Guy Daniels, TelecomTV (00:15):
Hello and welcome to Extra Short on Telecom TV and our final program of this year's AI Native Telco Summit. I'm Guy Daniels and it's time now to grab a brew and savor the key talking points from the online event. All of the panels, interviews, and live shows from last week's summit are now available to watch on demand and to guide you through some of the highlights. I'm joined this month by Chris Lewis, founder of Lewis Insight and our regular co-host at the Great Telco debate. Now, we opened the summit with a panel that looked at the benefits of becoming an AI native telco. There were lots of insights and ideas that came out of our first discussion. You really have to watch it all. But I wanted to select one particular thread that was explored by our guests that looked towards the bigger picture for telcos.

Rahul Atri, Rakuten Symphony (01:12):
I personally feel that every telco should have decided their big picture by now the moonshot. How do they want to do the business? How do they want to set up the culture internally? And the technology will evolve. I mean, every time open AI releases a new model, a lot of startup die, a lot of things evolve as well and I'm very sure it'll continue going forward. But we look at the brightest side as well. New devices are coming, new way of interacting with the systems are coming. I see this as an opportunity rather than something which needs to be thought through in the future. We really need to look at the moonshot. What next big thing telcos can do. This is again, a moment where the telcos can become more than utility and the center platform for everything.

Manish Singh, Dell Technologies (01:57):
Telcos cannot, must not, should not be OTD on ai, on generative ai. And the good news is there are already a number of telcos across different markets who are actually already working on ai, generative AI use cases. They are experimenting, they're actually trialing and in certain areas they've also started to run scale deployments.

Guy Daniels, TelecomTV (02:29):
So AI presents a great opportunity to make telecoms great again. Absolutely it does. It's a route towards a TechCo or DSP future if they can grasp it. They're still smarting from missing the cloud opportunity. And as Manishh referenced there, being outplayed by the OTTs, Chris, is AI the key that telcos have long been searching for?

Chris Lewis, Lewis insight (02:54):
Well, guy, I think in this latest wave of interest, another topic. So you say we had cloud, we had the metaverse, and now ai, I think AI is a set of keys. It's a bunch of keys that we need to be careful how we use to whether we unlock the opportunity or whether we lock it up and restrict access to it. I think the big picture is of course something which I look at a lot as an analyst. I don't think AI changes the big picture for the telco. We know that we need to reduce cost of operations. We know that we need to make it easier to work with the industry and we know that we need to be much more careful about how we interact with our customers. Now AI has a role in each of those in a different way. So I think we have to be careful not to say that it's one size fits all or one key fits all locks, but actually that we need to be a little bit careful about how we use it and we need to be very conscious that this is not a one hit straight away hit.

(03:54):
It's a long-term shift as I think one of the speakers there said in the culture as well as in the business process and execution.

Guy Daniels, TelecomTV (04:02):
Yeah, absolutely. Chris, absolutely see what you mean there. And AI will play a big part of our great telco debate in London this December. And of course we have a whole day dedicated to AI on Wednesday the 4th of December and more on that a bit later now. The summit also featured a discussion that explored where telco should focus their gen AI activities as well as looking at telco use cases for gen ai. The panel spoke about the issues around the inevitable marketing hype and it also addressed a growing issue around what exactly constitutes gen AI and how this is different to machine learning. Here's what a couple of our experts had to say.

Beth Cohen, Verizon (04:47):
I want to make it clear that I think we're a little bit mixing up gen AI and machine learning and they are two different very distinct tools in the AI tool set. And gen AI works quite well with sort of semi-structured models and machine learning works better with structured models and data sets.

Paul Miller, Wind River (05:13):
Machine learning has been around for a little while and in use in telco networks we look at things like predictive outage avoidance or event correlation or root cause analysis, things that can be run at the edge of networks to determine faults. Anomaly detection for example, is a great example. These are not ai. These are machine learning algorithms, processing data trained on certain actions that are very helpful to the service provider. The AI use that we're seeing at least initially is more in the core of the network where you're perhaps training a large language model on an API set and using that generative capabilities to generate API calls and dynamically manage infrastructure or to do more simplistic things like template generation and configuration management. That's happening more at the central part of the network where learning can happen for an AI system. Learning isn't really going to happen at the far edge that's going to be more AI inference and the execution of AI algorithms.

