at AT&T, and you are a founding board member at the ORAN Alliance. I wanted to start with actually an ORAN Alliance perspective before we come on to talk about Deutsche Telecom and AT&T and so forth. You've been involved since the beginning. How do you think, and just give it your reflections on how the Alliance is meeting the needs of the industry today and what it needs to do to maintain In relevance without maybe overreaching itself in future.

Yigal Elbaz, AT&T (01:01:06):
So first of all, good morning everyone. Great to see you and thanks for having me. I think that if we got to a point on which we have a panel on SMO, that tells you that there's a great success here. But a little bit more seriously, absolutely think that what the ORN alliance have accomplished over the last six, seven years is nothing shock of remarkable. And the breadth and the width of specification and capabilities and direction that it provides to the industry was extremely important. And at the end of the day, we should measure success but by what we are seeing in the marketplace. You've seen the panel. So first of all, unfortunately there's not too many, but if there's a new greenfield telco out there, they will probably build all stack from the ground up with an open RAN specification. You're starting to see established operators we included getting on the journey of moving to Open RAN.

(01:02:25):
So there's real evidence that companies like Deutsche Telekom, AT&T, Reliance and others, NTT are all in either in the journey, different stages of where we are in terms of going and being very declarative about this, about their journey and about deploying Open RAN in their network. I think a conversation with a vendor today, the key vendor in our industry, I don't think anyone will tell you from any of the incumbent vendors, "Oh no, we are not doing open or the conversation would be not productive." I think the vendor ecosystem is well aware of what we're trying to accomplish is playing along. They've opened their roadmap. So I think we are in a pretty advanced phase in terms of where we thought we were going to be from an open rent alliance. And obviously we have the specification ready, we have the documentation, we have testing environment.

(01:03:33):
So I'm pretty pleased with the state of affair. In terms of things that probably needs to be done a little bit different, it's the same theme you're going to hear me saying across all kind of conversation throughout the week. The pace on which we are doing things in our industry needs to accelerate. We live in a world in which the pace of technology around us is moving crazy fast. There's a new foundational model that is drop on us every three to four weeks. We cannot allow ourself to spend weeks, months, and years on agreeing on a specification. So that's in the heart of what I would describe as things that we probably need to do a little bit differently. And we can talk later on. I have a couple of other things, but-

Gabriel Brown, Omdia (01:04:30):
Yeah. Okay, good stuff. I'm sure

Yigal Elbaz, AT&T (01:04:32):
Thomas wants to chime in a bit.

Gabriel Brown, Omdia (01:04:34):
Yeah, of course. I mean, Thomas, chime in. I'll tell you what, I mean, you can add to that comment on ORAM, but why don't you give us sort of a Deutsche Telekom perspective a little bit as well. I mean, we've heard from two Japanese operators, North American operator. Give us a Deutsche Telecom hat here.

Thomas Lips, Deutsche Telekom (01:04:56):
Very happy to do, of course. And thanks for being back here in the Independent. As everybody knows, Deutsche Telecom is a founding member of Oregon Alliance. So our journey has started pretty early. We had early tests, early deployments, but what you see right now is in Germany, we are already deploying at scale. We will have next year about 3,000 sites already deployed. So it is not a trial anymore. It's really massive deployment, what we are doing there with our colleagues from Nokia and Onefinity. So that is one element. That's the element for Germany. I'm always a bit jealous on Eagle's size of network because we are in Europe, we have a lot of countries that we are bringing together. We announced it already, I think at the fuse event that we are going now for a big component-based RFQ. So really all those components are individually RFQs.

(01:06:09):
And while that one, we will find the optimal setup for us. And this will then include all the countries a really big step forward. The element that I heard quite often today, and maybe this is what we are most proud about, is all this AI that comes now. With the SMO, the rig and its capabilities, we are at the center of that one. We started it super early and we are doing it differently than most because what we are doing is we are reshaping the current run OSS. We implement dynamic slicing, own software. We do all the configuration, conflict management. We're doing it already today, even on the existing networks that are not yet ORAN, and we have a transition path towards SMO. Some might have read our press announcements. So this is not even only for Deutsche Telekom. We are in a partnership with Capgemini.

(01:07:17):
And in this partnership, we are developing our SMO rig together, but we are bringing it also to the world market. So we are helping that the industry is adopting, is able to use the capabilities in our current 5G networks, getting it hypercharged, I must say, with the AI capabilities of RAPs that are already there or that are coming.

Gabriel Brown, Omdia (01:07:44):
Yeah. Good stuff. I have read that press release among the massive slew of press releases coming through. I have noted on it. Maybe if you could, and I imagine there's some limits on this, elaborate a little bit on, you mentioned this RFQ that you announced at Fuse. Could you just elaborate a little bit on what you can publicly on that process, what stage it's at, what the rough timelines are? I appreciate you don't want to probably say too much, but give us a feel for it.

Thomas Lips, Deutsche Telekom (01:08:11):
Then my first sentence is, I don't want to say too much because a lot of decisions that you're making in an RFQ are strategic and that is not that easy to say it right now because not everybody is prepared to bid in such a component RFQ. Some that are focusing on certain components. I don't want to mention any suppliers right now. For them, it's easier. Others that are normally offering it end to end, we have to see. So therefore, I prefer for the time being neither to say to talk about volumes, timelines, and all of this beside that all countries are in and it will be an exciting journey to come to this one and we will keep you updated in the round.

Gabriel Brown, Omdia (01:08:57):
Yeah. Okay. Good stuff. Actually, a very good answer and enlightening in its own way. Thank you. Hego, maybe I could move back to you. And by the way, if you do have a question, now's a good time to raise your hand or start formulating it. I wanted to ask a question. Obviously you're in the ecosystem. The AT&T scale, as Thomas talks about, is significant. How much do you, and I guess it's in forums like this, how much do you collaborate outside of your home market with other operators in terms of learning, sharing? Obviously there's a lot of confidential information in the industry, but are you able to work with peers and other operators with aspiration and ambition in ORAN?

Yigal Elbaz, AT&T (01:09:40):
Yeah, absolutely. From the outside, it looks like a very big industry, right? Telecom all over the world. But at the end of the day, there's a lot of us that know each other for many years. It's actually not such a huge community of people. And I think that the shared mind on specific around things like Open RAN in our case always found themselves with the opportunity to collaborate, whether it's the Open RAN Alliance meeting, whether it's TIP, Rob was here on stage, whether it's MobileWatch Congress while we're here and seeing each others. And there's also occasions for us companies to meet regularly and sharing ideas about deployments and about what are we seeing in the industry and what are the things that we want to drive. So it absolutely ... It's a relatively collaborative industry when it comes to forming technology roadmaps and being able to provide some guidance to the industry about what we would be interested in.

Gabriel Brown, Omdia (01:10:55):
Good stuff. Thanks. Any questions out there? Surekar in front here.

Yigal Elbaz, AT&T (01:11:00):
You were just here. We didn't ask you a question.

Sudhakar Pandey, Rakuten Mobile (01:11:06):
Yeah. Okay. Thank you guys. So one question I have with respect to Rick, right? So as of now, majority of the development happening by operators, whether it's AT&T or Dustelecom, or we are doing something on Rakuten mobile as well. When do you guys see from ORN alliance perspective, we will be in situation wherein XAP, RAP can be ported out or other operator it can be given to use. Suppose it's developed by TNT for energy efficiency, some application, and it will be available in portable version to other operators. Maturity of the rec ...

Yigal Elbaz, AT&T (01:11:45):
It was a little bit hard for me to understand if you can repeat the key part of the question.

Sudhakar Pandey, Rakuten Mobile (01:11:49):
Yeah. So the question is that by when do you think that these X app or RAPs will be portable to other platform or suppose developed by other operator? Because as of now, things are happening in silo. So that's I think more role of ON wherein standardized and those apps should be portable to- Got

Yigal Elbaz, AT&T (01:12:10):
It.

Sudhakar Pandey, Rakuten Mobile (01:12:10):
Yeah.

Thomas Lips, Deutsche Telekom (01:12:13):
You're pinpointing and really relevant item because we are asking this all the time. If somebody has an RAP, can you put it in? Portability is not yet achieved because it's data models that they need to be aligned. It's every time a slight IT integration. But we have some initiatives now ongoing from a tip perspective as well from R& Alliance perspective to make this more portable, but this is not a four weeks exercise. Again, data models, it will take a while until we have this. We have use cases, we are some zooming in those use cases, and we try to frame them in a good way that why are such a use case portability can be achieved.

Yigal Elbaz, AT&T (01:13:00):
And maybe take this a little bit to a different space and maybe answering your earlier question about the challenges. I think this is what we as an Open RAN alliance community and industry community, this is where we need to make sure that we are focusing. We have to be laser focused on providing guidance on deployment. We have to start reducing the number of options. We have to have more clarity about what it will take to make it work across operators. I don't need to see roadmaps. Vendors who comes in with vendors, I'm not interested. I want to see evidence that things are open, portable, but also we as the open RAN alliance needs to be ... What happens, and it's not specific to the Open Run Alliance, this is how industry works. If we don't agree on something, we just adopt all of the opinion as an option, and we cannot carry on this way.

(01:14:11):
We're going to have to have, yes, some level of consensus on the main thing that are deployable and probably have couple of flavors. And once we are going to get into that place where things have clarity about how they're going to get deployable, how they're going to get operated, then I think we're going to be able to start answering those question and interoperability between data models and platforms and operators. But it has to start with simplicity.

Sudhakar Pandey, Rakuten Mobile (01:14:43):
Thank you so much. Thank you.

Gabriel Brown, Omdia (01:14:49):
Anyone else? Nine minutes left, so if you do have a question, get it ready. Iga, let's go to you and then Thomas. Thomas mentioned in his intro, there's different paths into ORAM for different operators, depending on circumstance and background and so forth. One of the big questions is the multi-vendor aspect. Do you have a feel for how multi-vendor a RAN should be or is there a ... Put it like this, there's a state where you have, in Sudoku's case, many, many vendors, and there's a case where you have maybe two or three or it's more constrained. Do you have a feel for how that goes or is it to be determined?

Yigal Elbaz, AT&T (01:15:28):
Sure. First of all, I don't think ever there are going to be an optimal number. I'm not sure I even understand the frame of the question. I think that every operator should have its own strategy about what kind of players in the ecosystem they want to work with, as well as long as you have the flexibility to be interchangeable in terms of who do you work with, then to me, that is what's the important thing. Now, different parts of the stack may have different answers. I don't think you're going to climb an AT&T tower anytime soon and you're going to see seven different vendors on the tower. Are you going to see in different domains, whether it's small cells or indoor or DAS or even on the macro different options? Yes. But there's a limit of what you want to do in certain domain of the network.

