Guy Daniels, TelecomTV (00:15):
Hello and welcome to Extra Shot on Telecom TV and our final program of this year's Next Gen Telco infra summit. I'm Guy Daniels and it's time now to get yourself a coffee, sit back and savor the key talking points from our online event, all of the panels, interviews, and live shows from last week's summit and now available to watch on demand and to guide you through some of the highlights of our annual next Gen Telco infra summit. I'm joined by my colleague Ray Le Maistre, who will also reveal the final results of our viewer poll. Well, we opened the summit with a panel that looked at the issues around building a future-proofed, telco infra, and amongst the topics covered, we talked about abstraction, and then we moved on to orchestration, simplifying the network, reducing the number of orchestrators. It's a very popular idea, but is it practical? Well, here's what our guests thought.

Beth Cohen, Verizon (01:22):
Orchestration is like standards. There's so many to choose from. So I think yes, that's the north star to have a centralized orchestrator. However, I think in all practicality, it's just not even possible to get there.

Fernando Castro Cristin, HPE (01:40):
You have to be able to go down to what is manageable, and that will be your centralized service orchestration for that area. And so that will create multiple ones. The important I believe is to have a strategy around it and the strategy of those multiple ways of managing have to be consistent. So it's manageable, but one single utopia.

Brendan O'Reilly, Boldyn Networks (02:06):
It is a utopia. It is hard to do. The great thing about our industry is we keep asking what if, so what there was one, what if we could go all general purpose computing? What would happen if we drive network abstraction to the nth degree? And it's only by asking those questions and by continuing to push, do we get to a point where we absolutely can see what can be done? I think the great thing about this industry is it just doesn't give up. And I think as we continue to ask these questions, we'll continue to make progress.

Guy Daniels, TelecomTV (02:40):
Well, telco has often run separate networks, separate business units with all the associated regulatory headaches, let alone the technical ones. So if we can't yet reduce the number to just one, then maybe just a couple Ray, anything to make the infrastructure cheaper and more flexible and easier to manage, surely must be a priority.

Ray Le Maistre, TelecomTV (03:01):
Yeah, absolutely guy. And I think if you look back five years, maybe 10 years and compare the way that networks and services were set up and managed, then you'd see a big difference to now that the shift towards cloud native processes and functions is starting to have an impact. It's happening slowly, of course, but if you even just look at the data transport networks and see how they're being consolidated into a single infrastructure with a much smaller number of management tools, then you can see definite advances in terms of simplicity, cost effectiveness, and energy efficiency. We won't get to utopia, but the industry is always taking steps towards it, that's for sure.

Guy Daniels, TelecomTV (03:48):
Yes, indeed. Thanks very much, Ray. Well, the summit also featured a panel on supporting AI and the emergence of AI factories, and we spent quite a lot of time discussing the emergence of these, so-called AI factories, which are essentially a new type of data center built around GPUs and what the opportunities or challenges may be for telcos.

Susan James, DSP Leaders Councillor (04:13):
They are very expensive and very few companies can afford to actually buy these GPUs to have them access to them all the time. So they're going to be a shared infrastructure as well. So I think you will see all the cloud providers being able to provide these capabilities going forward because that's a logical extension of their business, but it's going to be different from the runtime environment you would typically deploy your typical cloud workload on. So that's how I see these AI factories being built. You are going to use them while you're developing your applications. Once you've trained your models, you're going to deploy it and get off them as quickly as possible.

Guy Daniels, TelecomTV (04:50):
So very much designed for AI model training, and we moved on during the panel to look at the edge opportunity, especially around AI inference and the drivers there. But Ray, for telcos, it sounds maybe like the gates to the AI factories are firmly closed, at least in terms of telcos thinking about building these vast LLM foundational processing factories.

Ray Le Maistre, TelecomTV (05:16):
Yeah, I mean those gates are very hard to open for sure, and even once opened, is it a Pandora's box for the telcos? There are a few small instances of the term AI factory being used by some operators, but I think that's more for marketing purposes for GPU as a service type operations and not in a misleading way. There are some interesting AI infrastructure developments underway in South Korea, for example, but nothing in the realm of the hyperscalers, even if the hardware was available, who would have the funds and capabilities to actually acquire and deploy it. And of course, there's an interesting debate going on right now about whether the telecom sector needs any of its own foundational models, and that's going to be a very interesting discussion in 2025.