Guy Daniels, TelecomTV (06:07):
Well, there was certainly broad agreement on this and it's an issue that has perhaps not generated and excuse the pun much interest so far, but it does have the danger of confusing the hell out of everyone. Something that we continue to see with cloud native versus cloud and which we don't want to see repeated. Thank you very much. And Chris, I have the feeling that telcos are just as confused and unsure as to how to use gen AI as the rest of us.

Chris Lewis, Lewis insight (06:34):
Well, I think Beth nailed it in saying that the machine learning in that more controlled environment is what we are using now. That automation of the network, the gen AI bit is taking the human and compute combination and allowing us to to interact much more cleverly with an enormous data set. But of course the issue is we need to make sure the data is clean. So in that controlled environment within the network, as Paul said, it is clean. We understand what's going on there. As we get towards the edge where we start to interact with the customers and the variables are way out of our control as an industry, that's when we're going to have more issues of actually using the generative ai. But to interrogate to interact. And what we've seen so far is many examples where the industry is beginning to experiment with this customer experience side. So I think this combination, it is, it's a hierarchy from machine learning with a set to this unstructured set out of the edge where we will interact as humans with the technology in a much more open way than we've done before.

Guy Daniels, TelecomTV (07:41):
Sitting alongside Gen AI on the hype stack, if you like, are LLMs or large language models, how these are used by telcos and whether or not they need to create their own telco specific model was the topic of our third panel this year. Yet the most cursory glance at the investment pages would tell you this is not cheap.

Michael Clegg, Supermicro (08:03):
What does it take to build a large AI factory to do training? And you're talking about a hundred thousand GPUs. For many telcos, that's going to be a little bit challenging from a cost point of view. So there's training and inferencing, right? So the training side is the expensive side and that is areas where there can be some good collaboration across telcos. Think we've already seen that the telecom AI alliance number of operators getting together, creating this telco language model, this telco dialect we spoke about earlier. We could take all the standards that Etsy produces the work that TM four comes and feed it into there. There's no reason why that needs to be telco specific. That's an industry domain area. So doing what telcos have done in the past, collaboratively working together, they can sort of get the common denominators done, share that information, that learning cost may be amongst themselves a little bit and then they localize it into their own particular telcos operations. So I think it's going to be essential just given the cost and scale to build and particularly to train a LLM model that is going to be a lot of benefits in some corporation.

Guy Daniels, TelecomTV (09:14):
So the cost of going it alone to develop an LLM is obviously a major issue here, hence the need for industry-wide collaboration. And we're already seeing this with the likes of the global Telco AI Alliance, but the cost of not developing a telco specific LLM was also debated at length during the panel, yet not all telcos on the same page here, some such as orange don't see the value yet of creating their own telco specific LLM. Chris, there's a difference of opinion here, isn't it?

Chris Lewis, Lewis insight (09:51):
There is a difference of opinion guy, and that comes partly from the fact that the role of AI control or dictation within the telco is so different that some organizations have gone to sort their data out first and thinking about the models afterwards. Some have multiple relationships with LMS and I think for once actually, as opposed to perhaps the cloud era when a lot of them went out to try and do their own thing, they thought they should be in control of it. I think at this stage we're looking at leveraging what's been done out there in the broader world. And let's face it, we can't afford to go and build those enormous models that have been done by open AI and the others. So I think it's a much more pragmatic approach to it. We don't need to do all of that straight away.

(10:33):
It's not a massive shift and lift if you like, the change of the way we do things. But I think the reality is models are out there to be used, there are tools to be used as an industry. We do need to work much more closely together to collaborate if we're going to, as man said, fend off the OTTs and have a stronger role in the future, a broader digital environment. So we don't have to do it. Telcos do not have the scale that the hyperscalers and the LLM model developers have got. So they should use what they can have multiple relationships. Ultimately they have to build some form of model, some form of tool which builds upon their own data, which works with their own networks and which supports their own customer activity. And so you can't do that with a generic model. You do at some point have to refine it down and deliver things in the language, in the brand, in the tone of your own organization. So we're a long way from that. I think having multiple strands of activity with the different language models is absolutely what we need now refine it down to the small language models and then who knows even hyper small, hyper personalized models in the future.