(01:16:27):
On the other side, we just answer a question about SMO. I think this is actually one of the areas that maybe not be truly clear when we got on that journey, but that may end up to be probably the most important part of the oral architecture. First of all, it's software based, so you have the ability to adopt many options of RAPs from different vendors. And the whole vision of having new startups in Silicon Valley, building capabilities for telco, it can start to be realized when you're thinking about our apps. So again, it depends on the domain and it depends on each one of us. How do you want to see our ecosystem? I don't think there's an optimal number. It depends on ... And again, there's a point in time on which there's a little bit of hustle today. Let's be honest. If you want to integrate a radio from one vendor with a base station, from a different vendor, yes, we have specification, but it takes a little bit of work.

(01:17:34):
And hopefully over time, all of that work's going to get removed or it's going to get tested and interop in other areas. So when it comes to the operator, it just work. So we have a little bit of time together and once we get there, then again, I'm sure each one of us will have its own vision and strategy about how you implement this.

Thomas Lips, Deutsche Telekom (01:17:59):
Maybe complimenting because very clear that we are answering the same, there is no optimal number. We want to have a vibrant ecosystem of suppliers. We want to have competition. Competition drives innovation. This is what we need in our industry, that's for sure. But we are not defining how many we want to have. We defined in DT that we want to have the SMO as a strategic control point, that not every time I'm asking somebody, "Can you integrate?" We are integrating that one. Viabrand ecosystem on our apps, no limitation at all, but the rest depends on commercial framework and practicality also in the way how many you can deploy. What is different probably in the world today, we are not committing to seven, eight years timeframe, and then the big modernization comes. We can complete it a couple commitments on software, commitments, on servers, hardware. Maybe as Igal said, ready units, they are staying quite a while on the tower, but so therefore with these limited lifetimes, we are more flexible and the flexibility is what we wanted to achieve all the time.

Gabriel Brown, Omdia (01:19:15):
Yeah, quite a big change in process and concept, really. In the absence of one last question from the audience, I have one for both of you. Let's start with Thomas and come to you, Gal. How do you rate the quality of product from the ecosystem, the viability of the ecosystem? I mean, I know obviously you're going to try and be quite nice to everybody, but of course, if I look around as a vendor or supplier, everyone needs to get paid. It's a lot of R&D to support and to deliver a good product. Are you seeing fantastic products, pretty good products, or are you lacking products? Just give us a feel for how you see the quality. No names mentioned or anything like that, but just a feel for it. And

Thomas Lips, Deutsche Telekom (01:19:59):
Obviously once we are rolling out stuff in our networks, it has to be already on a high level. Will I ever give a great one? No, there's always a wishlist of things that come more, but what we see is all the suppliers, all our partners are picking it up and raising the bar on what they can do and how much they cooperate with each other. I would say it that way.

Gabriel Brown, Omdia (01:20:31):
So your process obviously is filtering and bringing out the best that you want.

Thomas Lips, Deutsche Telekom (01:20:36):
At least it helps that way, how we are working our way forward and motivating our partner suppliers to work with us. Again, I don't want to give a great because it's very dependent on the timing and the product itself.

Yigal Elbaz, AT&T (01:20:55):
For all the community out there, we shouldn't assume that openness compromise quality. That's not what we're trying to do here. We're building an open architecture that's going to allow us actually to move faster, to have a better operational efficiencies. As Thomas said, every time you have competition, there's innovation showing up. Our ability now to bring apps and do AI in the network now, that's what we are after. And no one ... I think Rob said us on stage a few minutes ago that everything starts with our customer experience and that is not negotiable in terms of quality and no one should assume our customer shouldn't know or shouldn't experience something different just because we are trying to pursue an open architecture. So that needs to come together. So everyone who wants to participate with us, and we are actually encouraging everyone to come and work with us, that is the bar that we are after.

(01:22:05):
And I think the way the Open Rent Alliance is driving the specification and the work that Thomas and Tim are doing is exactly around that. And I don't think ever anyone of us came in with the assumption that openness is compromising quality or speed.

Gabriel Brown, Omdia (01:22:33):
Okay. So on that note, Igal Thomas, we're going to close the session. Thank you both very much. Please join me in a round of applause.

Thomas Lips, Deutsche Telekom (01:22:41):
Thank you.

Gabriel Brown, Omdia (01:22:42):
Thank you, gents. Thank you. Thank you. Thank you very much. Thank you. So we're going to move to the SMO and RAN automation panel. That's going to be moderated by my colleague, Ruth, who's going to come and take the stage here. And if I could ask all of those people to come to stage. Ruth, please.

Ruth Brown, Omdia (01:23:27):
So thank you very much, Gabriel. I feel this panel probably needs very little introduction. Most of our speakers already this morning have already commented how important RAN automation is becoming, and especially with frameworks such as the SMO. So I know Rob and Yale have both commented just now about this year's had a lot of innovation. So I think rather than speak anymore, it would be great just to get straight into it. I'm joined by some fantastic speakers today. I'd just like to go down the line, just if I can get you to introduce your sort of name and job title. So we'll start with you at the end, please, Warren.

Warren Bayek, Wind River (01:24:06):
Hi, I'm Warren Beck standing in for Paul Miller, who you see his picture, that's not me. I'm the VP of Intelligent Edge at WindRiver in the CTO organization. So I oversee all of the intelligent edge strategy for Wind River.

Gabriel Foglander, Ericsson (01:24:20):
Hello, everybody. My name is Gabriel Fuglander. I work in Ericsson. I'm heading up strategical leadership working on Rand software.

Changsoon Choi, Deutsche Telekom (01:24:29):
Hi, I'm Chanceon Choi working in the group technology Deutsche Telekom, mainly driving the AI RAN, ORAN in the head ofcote organization of the R&D. Luthia

Lucia de Miguel, Vodafone (01:24:40):
DeMiguel, Bodafon, OpenRun, digital software and system development. I am accountable for the RAN software, but also for the SMO, RIC platforms and all the automation part. Hello,

Kanika Atri, NVIDIA (01:24:54):
Everybody. My name is Kanika and I'm part of NVIDIA and I take care of everything marketing and go to market. And I look after not just autonomous networks, but AI ran, 6G tracks, sovereign AI, and many more

Ruth Brown, Omdia (01:25:10):
Topics. So thanks everyone. So this really today seems like the ideal setting. This week already, we've seen a number of announcements around RAN automation. We've also seen a lot of forward thinking within the last year, starting to get towards the sort of end vision of using much more intent-driven RAN. I guess to start off this conversation, I'd like to kind of come to Janssen and Lucia first to get the kind of operator perspective around what's been happening within the last year, especially what the focus is in their journeys to run automation. You've both, Janssen, obviously from DT, Lucia from Vodafone, taken different paths inspose in terms of a kind of build your own approach versus using a vendor solution. So Chancellor, maybe we'll start with you first just to share your experiences.

Changsoon Choi, Deutsche Telekom (01:26:08):
Absolutely. And first of all, thanks a lot for having me here. And this is wonderful event. I'm really honored to be part of this section and then the summit. So maybe let me start from the listed Deutset comes angle. At least I can tell you that in the Deutsche Telecom, we see that this run automation, not as just one time the modernization, but as let's say continuous transformation journey toward to the autonomous and AI native network. I think that is the bottom line of our, at least in the journey in the last, at least one year and the 12 months. I'll say that if you ask me about our commercial network right now, like many other operator, we have also long use in the conventional lesson one configuration tool such as the Song. And it is true that it gave us the lots of the less important value, such as partner optimization and also enabling cell tuning and optimization.

(01:27:09):
And that is actually some value that we got. But when it comes to how we're going to bring the more AI enabled run intelligence into the commercial network, it is actually true that it is not really scalable with multiple AI use case. And this is something that we are also now thinking to implement as an RAP, and that is the situation. And in addition, we also see that it is not also easy to have the full observability when it comes to the order, and particularly from let's say all cloud side and AI run side. And also in the meantime, we clearly see that there are lots of divisions demand on, okay, how are we going to provide the dynamic slicing, which actually requires this order of dynamics, let's say, run domain orchestration functionality. So putting all this region together, we eventually decide to go and try to implement the run domain orchestration and then go into the SMO.

(01:28:13):
And what makes the Deutsche Telecom a little bit unique compared with the other operator is that like Thomas already mentioned, we decide to develop our own in our SMO, and this is at least the journey that we have started from the couples of the months ago. And that was, I think, some part that we say that last at least one year, that makes me very busy. And I think if you wanted to know the much more detail, I think you are more than welcome to come to the Deutsche Telecombus I think quarter to four today and that you're going to listen what we have actually suffering from and what we have developed so far for the DC in- house SMO in the non-real time rate. But briefly speaking about what are these some, let's say, key assets and what is the reason for this, let's say, run deal and then in touch SMO journey, maybe I'll say maybe two and three point, maybe that's not something that we can say.

(01:29:12):
The first one is, I think this basically allow us to consolidate quite fragmented run management platform and also automation logic, which is really essential. With this in- house SMO, we are also able to, let's say, like we discussed earlier, harmonize the different data model and also make some unified MLOps, which we can use for not only just like RA provisioning, but also other purpose. And that kind of like how we can harmonize the fragmented one management platform, and that's the first value that I clearly see. And second one is, I would say maybe faster time to market. Let me put it like this. And as you may also see how people are now discussing the run alliance, we probably need to also accommodate the several MCP or AI agent platform on the top of the SMO. And as long as we are able to develop this in also in- house SMO by ourself, we can easily make SMO to be fully supportive to any kinds of AI functionality that we need.

(01:30:24):
So that kinds of the time to market and also going into the more slicing and new types of the service, I think that is the second merit that we clearly see as an inoSMO. And last but not least would be, I mean, as we discussed earlier, this is also accelerating the innovation side. I mean, we clearly see that this many AI-based RF are also coming into the market, not only coming from this vendor specific app, also third-party AI-based item, and this is what we clearly see in the engineer market, and then putting this SMO with not real timely, and as As long as we can develop the distance by ourself, which we can easily support the multiparty without some other interface and every kind of tassel. So that is the value that we clearly see to accelerate the innovation cycle.

Ruth Brown, Omdia (01:31:14):
So thank you. Thank you. Obviously DT in the past have been very open about the challenges you face and you've mentioned data fragmentation. A lot of others have suffered that and it's one of the foundational levels you need to build up before you can start tackling that. I guess some of the use cases before we move on to talk to Lucy about her approach, perhaps you could mention a little bit more about the focus use case that you're maybe looking at first.