Guy Daniels, TelecomTV (06:08):
Yeah, thanks. Very absolutely. And if anyone wants to learn more about this, we held a dedicated panel on LLMs as part of last month's AI native Telco summit. It's well worth watching now. Each day of this summit also featured a live q and a show where our guests answered your questions. These are also now available to view on demand, and as usual, our viewers submitted some great questions. One of them asked how future AI demands and opportunities will reshape telco infrastructure, and here's what our panelists thought.

Fernando Castro Cristin, HPE (06:44):
Is the run revolution in terms of having an open run that going into the details, will that enable workloads to be influenced at the edge and generate some revenue for the telco or for the people using the telecom industry or maybe a shared situation in terms of revenue? Why not?

Sree Nandan Atur, Rakuten Symphony (07:11):
So telcos really are innovating here as to depending on the AI workloads that are running on your different layers of your edge network for each edge and core, depending on your AI workloads, the kind of inference that needs and the kind of data that is operating on. There are various AI models, and that is how I think telcos are reinventing their networks to bring in AI from the perspective of hardware and software,

Beth Cohen, Verizon (07:39):
AI and telecom infrastructure aren't necessarily the same thing. So a telecom could in fact build its own infrastructure to support its network and then tie it with a second piece of infrastructure to support the AI workloads.

Guy Daniels, TelecomTV (07:57):
Well, there's a lot more interest now around inferencing the application of the AI models and how telcos could use their infrastructure, especially edge sites and the RAN and Ray. It's going to be interesting to see how telcos can capitalize here if indeed there is a market and the extent to which these different workloads might be interconnected.

Ray Le Maistre, TelecomTV (08:21):
Yeah, I mean, I think this is now one of the most crucial infrastructure planning decisions facing telcos today as they make the progress from traditional siloed networks to general server-based platforms that can in theory run multiple workloads, should they plan for an all-in-one automated edge architecture that can apply the resources to the most pressing services or application requirement, whether that be the mobile services or AI workloads, big data processing or content delivery, or as Beth have different systems running in parallel. Again, we're looking at a utopian situation with a single platform, but for the majority of operators, I'd say that's a very long way off, but something to be baked into the long-term planning that SoftBank and Nvidia are for sure pushing the boundaries with their AI ran pilots, but that's going to be near impossible to replicate at scale by the telco sector in the short or medium term, at least in my view.

Guy Daniels, TelecomTV (09:27):
Yeah, thank you, Ray. Yes, there's going to need to be some serious discussions over future planning here, but telcos can't afford to wait too long. Well, we got a related question from one of our viewers on day two of the summit who asked if the telco infrastructure sector was progressively being taken over by other players such as hyperscalers, and if so, what can telcos do to counter this increasing global competition?

Susan James, DSP Leaders Councillor (09:57):
What they're providing to enterprises from a communication perspective has certainly decreased as the world has changed. So obviously other players like the hyperscalers are coming in there. They have sales teams in every country around the world, and when you look at the percentage of revenues coming from those enterprises for communications and IT services, of course, they're going up and up and up.

Eugina Jordan, TIP (10:22):
And if we look at the public or private cloud, if mobile operators should go with their own cloud or go with hyperscaler cloud, that is the question, is it going to be cheaper and more efficient for me to run those services in my own cloud or partner with the hyperscaler so I can provide better service and utilize my spectrum assets better?

Grant Lenahan, Appledore Research (10:51):
I think in understanding where is the competition, it's really separating running the core network and or services that run on the network where there can be real competition versus the other 85 or 95% of network costs, which are as we go out into the skinnier and skinnier RAN and broadband loop, including a lot of, by the way, the X fall for 5G and future generations of ran where it's totally impractical to put it in a data center because the data center 75 miles away. It's really that simple.

Guy Daniels, TelecomTV (11:26):
You just can't avoid the hyperscalers and this perennial friend or foe question. Can you, and again, this question reflects another concern from our telco audience about the services that will be required in the future and the infra that is going to be required to support them. Ray, as we have already said, planning for the future trying to play to your strengths, it's not easy.

Ray Le Maistre, TelecomTV (11:50):
Yeah, I mean it's very tough. Absolutely. And this is where the strategic thinking comes in. Of course, you can't ignore the hyperscalers. Telcos have never needed them as partners more than they do right now in this AI age. But how can service providers differentiate themselves and how can they become more agile and flexible? Those I think, are going to be the key attributes for the future. And let's not forget the hyperscalers for sure have tremendous advantages, but so do the telecom network operators, the access networks where the fixed or wireless are absolutely key here. Now, does that advantage erode as the access networks also become more IT platform oriented? I think by the time that happens, we'll know who the survivors are in the digital communication services sector.