Guy Daniels, TelecomTV (11:45):
Yeah, you never know absolutely though, Chris, and this is a discussion that all verticals apparently having at the moment as well. Well this brings us really nicely to our poll because we wanted to know what our viewers thought about the AI native telco. So we asked the following question, how can telco's best leverage AI innovation to improve operational efficiency and develop profitable new services? We offered seven possible answers. You can see 'em here on the screen and we asked you to select whichever ones you thought were the most appropriate and the results kind of divide into three groups. As you can see here, you've got partnerships, edge services and LLMs ranking very highly at the top. Then you've got gen AI software teams and test beds, pretty much neck and neck there in the polling. And then future standards bringing up the rear. And maybe that's in last place because it's not a pressing concern for operators right now or we're just bored with six G now.

(12:50):
I'm not surprised that partnerships ranked top. They also came top last year when we ran the poll. 72% of you last year thought it was important, not quite as important as their share perhaps I am surprised that gen AI capabilities fell below the 50% mark, but also I'm encouraged that telcos are still trying to make the edge work for them. And there was strong showing for telco specific LLMs, which Chris and I have just spoke about there, which validates our decision to run a dedicated panel on that topic. I think overall the poll results highlight a degree of uncertainty over ai. TURs are perhaps hedging their bets somewhat at the moment. They know they need to do this, but the how, when and where is all still a little bit vague. Anyway, thanks to all of you who took part in the poll. We really appreciate the time you put into this and the poll for November's next gen Telco infra summit is now live online. It's ready and waiting for your votes. Now we also had a lot of great viewer questions during our two live q and a shows of the summit and they covered a wide range of issues, but I wanted to highlight a couple of them. First we had a question on day one about how telcos can earn revenue from ai. We heard lots of ideas for near term and long-term services, but this point about trust is I think particularly pertinent

Scott Cadzow, ETSI SAI (14:20):
If you have trust, churn should reduce so you retain revenue also, if you have enough people trust in your network, trust grows. Trust is, I dunno, it spreads. It is like a smile as if smile, people smile around you. If you trust an operator, other people will begin to trust the operator. You build revenue, buy by trust. And if you can use the data, use AI to tailor that data, tailor that knowledge of your customer, then I think you'll get increased revenue just because you become a native partner of your customer. And if we do that, yeah, revenue will grow.

Guy Daniels, TelecomTV (15:01):
Chris, I know you like to remind telcos of this very basic question. What is it you are selling? It's a great question. I'm not sure AI makes this answer any clearer.

Chris Lewis, Lewis insight (15:12):
No, I don't believe it does, guy. I think it will make it easier for telcos to analyze the data from consumption both business and consumer and to better match the service to the customer. But fundamentally we're selling broadband. Broadband, whether that's fixed mobile, a satellite perhaps in the future as well. So no, I don't think AI does change it. What it hopefully will change is the way in which we interact with it as customers and given the options of all the various channels of interaction, I still think we need a human element in there, that interaction. But although of course the better the chat bots get, the more comfortable we will feel with it. I think we do have an element of trust with the telcos. I think we always have. We've always had a bit of a love hate relationship with them because they are national treasures and yet we also experienced the relatively low levels of service. So I think the gradual improvement of customer experience will enhance that trust relationship and it will reduce churn. And I think we as customers, we're not always looking to change providers because we're focusing on the higher level activities, the services that we use to support our personal lives and our business lives. So yeah, trust is important. Delivering the technology that underpins all of our digital lifestyles is absolutely critical and we should focus on making sure we deliver exactly what we say is on the tin.

Guy Daniels, TelecomTV (16:33):
And we've got a related question from one of our viewers on day two who asked how AI can help move telcos away from utility models towards DSP models and the conversation turned towards platforms and infrastructure based as a service opportunities.

Martin Halstead, HPE (16:53):
There's a lot of opportunities that we are seeing on the platform side of ai, IE the infrastructure GPUs, the ML ops environment to go and build the applications is kind of what the telcos are focusing on. If you have that AI platform, you also then have the capability to offer that as a service offering as well. So it's almost like the telco becomes a tenant of the platform for their own internal use, but also offer it as, for example, GPUs a service to their end users. So there's a lot of uses for it.

Guy Daniels, TelecomTV (17:32):
Well this is something we've highlighted in our previous telco as a platform summit and Chris, it's going to be interesting to see how the separation of applications from the platform is going to factor into a successful AI native strategy for telcos.