Changsoon Choi, Deutsche Telekom (01:31:48):
We are looking at actually multiple, let's say the use case, particularly for the 5G standalone and also let's say to increase the customer experience, but also we are looking at a bit of how are we going to improve the operation excellence. I think the one that level use case that we are also pushing quite a lot is that we try to use some AI for the anomaly detection and how we're going to bring into the loose cost analysis in the set mode and then this other area. And then when it comes to the more how are we going to increase the customer experience, particularly for the slicing aspect, we believe that I think how we going to predict this quality and then run capacity so that we can guarantee the certain quality coming from the slicing. We see that this has this really value from the customer perspective.

(01:32:38):
So I will say that one, Lutco analysis is anomaly detection and second one is the QS assurance. I think these are the use cases that we are really focusing at this move.

Ruth Brown, Omdia (01:32:48):
Brilliant. Thank you. And if we moved on to Lucia now, obviously Vodafone have taken a completely different approach to DT, which I guess also showcases that if you are using some sort of underlying framework, there are multiple approaches to be able to come up and use the same sort of open architecture structure. Within the last 12 months specifically as well, how have you seen things change? I mean, are your goals still the same as this time last year? Obviously there's a lot of momentum in the industry happening around this, but what have your experiences been?

Lucia de Miguel, Vodafone (01:33:27):
We follow, I think, seeing that the run automation is a started process that is deliberated evolution, but the difference is that we are moving from POCs or trial, small developments to the industrialization, or we can call it for more massive overall deployments. And this is key because you need to have the SAN enablers that we are seeing in the SMO. The difference is that we are not developing our SMO. We decide via I think a VCRFQ where not only we decided the runs of our vendors, also we decide some of the automation tools, especially the SMO and the non-real time rig where it's included. And we are integrated in our network as a basic, I think, point to deploy open RAN nodes, not only because we are trying to converge also with the traditional run. I think this is a bit the difference, but also it's a common approach because in the use cases, talking a bit what's changed is industrialization, you need due to the complexity to disgregate the hardware and the software.

(01:34:45):
To manage the cloud, you need to have some automation element. And we see the SMO ask this element to be able to do automatic deployments. That is one of the main use cases that we are implemented. Farther, of course, operational. So they want the operational part to do the cell healing or at least to detect the root analysis cost. And later as final the optimization, because we are evolving for the traditional songs to Rick. What is good in there, I think in the optimization is AI, that they are more AI native platforms and they are managing the disgregation of the hardware and the server. They put the infrastructure requirements for AI that later we can discuss a bit more about the AIs, but you need some computing where you are able to do these automatic tax.

Ruth Brown, Omdia (01:35:49):
And I guess before we sort of move on to bring in some of the others, have you noticed we were talking about pace of change. Does it feel the pace of development or these proof of concepts to moving to life? Does it seem like things are going faster, developing quicker, or is that just there's more of it and that's why we're hearing more?

Lucia de Miguel, Vodafone (01:36:11):
We are talking always pace because firstly, it's to decide what is the best solution. So it's not simple. Later, you have to identify the process. So automation, you can have the three, four main process, but later you have to, I think, to specify in your network how you are doing this evolution because it's not only about, I think platforms, technology, it's also process and also a skills of the people. You have to do it operational new model. And this is also, this is the way because it's space, because you need time to change, especially in the traditional operators to do it.

Ruth Brown, Omdia (01:36:54):
Yeah, thank you. So we're going to move on now and really kind of start discussing with Kanika, Gabriel, and also Warren, what we're seeing in the vendor spaces, have the strategies changed? How has this automation framework changed, evolved within the last year? So Gabriel, I'd like to come to you first really as a vendor of both RAN software, but also automation software. And just to comment what you've seen in the last year and also some of the key developments that have really started to occur.

Gabriel Foglander, Ericsson (01:37:30):
I think I was listening in on the panel before here, and I think Yigal and Thomas was mentioning it, both of them, that the R1 interface from the SMO platform is really enabling a wider ecosystem. So I think we had noticeable agreement with Nokia the other day to make sure that we onboard their our apps onto our platforms in the same way they resupposed to that and onboarding our R apps on their platform. For us, that I think takes the ecosystem to around about 90 vendors today available through our platform, let's say, that we provide. So I think that kind of notion of enhancing the architecture to promote ecosystem diversity and innovation is really coming through. And then I think we also, like we work with both the distinct colleagues here on the panel and we see that the starting point for all the vendors are a little bit different, fragmentation on the northbound side of things and the path that is correct for each customer is not necessarily the same for the other one.

(01:38:34):
So I think we work very intently with pairing the automation that you can trigger from the small level of network, slower loops, access to wider datasets, excellent data sources in tandem also with the intent-driven automation at Genobe where instant decisions are taken in the resource domain. So for someone like Chung Zung that built this on top, of course, our ambition is that the interworking between the round software and the SMO platform is there regardless of whether it's our product or an in- house product. So that's really the aim to really go for the advantages that you say, reducing the complexity, making real-time decisions to improve user experience, and also industrialize the development that we see now with differentiation of connectivity and monetizing that.

Ruth Brown, Omdia (01:39:26):
So thanks for mentioning that. And obviously you mentioned you've got a number of vendors, our app developers on your platform. Lisia and Chanceen mentioned their kind of use cases focuses right now. Are you seeing though you've got a lot of our app developers, are people taking really different strategies in terms of what's their first use case, what's vital for them?

Gabriel Foglander, Ericsson (01:39:50):
Yeah, that's my perception as well, similar to what you were sharing, right? It depends on a little bit on point of emphasis. I think for some simply the fact of consolidating to a single platform is unlocking a lot of the benefits of kind of decluttering, if you will, the ecosystem and cleaning up the data structures. But then on top of that, I think the some type of optimization use cases are existing before, right? And I think what you were mentioning here is we see that moving now into an SMO plus RP domain, which I think will be a really healthy movement.

Ruth Brown, Omdia (01:40:25):
Thank you. And then I guess also part of having these open frameworks and one of the major industry observations that we're noticing is having these kind of frameworks is making a much more collaborative environment ecosystem. We've spoken about the vendors on your platform, but you're also working with a number of different vendors to sort of create these whole automation solutions. So Kanika, I'm going to come to you. I know we were sort of speaking around a video obviously has a number of focuses across the man automation space, but I'll let you put in your own words around what your focus is, what you're kind of thinking and how it's like linking in with others in the ecosystem.

Kanika Atri, NVIDIA (01:41:11):
Absolutely. I'm a little bit surprised that the word autonomous networks hasn't come up in the first five minutes of the conversation, but let me bring that up. And the first thing I want to say is automation is not the same as autonomy. Autonomy is really about self-configuring, self-optimizing, self-healing, bringing the human in the loop when you absolutely have to. And we believe that that journey from automation to autonomy is accelerating at a breakneck piece. I see this every day through our partner ecosystem, through many of our customers, and it's built on AI. And if I have to peel it back, there are four layers to think about. At the very bottom is what we call foundation models. And everybody would be as aware of every month there is a new foundation model that's available with breakneck capabilities, right? So you take the foundation models and you turn them into what we call large telco models.

(01:42:21):
These are models that are trained on network data. They understand the network language. They understand scripts, they understand architecture diagrams, they understand logical VPN tunnel mapping, they understand logs, counters. And it's not limited to RAN. That's the other part about autonomous because end of the day, you've got to deliver a KPI. So it's all across the network. So it starts with large telco models. And on top of that, you start building agents and between there somewhere, you've got to build the guardrails and you got to make sure that this agent is saying the right things. It's secure. It's of course, whenever it's not confident of an answer, it knows when to stop. So in between, there is a whole data curation and whole pipeline that teaches the agent on what it can and cannot do with 100% confidence. And then you have agents. And then on top of that, you have simulation because in order for agents to actually do something useful, they want to validate their recommendations into a digital twin, check it before actually actuating it.

(01:43:31):
So in order to achieve this full autonomy, you got to build all these layers of the stack. And this is where the phenomenal amount of progress is we've been seeing. And the other part of automation and the autonomy framework that I want to talk about is, it's not just the operations part, which is to think of it as not an online thing, not in line. The other part of it is the whole agentic RAN itself, which is at the layer one. In your inline pipeline, when RAN is making those decisions on where to point my beam, that is where a whole lot of agentic applications are coming forward. That is where the idea of DAPs is absolutely the next foundation that will allow you to do site-specific learning, different behavior on different sites because these D apps are integrated inline into that radio behavior. So when we talk about autonomous networks, it has both the inline components as well as the operations.

(01:44:35):
And all of that is built on a full stack from large telco models to data pipelines to agents and then digital twins. They all work in tandem to get there. So that is what our vision is, and that's what we are working with a very large ecosystem to drive all of this.

Ruth Brown, Omdia (01:44:52):
Thank you. And that's like a really good summary of everything you're working on. I know you mentioned DAPs, and we're going to come back to that in a minute as our next point, but I'd like to also segue to Warren quickly just to ask about what your focus really, obviously as that infrastructure layer in supporting all of these others within the ecosystem to ultimately deliver these solutions that the service providers need.

Warren Bayek, Wind River (01:45:19):
Yeah. As an infrastructure provider, we see the critical part of having any of this work is to have the highly available, deterministic, low latency cloud infrastructure. So at the bottom of all of this, all of the things we're talking about is this infrastructure requirement that we see the past year. It's undergone a major transformation in that as this ORAN alliance and as the implementations have gone from small scale to very, very large scale deployments, as someone mentioned earlier, the Samsung rep multi-tens of thousands of nodes in the Verizon North American network, being able to manage at that scale, as someone mentioned, automation isn't an option anymore, right? It's a requirement. And it can't be just human automation, as you mentioned. It has to be somewhat autonomous, meaning humans, this is just far too complex for humans to deal with anymore. So that's something, a big change we've seen is that the automation requirements are being pushed more and more to the bottom, all the way to the top.

(01:46:28):
So as an infrastructure provider, we feel our real value is to provide all of the telemetry from the infrastructure in a very deterministic way to all of the SMO, through the O2 interface, to the SMOs, to everyone who needs that information in a very reliable, highly available, real-time method, because that's the only way things like the apps are going to get the information they need to make the changes they need. And the second part of what we do as an infrastructure and that's becoming more and more important is to allow the, for instance, DAPs to be hardware awarely placed. In other words, that we can decide at the infrastructure layer where it's appropriate to put the various RAPD apps, X apps, and to do that in a hardware aware way so that the applications are placed at the correct part of the infrastructure where they need the inferencing information and they can act on it in the way they need to act on it.

(01:47:29):
Some of the applications such as the apps have to live at the very low layer they interface directly with the radios, they can't go back to the cloud for information. They have to make their decisions locally. Other applications, it's not so important that their information is handled in real time. So as an infrastructure provider, deciding and being able to give the hooks through to how to place those applications in the right place, very important part of what's happening in the ecosystem going forward. The other thing that we've seen is, and I'm glad someone mentioned, I think one thing we always forget as technical people, right? We love our science fair projects. We love doing these cool things. In the end, we're all here for one reason, the customers, right? I think we don't talk about the customer experience enough. The reason we're doing all of this is to help the customer experience.

(01:48:18):
And by helping the customer experience, that's how operators and infrastructure providers and all of us are going to make more money. So in order to keep the customer experience as high level as they are used to, we have to do a lot of different things as we open up and virtualize this entire telco environment.

(01:48:42):
One of the things we've seen a big change this past year is something as simple as the maintenance windows, right? We are updating AI models, we are updating hardware models, we are updating NIC drivers,

(01:48:56):
Putting new security in place because security's becoming more important. Being able to do things like predictively understand as we're going into a maintenance window where we have to upgrade two or 3,000 sites, predictively understand, are these sites able to be upgraded tonight to all of these, to the various things I'm doing to the network? And there can be a variety of reasons why a network wouldn't be ready for that. There might be some network anomaly happening with some site in some remote region. There may be an old bias version that somehow drifted from what it was. Anyway, there's a variety of reasons why a maintenance window could fail. So being able to do something as simple as saying, "You know what? These 10 sites, we're not going to do those tonight because they're not ready." That has helped operators make sure that they can do the upgrades that they need and then the customers aren't impacted.

(01:49:49):
So we hear a lot about anomaly detection and predictive maintenance. That's kind of one example of where we've seen a lot of improvements in the past year of being able to keep the customer experience high by doing predictive maintenance prior to outages. It's great to go in later and have a great root cause analysis, but what's really more important is to prevent the problem in the first place, right? So being able to make sure that the customer experience isn't impacted ahead of time is another place we've seen a lot of advancement in the past year.

Ruth Brown, Omdia (01:50:25):
So would you say that the day two operations remain probably the biggest challenge in terms of your perspective?

Warren Bayek, Wind River (01:50:34):
Great question. I mean, everything about this is a majorly difficult as we virtualize these networks. The reason day two operations, I think does stay at the highest level is, as we've heard many panelists say, this ecosystem is now many, many, many, many vendors in one solution. It's no longer kind of one closed box, here you go top to end, I mean, top to bottom. So making sure that all of these pieces fit together and that all of these pieces are going to interact correctly is becoming more challenging. And frankly, as someone mentioned earlier on an earlier panel, ORAN adoption has been slower I think than most of us would have liked. Let's be honest. And I think one of the big reasons is the perceived, and I do say perceived for a reason, the perceived complexities that are difficult to overcome for operators. But as we as a consortium work together more closely and do things like the ITEL lab, the digital twin idea, doing things ahead of time makes day two operations a lot more deterministic and a lot less complex for operators to deal with.

(01:51:53):
But I still do think day two operations remains one of the most difficult problems operators face moving to virtualize the opener in architectures.

Ruth Brown, Omdia (01:52:02):
Yeah, thank you. So we're going to move on and kind of pick back up on some of the points you were making, Kanika, about really agentic AI is making this massive explosion moving with some vast speed acceleration, but really this has got impacts into the way, I suppose, operators and operations are going to happen in the ran automation when we consider things like who is going to take the overall decision and really these are going to feed into some of these SMO frameworks. Now you sort of mentioned as well about the importance of working together, thinking about where you're putting the intelligence. So I wonder if we'll sort of segue and move back to you really just to talk about that and a little bit more focus before we hear from each of our panelists about what they think AI and Agentic AI is really going to do in terms of running evolution and run automation story in the future.

Kanika Atri, NVIDIA (01:53:07):
Yeah. I think from NVIDIA, we are a huge proponent of open everything, huge proponent of open source as well. In fact, at Barcelona on Sunday, we made a few announcements. We announced a first large telco model that was fully trained on network data, open sourced it. That's now available as part of the GSMA Telco AI initiative as well, available to everybody. With that, we also published guides like how do you take this LTM, train it with your own data and make it your own, make it available to everybody. Then we build things like blueprints. So for example, we launched two blueprints. One is for energy efficiency. One is for configuration management. Again, these are daily problems that every telco has to take care of. And these blueprints are like recipes of what data do you pick from where and how do you use this large telco model to create an agent that, for example, in case of configuration is smart enough to do complex trade-offs between KPIs and be able to continuously manage the parameters automatically such that your KPI is always met.

(01:54:24):
So what is that blueprint and that recipe to build the agent? Similarly, building agents and energy efficiency, you're going to see hundreds of these and we will open source all of them, not just open source the how-tos and the blueprints, but also the work that we do ourselves and make it available to everybody. And the same thing on the digital swim side, we have a platform called Aerial Omniverse Digital Twin that not only creates the entire physical replica of your environment, it actually simulates your rate tracing and your actual radio propagation behavior, it simulates your traffic, different UEs, right? And then now you can start running models. That entire platform is open source and that is how you accelerate the innovation and bring partners in. In our autonomous networks ecosystem, I can take 25 names and all of them who are very important partners for many telcos are building and using all these things.

(01:55:25):
So to me, that journey from just automation to autonomy is not that far and it starts with a very open foundation and we're a big proponen and we'll continue to make big moves and push everybody to build on these open components.

Ruth Brown, Omdia (01:55:44):
So when we spoke the other day as well, we spoke about data, and I'm going to move to Jensen after you, but we were talking about synthetic data and its role in supporting this development, perhaps you could sort of comment from the-

Kanika Atri, NVIDIA (01:55:59):
Absolutely. That's actually one of my favorite topics is we all have a challenge of how much data is available for training. And of course, every telco also is guardrail that data, very rightfully so. It's very hard to kind of open up that data to train. However, with synthetic data, even if I have, let's say, one real data of a site and how a particular radio behaves on that particular site, now I can take that same data and turn that into a thousand variants with simulation, creating that synthetic option saying, what if that scenario had 50 more trees? What if it had 10 more buildings? What if the actual infrastructure was something different? So you can take that same data and create a thousand different variants of it and use that data to train your models. And that idea of synthetic data generation, we actually build models to create synthetic data.

(01:57:01):
In fact, for physical AI, that the world starts with training and it happens in the simulation world. In fact, physical AI is born in simulation first and 6G will be born in simulation first. So that idea of synthetic data generation is absolutely critical and every telco will have access to that. Again, it's upleveled, so it's not about privacy and all those kind of issues, and that is the starting point for the next generation models we'll see.

Ruth Brown, Omdia (01:57:33):
Okay, thank you. So Tensor, I'm going to come to you because when we first spoke, you were talking about one of the critical changes for DT has really been getting your data organized at that foundational layer. Do you think that the synthetic data is a complement, is enough on its own? How will you work with both forms? Because obviously just in both your scenario and Lucy's scenario, Vodafone, DT working very differently, very different networks. Maybe you can comment on how this is going to work to support your automation journey.

Changsoon Choi, Deutsche Telekom (01:58:12):
Yeah. Maybe I can bring the two different angle, more like what we are now focusing and how we're going to try to get improved, and that is the first angle. And second one is how we're going to try to evolve the desire end to tend to network in a way that maybe we can give the better customer experience, but then how we can use the synthetic data for the second part. The first one is, I think this is what I mentioned earlier. If you look at what kinds of the data that we are actually collecting from the multiple, let's say multiple network, of course, coming from direct models to different vendor, I think most of the people have the similar experience. I think there are lots of the missing point, for example, PM counter and everything and semantic are absolutely different. And when it comes to even one more detail into the PM counter and how you're going to make a proper leveling and everything, sometimes I see that some of them are actually meeting and some of them having some different semantic in the labeling.

(01:59:15):
So this somehow makes MyLab really measurable. I really have to say when I such that this old AI and the run automation journey and how we going to use the synthetic data based on the reliable AI model, that was to super help to fill the gap between what we have and what wanted to have to train the proper AI model. So that is the first point that I really wanted to mention. And second one is, I think this is what Kenika, you have just mentioned. I mean, now when it comes to the 6G and bit of the AI native, we believe that maybe we're going to bring this AI agent more into the real time processing, not only in the same or in the not real time rig area, but also maybe going into the one level deeper, like a near real time, maybe next set or the app, it may be in the CEO and the DU in the user plan and everything.

(02:00:14):
So that would means that maybe we're going to have some multiple agent where one agent can stay in the CMO, but several agents might stay in CU and the deal maybe as a DM, maybe this is what we have to discuss earlier. And main purpose to bring that kind of the real-time multi-agent angle is we try to bring the more close loop to maintain and provide the quality that we have to provide to our customer. But then we always have a problem of the divergence when it comes to the close low, right? I think you cannot actually touch all of the close function in the real nettle, you have tested everything into the lab environment, but when it comes to the lab environment, we don't have any data that we can use for the This entire closed loop and testing and everything. So that is the area that we actually need really synthetic data to emulate and test all of the close loop in the real time.

(02:01:12):
And this is what we're going to actually push into the 60 direction as well.

Ruth Brown, Omdia (02:01:17):
Yeah. Thank you. So we're going to move on and you've nicely started the conversation along with Danique and also Warren did as well, really to talk about how we distribute the AI. And you've already spoken about centralized versus distributed these kind of D apps, et cetera. And Elisa, when we were speaking the other day, we were chatting about it. So I wonder if you can give me your thoughts on what you think is going to be really important around utilizing this in the future, especially when we start lifting and uplifting towards agentic AI, how this is going to make a difference.

Lucia de Miguel, Vodafone (02:01:59):
Yeah. I think when we are talking, I think different application, you have to look for the use case. And at the end of the delay that you need, it's more real time, it's more closed loop. And this is the part that we need to analyze because thinking in the apps, you are thinking something very quick. I think sensing, for example, the ISAC part. Okay. You have to look through, but later you have to balance, to open the network efficiency, the KPIs, the performance versus I think the part of the use case that you are implementing. So you need later some entity orchestrator that they can balance this, these AIgens. Later, I think you have the X apps that this is more near real time rig or the re apps that they can have more delay in the application. But you have all the time balanced performance, customer experience with, I think the use case that you are implemented, the infrastructure that you need it or the benefit that you are getting off it.

(02:03:09):
But I think the open to the AIgens, I think it's very, very interesting, especially because you can monetize.

(02:03:20):
If a different use case in terms of the operator, how you offer to the customer, what you can offer to the customers that is valued and they can give also the data. Because I think for the collaboration, the data, something important is you have to keep it the privacy for the customer. So it's generated real data with a mix of emulated and later the digital twins. And this is an angle for collaboration, sure. How we are able to create the data and require it for the AI agents. Farther of this collaboration, we didn't touch it, but it's the collaboration between all the ecosystem for the standardization because when you are talking of the DS, what is the interface that you need to open? If it's not standardized, at the end, you are creating some, I think, blocking solution or locking solution that you have it now in the radio.