Chris Lewis, Lewis insight (17:49):
It is guy, I think the confusion, the temptation as we said earlier on is to go and try and do everything relating to ai. There will be some telcos and as we've always said, every country market is different, every telco is different. There are some who are already starting to offer GPU as a service, but I suspect for the vast majority this move towards new revenue is going to be leveraging what we've got in place, opening up the network, opening up the APIs, and actually allowing the communication services to be embedded in other activities from other players. So going back to your partner issue earlier on in the poll, actually I think it should be seen as more important there. So I think the danger is going after everything to build everything, to expect everyone to then come to the telco for the service. Some will manage it, some will achieve that most will not even try. But actually I think it's much more important to try and find that balance, the relative positioning of what the telco offers, that product, the broadband and so on. And actually embed that into other value chains rather than expecting other value chains to embed themselves in the telco.

Guy Daniels, TelecomTV (18:58):
Well we also ran an interview with Alex Choi, formerly the chair of the O-RAN Alliance and now the chair of the AI ran Alliance do keep up at the back. Now it's early days for the Alliance, but he set out their objectives and he explained the rationale behind their three separate working groups.

Alex Choi, Softbank, AI RAN Alliance (19:18):
First AI for ran. And this group is focused on exploring how AI ML technologies can be applied to improve RAN performance. For example, spectrum management, traffic management and optimization and energy optimization, et cetera. And the second working group AI and ran this working group is dedicated to developing a unified computing ran computing platform that can run both AI and RAN workloads concurrently. So their short-term focus is on GPU based platform to handle the high computation needs of AI applications while simultaneously managing ran function workloads. The last one, AI on RAN working group also we can call it a RAN ai, I mean the RAN for ai. And this group is exploring how AI applications such as Edge AI and the real time inferencing can be run effectively on RAN infrastructure.

Guy Daniels, TelecomTV (20:29):
Well obviously there's a lot of interest in applying AI to the radio access network, both for its internal operation and for supporting external AI workloads. But Chris, do you think a dedicated organization can help you and can be strong enough to influence the vendors and the standards and the specification bodies?

Chris Lewis, Lewis insight (20:52):
Well guy as ever, the vested interest behind many of these bodies has to be taken into account. And of course the promotion of this disaggregated notion of networks and the importance of the RAN is not in dispute. It's a key part of the telco environment. But I think if we go down too narrowly into each of those elements, we're going to lose the big picture that we talked about earlier. So if we don't have that broader framework within which we want to improve the relative role of the telco and the connectivity services and associated services, I think going too far down, having too many bodies trying to define what we're doing is actually going to do the opposite of improve the network. It's going to fragment activities. And once again, this is not just a telecom industry issue. The people getting involved in this are coming at it from the broader infrastructure, the broader platform, the broader IT and cloud side.

(21:47):
And we do need to be very conscious of telecoms working more closely with those players, not trying to out flank them, but actually to work more closely with them because they want stronger telcos to deliver the underlying services. But perhaps that's the ultimate issue is that where does the telco fit into that hierarchy in the future? Because they're national players, they don't have the same scale as the hyperscalers and they need to find that point of equilibrium where they're happy working with all the partners. They will find new revenue streams by doing that, by not trying to control. So I'm against too many different bodies to try hide what's going on down at that micro level.

Guy Daniels, TelecomTV (22:28):
Yeah, thank you very much Chris. That's one of my key takeaways from this summit is, as you've just said there, keep an eye on the broader framework. It's so important. Well that concludes our AI native Telco summit for another year and Chris will join us again in December for the great Telco debate and the Telcos and AI event both being held in person in London. And Chris got any surprises for us in December?

Chris Lewis, Lewis insight (22:55):
Well, guy, I think we always managed to find surprises during the great Telco debate because the format generates people's interest, gets people taking a decision, trying to vote either side of the motions that we produce. Now we are going to of course touch upon six G you already mentioned we'll touch upon ai. We're going to delve into what a platform means for the telco and perhaps even APIs might get a mention there. Whatever we touch upon, we know that the witnesses that we have on stage, the people in the audience, the questions coming either from the room or from outside online. We know it's going to stimulate debate and that's the whole purpose of it. But sort of wrapping things up towards the end of the year, give people a chance to perhaps reshape their thinking. And as we've always said the tagline, we hope you'll never think the same about telecoms again.

Guy Daniels, TelecomTV (23:44):
Yeah, absolutely. Thanks very much Chris. Well please join us in London if you can and if not, you can watch us online. Don't forget, we are back next month for the next Gen Telco infra summit. So do start sending your questions into us and please have a look at the poll. For now though. Thanks for watching and goodbye.