(02:04:20):
So I think this is some angles for collaboration is you need to work in the standardization altogether and not only standardize, also implement it, and later to decide what are the best use cases.

Ruth Brown, Omdia (02:04:33):
Yeah. And thank you for bringing up the point about the other interfaces as well. I know we've spoken about the maturity, I suppose, and the recent announcements we're seeing on R1, but obviously they're as important, uplifting and ensuring that we fully mature these other key interfaces for the future. Gabriel, maybe you'd like to comment a little bit on that and how you're working with operators really to support what type of apps they're using where they're distributing their intelligence to build this kind of long-term path towards this agentic and full autonomy, as Kinika mentioned at the start.

Gabriel Foglander, Ericsson (02:05:19):
Yeah. I mean, I think I'm actually very optimistic, just to be clear, we're doing it now. So we have multiple level four certifications based on autonomous networks right now, specifically focusing on the energy function that you focused on. We have implemented intent-based energy management in the node that actually combines the setting a target for the user experience and then optimizing energy consumption around that. So intent framework will work very well together with AI-driven applications like agentic approach or something like that on top, but it's really important that you have an intent framework that can actually execute at the Genobe or the node level of the network because that's where ultimately the resource decisions are taken. So that's, I think we're a little bit running ahead of standardization, to be honest. I think that's one area where I would like to see the pace picking up on defining specifications for intents, but in the meanwhile, we can't wait for innovation, right?

(02:06:22):
So we're running with it. If you talk specifically about interfaces, I think O1 and the work on data models related to OR1 I think is going to be an important area to enhance if we can. I think it's a little bit too slow at the moment as well. So that's something that we're working with, you're smiling because you know. So that would be maybe one example of where we as a community try to accelerate.

Ruth Brown, Omdia (02:06:51):
Thank you. And Warren, I don't know if you want to comment or add on to what Gabriel was just speaking about.

Warren Bayek, Wind River (02:06:58):
Yeah, I'll probably build a little bit on what Gabriel said. I think one of the things we see is the importance of, and of course we all know synthetic data, we all need to use synthetic data, let's be honest, at some level. But what we're seeing a lot of is the importance of having telemetry. Once you've deployed these models and they're actually doing something in the physical world, changing the radio, beam forming something, being able to receive telemetry to be sure that a few things haven't happened, that a few things are happening. One is that the data that you were acting on or thought you were acting on is still relevant and is still accurate, right? So you kind of have data driven potential. So being able to pass that information up the stack, like you said in an open way, all of this has to be open so that the implementer of the AI application can be sure that the data that I wrote this algorithm for is still relevant.

(02:07:55):
That's one thing, to be sure that the telemetry is available so the data is reliable. A second really important thing is, okay, the data's reliable, but am I actually seeing the results that I thought? Am I getting the improvement in the real world that I was seeing in my simulated environment? So the same sort of thing, right? Bringing the telemetry up from the infrastructure to the application developers to be sure that the AI models that I built are actually doing the things I thought they were going to be doing. And then, of course, it never is, right? There's always something you need to tweak and change. Being able to change those models quickly, push those models back out so we immediately get another feedback loop happening, right? So that ability to implement and deploy these systems at scale in real time and get feedback immediately to be sure that we're doing what we think we do, we're doing in the AI space.

(02:08:52):
And I think that can only happen if things are open. And I'm glad that keeps being brought up. It's very important to all of us that we stay open and that we let everyone get into this ecosystem. And funny, one thing I'll say is I'm always asked, what's the killer use case for this? What's the killer AI use case for RAN? And I can give you my answer. I think everyone on this panel can give you three or four, but it's a little flippant, but my answer's always the same. It's that when Apple developed this, they didn't know what the ultimate use case was going to be, right? They couldn't give you the killer. I say the same thing in RAN. I don't know what the ultimate AI use case is going to be that changes the world, but I will guarantee you this. In five or 10 years, your lives will be completely different because of the AI implementations that people in this room develop.

(02:09:44):
Exactly what they're going to be, I'm not sure. But if we create an open ecosystem and let everyone in, a lot of really smart people are going to develop really cool applications in AI that will affect our lives using RAN in a way we can't even imagine today. So I want to be sure that we all buy into that openness. And I think being an open RIN, we do. But to me, that's the critical piece to AI that we have to always keep at the very forefront. It has to stay open. It has to be available to everyone.

Ruth Brown, Omdia (02:10:17):
Yeah. And a good point you mentioned about keeping open because obviously the more we get into starting to implement AI solutions, we can sometimes, unless we're keeping this kind of open ecosystem, get trapped in things that might become a little bit sort of locked down. Janssen, I'm keen to talk a little bit before we go to Q&A about, we've seen a lot of this year, especially at Mabel World Congress, Agentic RAN apps, so apps that are going to start calling on some of perhaps the R apps that we've traditionally started to develop. Is this something you want to speak on, you've been looking at in the lab and developing with some of your partners?

Changsoon Choi, Deutsche Telekom (02:11:12):
It is very clear that I think now we are trying to implement the AI agent in the OSS and then this domain orchestration layer. This is quite obvious. And we are also really pushing really a lot to develop up the, like I mentioned earlier, so in- house SMO together with the not real time re. And then once the not real time rig is ready, then we try to put this AI agent as an item, more or less like a containerized platform, and then try to use this one as early as possible. This is the clear goal that we're going to try to do. But then, I mean, the question even to ourself, which we have not had the proper answer is, okay, in case that this agent might need to speak with another agent and this agent might stay in the certain real time area and how we're going to communicate, how we're going to enable the certain communication between this multiple agent.

(02:12:11):
I mean, the traditional A2A, I really don't know whether this is going to really work. And then we can also use the MCP platform to make it happen or to code the different natural function. But again, I mean, this is what we need to figure out, how we're going to implement in the top of the order framework. So I would rather say that still there are several things that we have to work together with the many different partner to make it happen.

Ruth Brown, Omdia (02:12:37):
Gabriel, I see you sort of nodding in agreement with Henson. I don't know if you want to comment. Yeah,

Gabriel Foglander, Ericsson (02:12:43):
It resonates. I think we have a couple of our apps in the portfolio that is implementing agentic procedures. One is, for instance, for red tilting of the antennas, which just a single percentage point can have a huge impact on the sell edge, right? So that's where we do it. I think one of the things that I'm curious about is

(02:13:03):
This is not just about tech, actually. It's a lot of how do you transform processes to trust automation. And I made the point, I think both of them nodding. This is, I think since we're dealing with critical infrastructure and even more so going forward, just having the way of transforming the operations into auto autonomous first framework and providing what you were mentioning before about observability, transparency of decisions, how can you unblack box what an agent is actually doing. Once we start to get ahead on the technology problem, I think this will come more to the forefront as well in terms of scaling and deploying.

Ruth Brown, Omdia (02:13:48):
Lucia, do you want to come

Lucia de Miguel, Vodafone (02:13:49):
On in? Yeah. And in the case of Bodaphone, I think we have two angles. So it's common to have it AIs in the reaps more in the centralized part, more for the real networks. And later we are doing some POCs trials, more linked with the concept of the XAPs, the apps. It's not something new. I think some years ago we did with SAN, I think Spectrum, I think efficiency, and now we are beginning to test in ISAC. So more the sensitive tracking, I think drones, people, and with the concept of the apps, but I think it's not easy to move to the real network. So now we are doing more the trial, saying pushing the industry to have more standardized and to be able to put close to the customers.

Ruth Brown, Omdia (02:14:44):
Brilliant. And I've got an eye on time before I get to hold off for not asking questions. I'm sure we've covered a lot today. I know it's always such a short amount of time we could talk about a lot, but I'm keen to take questions from the audience. So please feel free to raise your hand if you've got a burning question about automation. I'll try and look out for you and we'll get a mic to you. We've got a question just at the back here. If you'd like to stand up, we'll get a mic to you. Do you want to stand up and then we can hear you? Yeah, thank you.

Louis - Audience member (02:15:17):
Hello. Yeah, good morning. My name is Luis. I am with Brown iPods from Germany. Thank you so much for your comments. I mean, it was very interesting for me to hear the different use cases that you guys are working very hard on using AI power autonomous networks towards improving the quality of experience of your users. But for some of us, like I'm a network engineer, the idea of having an agentic AI fully closed loop system that is performing changes in the real network might be a little bit terrifying. But the question is, how can we build trust on these systems, especially in the transitioning process when moving from the lab and the emulations onto the real network?

Ruth Brown, Omdia (02:16:13):
Okay. So maybe we start off with Kanika, because I know you mentioned digital twin and then we'll chat with the other two.

Kanika Atri, NVIDIA (02:16:19):
By the way, this is a great question because trust is the biggest challenge today on AI adoption, right? And how do we overcome that? And there are different levels. First is, of course, digital twins kind of tools. They give you a little bit more confidence that what the agent has recommended, it has validated, you can see that it does well, it will give you maybe 20% more confidence, right? I'm just talking as a human if I were this network engineer and trust me, I've been one. The next part of it is the beauty with AI is that it learns. So it may get it wrong this time, but the next time it's going to take that data back and there is a human that actually tells it that, "Hey, you did not get this part right." And it learns. And that cycle is so fast that very quickly that trust will begin to establish.

(02:17:15):
And even if I take out this network automation for a minute and talk about technology, any new technology, it goes through three phases. First, people invest in it, right? They're like, "All right, let's find out what can it do. " Then they start seeing some interesting benefits and some ROI. So they start adopting it. So let's say now I deploy it for anomaly detection or I deploy it for some fault management, right? So that's adoption. Eventually, when trust builds, they come to the third stage, which I call the adapt. Their entire processes, their people, their behavior adapts around that technology as if there was no other way to do something. Our children are born, and for those who are going to be born, are all going to be natives. Our children were digital natives. They only knew one way of looking at photos. That's called adaptation.

(02:18:16):
There's no other way of doing it. We're not there yet in telcos networks, nowhere close. We are somewhere between invest and adopt. And from going from adopt to adapt, we have to cross the bridge of trust. So eventually there are lots of technology tools that enable you to do that, but I just want to acknowledge that is a very big challenge and the collectively we think we can solve that.

Ruth Brown, Omdia (02:18:44):
So do you want to comment about how you're maybe from lab to building the trust?

Lucia de Miguel, Vodafone (02:18:50):
Yeah. I think from an operator point of view, I think that a lab is not something that you emulated. So you can have different laboratories, but the final one is the most close to the reality. So you are not able to put something in service without testing with the same data or with the same configuration that you have it later in the real environment. And also the approach is that normally you put in a small piece of your network with some traffic and you are able to do some, I think, roll back. So this is basic. You do some changes, you see that I think it's seamless for the customer or at the end is that you get the benefit that you expect it or not. If you don't get it, you do the rollback and begin again in the laboratory to test it what is failing.

(02:19:46):
But it's a normal practice as new technology. So you are not putting something and do it without any observability, any control, because especially for our operation guys, it's impossible to do it.

Ruth Brown, Omdia (02:20:02):
Thank you. Have we got any other questions? We've probably got time to squeeze in another ... I'm just sorry, I can't see with the lights. I think we've got another one at the back if you'd like to stand up. Thank you. Hi.

Sahana - Audience member (02:20:14):
Thank you. I'm Sahana from Fronhofer. I head a department related to wireless systems and standardization. And one of the challenges that I've seen is when we talk about openness, as much as it is necessary, it's not easy to adopt or easy to let everybody into the world of open data or open interfaces. So what do you see as the roadblock to it? So if we want to be able to propose a better solution or better trained algorithm, how can you or as the telco operators or as OEMs, what kind of information can be made available to public or to adopters or research institutes to come and propose something new or something better or to validate it? Is that feasible or does it hurt security and privacy on the other hand? Okay.

Ruth Brown, Omdia (02:21:12):
Would anyone like to go offer to answer that one?

Gabriel Foglander, Ericsson (02:21:18):
Yeah, it's a big question to answer,

Ruth Brown, Omdia (02:21:20):
To be honest.

Gabriel Foglander, Ericsson (02:21:21):
I think you're happy that you didn't get it. No, no, I think it's right to respect the challenge there. I think it is not easy. When we talk about tremendous amounts and volumes of data as well, besides the fact that RON is an incredibly complex system in its own, and then on top of that, you have every operator has their own topology, their own configuration settings, et cetera, which by nature means that you get into exclusive data sets. And I respect your previous comments on divergence between PM and so forth. So I think the answer lies somewhere and finding a reasonable level where you can get commonality, but still allowing for additional divergences if it doesn't matter for the end goal. So as long as you're not compromising what you want to solve, you can still allow for flexibility in what you have underneath because standardizing all of this with what I just said is not taking us to the right objective of excelling for our customers in the end.

(02:22:28):
So it's not easy, but yeah, that's my take.

Ruth Brown, Omdia (02:22:33):
Thank you. Right. Sadly, we are at the end of time now, but I'd like to thank our brilliant panelists for their candor, for showing their use cases. So please let me thank Warren, Gabriel, Chanson, Lucia, and also Kanika for sharing their views. Thank you very much.

Kanika Atri, NVIDIA (02:22:49):
Pleasure to be here.

Ruth Brown, Omdia (02:22:56):
Thank you. Thank you. Thank you. Thank you. Thank you so much. Back over to Gabriel. Thank you,

Gabriel Brown, Omdia (02:23:07):
Ruth, and another panelist. So we come to the end, final session now. We have 30 minutes on ORAN and 6G. Obviously, lots to get into, and so I'm going to ask our panelists to come to the stage as and when they're ready. We're going to hit this. We have FEMI here. We have Rob, I think Lorenzo somewhere. Here we are in fact. And Brad, please join me in welcoming the panelists. So look, as before, I have a bunch of questions, but do bring your own as well to the party. So terrific panel that lots to kind of chew on and follow through here now on 6G. Let's start with some introductions. Femi, perhaps Rob, we know, but it can restate. Femi, please.

Femi Adeyemi, 1Finity (02:24:12):
Good. Still good morning, everyone. I'm Femi Ademi. I'm from Onefinity, Ifujitu company. I'm part of the RAN systems responsible for our strategic product management. It's good to be here.

Gabriel Brown, Omdia (02:24:31):
Rob, we know you, but just restate for anyone who joined late. Sure.

Rob Soni, AT&T (02:24:33):
I'm Rob Sony. I have RAN technology architecture planning standards for AT&T.

Lorenzo Casaccia, Qualcomm Technologies (02:24:39):
And I'm Laurenso Kazachan. I run the Qualcomm technology standards department, so on the radio standards and several other areas of standards under my purview.

Gabriel Brown, Omdia (02:24:51):
Fantastic. So we couldn't have two people with standards right under their remit. It's perfect. Lorenzo, I'm going to start with you. Perhaps you could just, you're head of standards at Qualcomm, right? So you're going to have a good set of views here. And you're in ORAN alliance at Qualcomm. Just set out a little bit the two timelines we're dealing with, I suppose particularly the 6G timeline as we understand it today.

Lorenzo Casaccia, Qualcomm Technologies (02:25:13):
Yeah, of course. Thank you. First of all, thank you for inviting me and having me here and thank you for the audience for listening to us. So the 6G timeline is more or less well organized in the cellular industry. Free GBP is studying 6G concepts as we speak now is the so- called study item phase. And there is a tentative conclusion of both the study and the work item. Work item means where the specification is actually produced for the beginning of 2029. So first quarter of 2029, we should have ink on paper on a 6G spec. Then they're very important. There has been a division of work between FridgeVP and Oran Alliance where certain areas of 6G will be the responsibility of Oral Alliance, very importantly, the frontal, but then also the so- called SMO, probably many people are familiar with. Specifically for the frontal, which is very important because it touches the air interface in mid of this year, so just in a matter of couple of months, or an alliance will start also a project for 6G.

(02:26:27):
And therefore, from more or less June of this year, you will have these two parallel tracks of 6G projects for the open frontal and the organ alliance and the 6G study in FridgeVP. They will run parallel, and of course, they will have to work with each other under the general framework that it was discussed last year.

Gabriel Brown, Omdia (02:26:48):
And if you could just back in a little bit to the 6G timeline, you're talking about first quarter of 29 as the sort of first spec date, I guess. In the interim period, what are some of the key decision points and things we really need to be looking at?

Lorenzo Casaccia, Qualcomm Technologies (02:27:04):
Yeah, so this year there are very important ... This year is the more researchy phase, so where people look at fundamental concepts for 6G. So from a certain point of view, it's more the areas of researchers, but at the same time is where there are really fundamental decisions made. Let me make some examples. So now, of course, there is a general momentum on the industry to stay in a sort of continuity path with 5G. So 6G will continue to use LDPC as a data channel coding. However, there is a discussion on improving LDPC channel coding for the purpose of reducing complexity, especially on the devices. I mentioned this is a topic very close to our heart. LDPC channel coding, especially for very high debtor rate, has a lot of impact on the hardware, both on the network and therefore in the ON Alliance purview, then on the devices.

(02:28:06):
So this is very important. This will be discussed in the next few months. Similar discussion, modulation, similar story. There are already modulations in 5G will be reused, but can we do something better to reduce the total cost of ownership? Third area, spectrum. Can we have larger carrier bandwidth, 400 megatz, for 6G that would enable new use cases like sensing. And then finally, there is an overarching big discussion very much relates to the prior panel as well on whether to have a new core network or to have just an evolution of the 5G core network. There are different points of view that relate to the role of AI and compute that would have in a 6G network. These are, I think, four big milestones that give you an idea of what gets discussed in calendar year 2026.

Gabriel Brown, Omdia (02:29:05):
And just leaving aside the core network on those RAN issues, do you feel there's ... How would you assess the level of consensus within the industry as it looks?

Lorenzo Casaccia, Qualcomm Technologies (02:29:16):
I think there are medium. I would assess this as medium. I think there are very well justified differing points of view. So for example, let's take the core network. I'll mention the core network because it's a little bit further away from Qualcomm's area. We provide now also Silicon for the radio access network, and of course for devices, but not for the core network. So I feel we tend to be a little bit more neutral party. And I think there are two very legitimate points of view. If you look at the adoption of 5G standalone, 5G standalone has a very high level of adoption in China, for example, and a fairly low level of adoption in many other parts of the world. And so it's perfectly logical from many other parts of the world, in my opinion, to say, "Hey, I don't want to do a new core network for 6G.

(02:30:13):
Let me just continue to evolve the 5G core network because I haven't even fully deployed standalone, so I don't want to change the engine midflight." That's very rational position. I'm sure some colleagues here on stage would agree with. At the same time, there are other parts of the ecosystem, especially in China, they say, "Hey, there is this general ... We already have had a standalone for a number of years. There is this general trend of compute integrating with the telco network. How about we redesign the core network in order to accommodate an integration of compute workloads and telco workloads?" I think that is also a very rational point of view that can It comes from some parts of the Chinese ecosystem. And I think on that, I don't think there is an industry consensus at the end. At the end, it's going to boil down to the real experts to find a compromise in the details.

(02:31:14):
So I say consensus is medium. I don't think it's anything particularly bad at the moment because I think it's a creative lack of consensus. The lack of consensus is also due to the fact that there are different business models and different ecosystems at play. So it's all good from my perspective.

Gabriel Brown, Omdia (02:31:35):
Yeah, inevitably a learning phase. I think the network computing is only in China. I mean, there are definitely Western advocates of that model as well. Rob, at AT&T, I invite you to reflect on that. First of all, just scope a little bit how you're thinking of 6G just in the broadest terms.

Rob Soni, AT&T (02:31:55):
First of all, thank you for inviting me to come back here on stage and to talk more again. And again, look forward to talking to you all and thanks to the ORAN Alliance for encouraging a discussion about the evolution of cellular and what we can do overall. I once heard our COO say in front of a very large audience in Sweden, "We don't want 6G." Did that wake anybody up?

(02:32:22):
So this room feels a little quiet and a little stale, so I'm going to just try to see if I can't get you guys to wake up. It gets very technical, very deep, very quickly. And we always say to ourselves, "Why would somebody say they don't want 6G?" I spent a long time thinking about that on the plane back from Stockholm saying, "Why in the world did he say that? What was the reason for that? " Because operators are busy. What are they busy with? Modernizing and refreshing the network, adding new spectrum, layering on new services for the technologies they have today. And then every once in a while, our good friends around the world show up and say, "Hey, we got this shiny new box for you. We want you to go and put it there. And that's called XG, whatever X is. " And then we get into a situation where we say to ourselves, "Hey, we haven't finished deploying the spectrum that we want here.

(02:33:11):
We haven't modernized and moved to cloud. Now we're being pressed to go to AI native or AI." How many different AI adjectives have you heard this week?

(02:33:22):
I didn't know that you could have AI integrated, AI native, AI sensed, AI built, AI born. I don't know. I heard everyone this week. So I think we're in a space where we're all being asked and challenged to figure out how AI can transform our network today. You heard it all say a new model gets dropped every four to six weeks. We see last year was the year of generative, this is the year of Agentic. What is next year? So when we have these discussions about 6G and about changing the mother code rate on LDPC, and I'm not picking on you, please don't think that way. Even though we don't support that. You know that, but your team knows that. But I think the simple thing is we are looking to be software driven. Not because we want to choose a G, which is software driven, because we are moving the fundamental value stream away from hardware into software.

(02:34:17):
The telco world needs to embrace this and accept it, make things more programmable. And this is exactly what ORAN does for us. ORAN makes the network open and programmable. It allows us to bring assets best of class from different folks. So table stakes for us of 6G, of course, is open. Table stakes for us around 6G is it should be AI native. Table stakes for us is it should be scalable interface. I don't need to have the same conversation with our product folks. Oh yeah, we're going to introduce IOT in the last seven years of a G. Okay. We got CADM in what? 2018, 2019. We're getting REDCap here in 2025. Okay, we'll get EREDCap in 2027. Why is it not there on day one? So we want a scalable interface. That's what we're looking for. We don't want to get that box either of features.

(02:35:12):
We don't want to get everything all at once so that somebody can sell another chipset. Sorry, I feel like I'm picking on you.

(02:35:19):
So in the end, we can't have a situation where features and capability are delivered as a big box and then people struggle to consume them. We can't be in that situation anymore. We need to break it into pieces. Pieces we can consume. We need partnership from the device makers, partnership from the infrastructure vendors so that they deliver features continuously. That's the way it works in other domains. Why can't it be that way for us as well? Why do we need to drive everything towards five year or 10 year cycles of innovation? Yes, we get a new release every 18 months. I've had this discussion. Yes, new release every 18 months. Yes, we're getting interim drops. But even the big release, it's always the momentum around the big release to drive big scale investments. We know when we look at this and say, "Hey, here's the opportunity for insertion to change and make big decisions." Guess what?

(02:36:07):
We don't make decisions about what vendors enter or exit our network based on a G boundary. And I think very few operators actually do. So stop playing that game. Focus on innovation and delivering it continuously. Build the best product you can today, not for five years from now. That's what your focus should be on. That's what we've been looking for.

(02:36:28):
Just to finish.

Gabriel Brown, Omdia (02:36:30):
Yeah, of course. Sorry.

Rob Soni, AT&T (02:36:32):
We're trying to wake everybody up. We do want AI native. We want the continuous evolution, but we don't want to wait for a 6G for that, but we do expect AI innovation to be present. We've heard a lot of discussions about how ORAN and the ORAN Alliance could participate in Agentic and change things there. And then of course, we're excited about ISAC. And for a long time, I was kind of scratching my head saying, "Is ISAC only a 6G technology?" No, it's not. We're using the RAN as a sensor for spectrum sharing. We're looking at it specifically for environmentals. It's not just about object detection. And by the way, even object detection with the RAN can be done today. We don't have to wait another five years to do it. And we have to figure out what's the efficient way. And then finally, what's the business case for sensing and communication?

(02:37:20):
Do we have one? Do we agree it's worth dragging the whole cellular industry into this morass of saying, "Hey, you now need to build capability, whether it really affects you or not to support a sensing system in every device, in every bit of infrastructure." But at the same time, we're very interested in what 3GPP will produce in terms of new technologies. Very interested in seeing what ORAN can do to layer open underneath it. We fought quite a lot with a lot of people, some of which are in this room about whether we should even use the word open or even have 3GPP refer to ORAN specifications. I don't understand that. For every operator I talk to, 6G table stakes should be open, but some people don't like that. Hopefully, I'll stop there.

Gabriel Brown, Omdia (02:38:09):
I want to bring FEMI in, but before we do FEMI, let's just get right to that then, Rob. The idea was ORAN came after 5G more or less in the timeline. 6G was going to be a chance to make open native. I want to put it that way. I don't know if that's quite the right term. What we've ended up with, it appears from the outside is kind of two tracks now, a closed proprietary fronthaul track and an open track. Is that the right understanding or ...

Rob Soni, AT&T (02:38:38):
I want to respect my good friend here. No, it's fine. He's actually an expert on open front hall. So I'll let him answer the question.

Femi Adeyemi, 1Finity (02:38:45):
What was your question? You are directing it.

Gabriel Brown, Omdia (02:38:49):
It looks like as we go to 6G with Openfront Hall, we've ended up with a sort of non-open track, the 3GPP, and then Open is being offloaded almost to ORAN. I mean, that's an unfair characterization probably, but it looks a little like that from the outside. So we ended up now with two tracks when the whole idea with 6G even more broadly was to streamline and have fewer options. Well, right away, we've gone back to two.

Femi Adeyemi, 1Finity (02:39:16):
So as a community, we need to be consistent. As Rob said, we fought quite a bit with the number of people in this room to have open interfaces, to have standards that are open, so that we will have a healthy ecosystem. And so as we go into 6G, we actually believe that continuity is very, very important. We need to maintain that so that we're not ... For network operators, we don't complicate their life. So there still needs to be a lot of work to be done from a standards point of view to actually aggregate all of that together so that the industry actually benefits. When we build a platform, and then instead of adding to it, we are tearing it down, it complicates network. You heard them, simplicity, easy to consume, open interfaces. That's where network operators benefit.

Rob Soni, AT&T (02:40:17):
Simple. We don't want two products. We don't want vendors building two products.

Lorenzo Casaccia, Qualcomm Technologies (02:40:21):
Yeah. I think to your question, maybe it's yes, it's a little bit unfair because I think there's not really two tracks the way I look at it. You have to look at how it was for 5G. Actually, for 5G, yes. Maybe of course in the mind, in the mental model of many of the people in this ecosystem, things were clear, but on paper, things were not so clear because remember that for 5G, the REU does not exist in 3GPP. So I happen to.

Rob Soni, AT&T (02:40:53):
There are vendors today who don't build an open product. They only build-

Lorenzo Casaccia, Qualcomm Technologies (02:40:57):
No, but I mean, in 5G, it was even worse because from a philosophical point of view, in the 3GPP specs, the RU just does not exist. It's like a UFOs and it's just not there. But for 6G, there is. So in 6G, the industry recognizes that there is an entity called RU. And all it says is that the interface between DU and IU is specified by Organ Alliance. Now, of course, if you were asking me personally, even not even my team, it's just me personally, we could do everything in HPP, but there is a certain inertia. And so now the way I look at it is just division of label. It's recognized that this interface exists and okay, it's just done by another group and everybody has to work together. To your point, of course, some vendors do not build that interface, but in a certain sense, you cannot force all vendors to implement all the interfaces in the standard.

(02:41:58):
I

Gabriel Brown, Omdia (02:41:59):
Think, Rob, I think I heard you say you don't want two tracks, but just a chance to amplify that louder. Is that what I heard right?

Rob Soni, AT&T (02:42:06):
We have two tracks.

Gabriel Brown, Omdia (02:42:07):
Yeah. But you don't want to implementation technically. Some

Rob Soni, AT&T (02:42:10):
Vendors may feel compelled to build a closed 6G solution ahead of an open one, and it doesn't matter where they live. They may feel compelled to do so. And yeah, there was a lot of fighting over the specification. And did we end up with a requirement on an open front hall? Not really. Did we end up with a definitive reference to the ORAN specification that says the front hall shall be defined by ORAN? No, we didn't. So I think that's the challenge. There will be closed product for 6G. There will be.

Gabriel Brown, Omdia (02:42:46):
I guess there's also though a choice, a decision for the ecosystem and suppliers and so forth to make as well to kind of ... Even if it's not in 3GPP, in a sense, the weight of opinion will kind of carry today. I

Rob Soni, AT&T (02:43:03):
Think the vendors that we work with are all committed to open, but what I worry about is that the vendors that we work with will feel compelled to compete with the vendors who potentially might go faster with a closed product or believe or position it as competitively different, that they would ultimately develop both a closed and an open product. And that's ... We had that question in the forum I was in before about fragmentation. That's real fragmentation. It's actually two products.

Gabriel Brown, Omdia (02:43:35):
Can you have any questions out there? Gentlemen, in the third row, thinker gentlemen.

Ren, MIC - Audience member (02:43:49):
Hi, everyone. And thanks for the speakers for sharing so much things. And speaking to the 6G, I remember that one speaker was talking about the spectrum. So my question is, what kind of spectrum you think that we'll use in the 6G? Yeah.

Gabriel Brown, Omdia (02:44:08):
Could just give us your name and your affiliation.

Ren, MIC - Audience member (02:44:11):
Oh, sorry. My name is Ren and I'm working for the MIC, which is an institution to work for government in Taiwan. Yeah. And we do the research. Yeah.

Gabriel Brown, Omdia (02:44:25):
Okay. Thank you. Spectrum Femi, you have a- Yeah.

Femi Adeyemi, 1Finity (02:44:29):
So there are a number of spectrums being defined for 6G. Primarily, you will have between seven and 24 gigahertz, as well as hundreds of therahats that have been defined. From an infrastructure point of view, now we have to worry about building radios and faster processors to be able to keep up with such spectrum. But there are spectrum today that has been defined for 6G, seven to 24 gigahertz specifically, and then 100 to 300 terahats to be used.

Lorenzo Casaccia, Qualcomm Technologies (02:45:11):
But I think in practice, there is a focus on more tangible piece of spectrum, which is upper six to seven gigatz. So if you look, I want to say with the exception of the US, there is a general momentum on the planet for the so- called upper six. This is from 6.4 giga to 7.1 gigatz, or also going to 7.4 gigat or 7.7 gigat. So we are from mid six to mid seven. So that's the case for China, that's case for India, it's case for Brazil, that's already three billion people. And there's an upper six also for Europe. Then there is Japan has had some issues to identify that spectrum, but not because of this global momentum, I think they're turning around. The US, there is the difficulty that this spectrum is already going to wifi, but the US is now working very hard to look for spectrum, especially 4.4, 4.9.

(02:46:25):
So in that sense, the US is a little bit outlier in this global momentum, and that leaves Korea that I think has a lot of problems, which I'm not sure how they will really solve. I'm sure my Korean friends will figure something out.

Rob Soni, AT&T (02:46:45):
Yeah. I'll just give a little bit of the US perspective and build on what Lorenzo just said. I mean, admittedly, it's a challenge to identify spectrum, especially this requirement around 400 megahertz contiguous that some of the vendors are pushing for, and we have had them advocate to us to advocate back to the US government to open more of the seven gigahertz band. Currently, they've identified 250 megahertz. The problem is it's not just 400 megahertz. There are three large carriers in the US, so that's 1.2 gigahertz you need to identify. There's no 1.2 gigahertz. Actually, ironically, in the regions where there are healthy operators who have the ability to invest in capital and there's 1.2 gigahertz available, there's not a lot of places in the world that's true.

(02:47:41):
So even though we say, okay, there are places where it's possible, remember the seven gigahertz, if we're going to actually turn it into useful frequency for macro use, it's going to be expensive. We have folks coming to us and telling us, "Oh, it's going to be 1,024 antennas. It's going to be 256 active elements. It's going to be a lot bigger and then span a lot more bandwidth than we have today." We're already paying a lot for a massive MIMO and we already have significant wind load on our towers for them today. To put those up is a real operational challenge. We think it's going to be difficult to find and identify around the world and have a golden frequency for 400 gigahertz. We also think that there's the industry's overselling the benefit of seven gigahertz as a useful frequency. We're an outlier presumably in the US, but there's a lot of spectrum that's going to be available in the timeframe that 6G could be launched between upper C-band, between 2.7 and 2.9, between 4.4 and 4.9, with potential relaxation and potential reauxion of CBRS.

(02:48:47):
There are other places to go perhaps other than just simply focus on seven gigahertz. So we really want to change the conversation around that for that reason to say, let's not couple 6G so strongly towards seven gigahertz. Let's not make that a requirement because what happened in the US the last time is everyone said, "Oh, I need large contiguous bandwidth in order to support 5G." So some operators went off and deployed millimeter wave. AT&T was one of them.

(02:49:13):
Did it provide scale? Did it really fundamentally change user experience or did it just cost a lot of money? I'll let you be your own judge. So I'll

Gabriel Brown, Omdia (02:49:23):
Leave that. Even if you don't have three times 400, there aren't many markets that are going to have that. I mean, definitely not in Europe easily. That doesn't necessarily mean it shouldn't be specified as within standards, does it? I mean, you're going to have many channel width options. So it's not a-

Rob Soni, AT&T (02:49:42):
Let's have more options we don't implement. That sounds good. The other-

Gabriel Brown, Omdia (02:49:46):
Sorry,

Rob Soni, AT&T (02:49:46):
I'm picking you down too.

Gabriel Brown, Omdia (02:49:47):
Well, I mean, someone could, that itself isn't probably the worst option to have. That's not going to fragment things too badly particularly. What I wanted to, Lorenzo, you can come back if you like, but also maybe whoever wants it, the question's a good one because you answered it as fresh spectrum essentially. Rob then identified some other fresh spectrum, but should we not also be thinking about the value in the spectrum already allocated for 4G, 5G? I mean, you could argue, and technical people argue a lot about it. It's under exploited. There's been under innovation in some of these FDD, midband FDD, low band FDD that we could do a lot, lot more with. Might have more impact on us than a seven gig Ultra MIMO.

Lorenzo Casaccia, Qualcomm Technologies (02:50:31):
I think you answered yourself very good. I would have not been able to say it better.

Gabriel Brown, Omdia (02:50:36):
Okay. Well, I'll turn that into a question again.

Lorenzo Casaccia, Qualcomm Technologies (02:50:38):
No, it's a very good point on FTD. Actually, I forgot to ... I just went on a tangent on the fresh spectrum, but you're totally correct on FTD. Absolutely. During 5G, you can argue that FTD was maybe not really a focus for 5G for better or worse. It's not even clear why from a certain point of view, but it's a fact.

Gabriel Brown, Omdia (02:51:00):
But all this uplink challenge now and there's

Lorenzo Casaccia, Qualcomm Technologies (02:51:02):
All this fallow spectrum. And I think in general, the industry now for 6G, we need to take a look again for FDD.

Gabriel Brown, Omdia (02:51:15):
Mari, over there.

Marie Silby, National Spectrum Consortium (02:51:16):
Hi, thanks. Thanks for that. I'm Mari Silby with the National Spectrum Consortium. So I heard a very interesting division there, which is 6G is going to be seven to 24 gigahertz, and then the operator on the panel saying, "Gosh, we don't even want to go above six gigahertz at this point. We've got a lot below that. " Arguing how much better you can make use of the existing work under six gigahertz. My question is, if you start to go above that and yet the operators only want to do things in software, how do you reconcile the fact that ultimately you're going to move above six gigahertz? How do you reconcile that with the hardware investments that will just necessarily have to come with that?

Gabriel Brown, Omdia (02:52:02):
Fannie.

Femi Adeyemi, 1Finity (02:52:05):
It's a good question in the sense that I will not build anything that the operator will not buy. So the conversation starts with the operators. What do you have? Are you going to refarm? Do you want to do 6G? How much wind blow do you currently have on your antenna? What benefits your network? That's what I will build. So the standard bodies, they make standards, but at the end of the day, it needs to be deployable. It needs to be software. It needs to be open so that the whole industry can benefit from it. If we define non-contigured spectrum, which makes my life difficult to build, then we have problems. So we work in conjunction with operators to build what is deployable, what is manageable, what is easy for them to consume. And also, it's a global standard. So you will have from left all the way to right in terms of what is available.

(02:53:08):
That's what I define. But from different regions, we have to look at what is it that we can build that others can consume.

Rob Soni, AT&T (02:53:17):
Maybe just a comment. I mean, look, the ecosystem has benefited from universality of spectrum outside of the US. So classic definitions for GSM were in common bands, classic definition for band one, band three, et cetera, for UMTS benefit outside of the US because those became canonical as the bands for 3G. We didn't really have that in 4G. There wasn't an agreement on that. We definitely didn't have it in 5G. And so now again, I think there's this pressure to say, "Oh, we need to have a golden band for 6G, otherwise we won't gain the scale deployment." And the TCO at the moment for seven gigahertz looks quite challenging in terms of the deployment and the management of that. You asked about software and it's a fair point to ask, okay, if an operator just keeps telling you software driven, how am I supposed to address a new band?

(02:54:09):
You need new hardware to address a new band. That's fair. And so seven gigahertz will come with new hardware, and that hardware, as I've just mentioned, is going to be expensive. What it also I expect is it will change the base band as well. It's not just going to change the radio because I expect that the air interface definition will shift in a way that probably will break hardware and especially somebody wants to get 5X the peak rate. So for him to get 5X the peak rate, that will probably change hardware, if not devices anyways. So from that perspective, we do believe in hardware needs-based hardware refresh. So if it's a new band, probably need it. Do I need it for 4.4 to 4.9? I don't think so. I mean, radios will be new. Baseband, hopefully not. Upper C radios will be new, but baseband hopefully will not.

(02:54:58):
And for us, that's the big investment thing that we watch and watch carefully is that we change actually, ironically, we change baseband a lot more frequently than we change radios today because we break things in standards.

Thomas Lips, Deutsche Telekom (02:55:13):
Thank you.

Gabriel Brown, Omdia (02:55:14):
One right at the back, two at the back. We're going to be quick though because we've got two minutes.

Simon, Huminis - Audience member (02:55:20):
Hello. I'm Simon from Huminis. We are building digital twins for SafeAi agent deployment-

Gabriel Brown, Omdia (02:55:26):
Sorry, I can't quite hear you very clearly. Could you just- Oh,

Simon, Huminis - Audience member (02:55:29):
Sorry. I was introducing myself. I'm Simeon from Humanist. We are building digital twins for Safe AI agents deployment on the RAN. And I wanted to ask you if you think if 6G has already defined how much computing power will be available on the edge and what type of computing power, if it will be GPUs or if it's still up in the air.

Gabriel Brown, Omdia (02:55:52):
That's a big question for one minute, but-

Femi Adeyemi, 1Finity (02:55:55):
Can you repeat the question? I

Gabriel Brown, Omdia (02:55:57):
Hear

Femi Adeyemi, 1Finity (02:55:57):
Him very well.

Gabriel Brown, Omdia (02:55:58):
Well, I've got part of it. Okay. Is it already clear in 6G how much computing power will be needed at the edge? Some version of that.

Femi Adeyemi, 1Finity (02:56:06):
Okay. I mean, a lot of things are still being studied. For innovators in the room, in my mind, 6G creates very good opportunities, whether that's if you want to build hardware, even do studies, software. We've just scratched the surface on what 5G is doing today in terms of MIMO, in terms of apps, in terms of small. So 6G just gives us the opportunity to be able to do a lot of that. But to his question, a lot of that is still under study.

Gabriel Brown, Omdia (02:56:42):
Yeah. It feels like the industry's not re-litigating exactly, but going back over the edge discussion again at this point with the-

Rob Soni, AT&T (02:56:51):
It's been definitely being revisited. And I think people are looking today to revisit edge processing, edge compute for inference way ahead of a 6G era, way before it would be standardized via 3GPP. And it's an open question as to whether it needs to be standardized in 3GPP, or it's something that could be handled in a forum like ORAN instead. And that might be logical itself. I think we need to understand a bit better of is there really a set of use cases because edge compute Mac goes back to the 3G era and we don't see wide scale deployment. A lot of companies have tried to succeed at this. AT&T tried well before my tenure to come up with an edge compute architecture and scale it and actually be able to sell it. And unfortunately, like many companies and it didn't scale. So the real question is, is it going to be different this time around?

(02:57:44):
Is there going to be value in it? Definitely you've heard a lot of this from one particular company about how they can sell more chips to use for that purpose. But I think that does create a question of, do we actually have the use cases we need for it?

Gabriel Brown, Omdia (02:58:02):
Lorenzo, let me come to you quickly on this. There is clearly a standards and architecture implication of this because let's say service-based architecture to the RAN, say you're running a sensing application in a way is a little bit person's question. It's an edge application. So it's got to be kind of thought of at this stage of-

Lorenzo Casaccia, Qualcomm Technologies (02:58:20):
Yeah, I think there are two different things related to the computer at the edge. One, I think is the more difficult question. The more difficult question is to, of course, find business cases for operators. In our view, there are use cases like sensing. If we believe that you can have many more devices that produce sensing, they sense data in the uplink, you need to have very low latency to detect drones at the border or things like that. Then that creates demand for computer at the edge. Does it apply to all operators in all countries in all scenarios? No. Will it apply to some operators in some scenarios? I think yes, of course already you go in some countries that applies. Then there is the easier question, which goes back, I think, to the previous panel, which is a simpler fact of moving to software-based run and therefore moving from dedicated ASIC to generalized processors for the use.

(02:59:18):
That is basically what's the focus of the prior panel. And I think in a certain sense, it's an easier question. It's an easier question because I think there is a general momentum. We can argue it's slower or it's faster, but I think there is a movement in that direction.

Gabriel Brown, Omdia (02:59:32):
Okay. General momentum is a great place to close the panel. Thank you

Thomas Lips, Deutsche Telekom (02:59:37):
Very

Gabriel Brown, Omdia (02:59:37):
Much. Thanks everybody for attending. We're going to close up right as we walk off the stage here, but please give a round of applause to our three panelists. Thank you very much. And that is really the end of the summit. Thank you very much.