Guy Daniels, TelecomTV (12:46):
Yes. And that day is fast approaching, which brings us very nicely to our poll. We have a viewer poll for each of our summits and they never disappoint us. Ray, what did we learn from our poll this month?

Ray Le Maistre, TelecomTV (13:01):
We wanted to know what you thought were the most important investment areas to create a future proof telco infrastructure summit. Attendees could select as many options as they wanted, and that's why the total adds up to way more than 100%. And the big picture answer is, well, they should pretty much invest in everything. While service-based architecture attracted the highest proportion of votes with 52%, building a containerized core and investing in a centralized service orchestration platform, if they can find one garnered nearly as much support with 47% each and not far behind with 43%, we have AI factories and AI edge compute while we had the same score for improving capacity and availability. I mean this essentially is a CapEx planners nightmare.

Guy Daniels, TelecomTV (13:56):
Oh, yes, it is. And as you say, the answer is pretty much everything. I wonder how these choices will change by this time next year. Well, we also featured two executive interviews during the summit, both of which Ray conducted at the recent Fyuz event in Dublin held by the telecom infra project. Ray spoke to Cayetano Carbajo of Telefónica and asked about the delivery phase of AI and the associated inferencing, which will likely place an additional overhead on the network infra.

Cayetano Carbajo, Telefónica (14:30):
Everybody's talking about AI and a lot of people is talking about how AI can help in increasing the efficiency of bio automation, but that's another angle that is what we have to do in network for facing the challenges that AI poses delivering. AI services will not be easy. The level of traffic will increase maybe more uplink than downlink. There will be the need of low latency because many of the AI inference application will need low latency. So we need to distribute edge, we need to distribute our core nodes. We will need a very efficient transport network. And I think by the end of the decade, all this transformation has to be done

Guy Daniels, TelecomTV (15:17):
Well, yet again, Ray, we end up talking about inference and the huge demands it may place on the network. It seems to be emerging as one of the key considerations for NextGen Telco infra.

Ray Le Maistre, TelecomTV (15:29):
Yeah, I mean, this is a major focus for all of the tier one telcos I've spoken with recently, planning network capacity to deal with the impact of AI workloads and in particular what the demands on network infrastructure will be once we hit the era of multimodal ai. That's what's really freaking out the CTO offices. How do you avoid bottlenecks without breaking the bank? Interestingly, Telefonica believes that investing in disaggregated platforms is the only way to tackle this, and importantly, to deal with these capacity challenges in an energy efficient way. And that's something that Kayano Telefonica has been looking at for some time already.

Guy Daniels, TelecomTV (16:16):
Yes, it's all really interesting. Thanks very much, Ray. Which brings us to the final program of this year's summit. You interview with the new chairman of tip, Rob Soni and Rob talked about the role of the telecom infra project and developments in the industry, including the challenges faced by both telcos and infrastructure vendors.

Rob Soni, TIP & AT&T (16:38):
I think there is a strong pressure from the operators to try to leverage what we've deployed for 5G or what we will deploy for 5G for what comes next, six G or next G or whatever. So from that perspective, there is a strong push from the operators, is there yet coherence to say what would we consume and how would we be able to allow for things that are disruptive while also preserving as much as what we've deployed, especially as it starts to move on different cycles of hardware upgrades and software upgrades. Because now the underlying servers themselves will specifically be now not generation specific. The cloud infrastructure layers will no longer be generation specific. And to a degree now we expect the network management layers to no longer be generation specific. So this is an exciting time because we've never seen that before. So from that perspective, I think we're aligned on that when we go with a cloud-based solution. But today, 99% of the world is still appliance space. It's still purpose-built hardware and the purpose-built hardware. The question is, is can it absorb the change? And so that's the challenge for the infra vendors.

Guy Daniels, TelecomTV (17:46):
And Ray, this issue about trying to leverage what telcos already have deployed for 5G or are still planning to deploy and the way that cloud-based architectures will evolve differently to, it's absolutely fascinating, but again, it makes planning next gen infra really difficult.

Ray Le Maistre, TelecomTV (18:05):
Yeah, I mean this is where I have a lot of sympathy for the tier two and tier three network operators. Many of the large tier one telcos have the scale and resources to be able to adapt, try things out either in one territory or even in one region. And we're seeing that to a certain extent with open ran for example. But the majority of operators don't have that luxury. They're stuck with those traditional 5G investments for some time yet. And 5G standalone has proven trickier to deploy than had been anticipated. Once an operator's migrated to that cloud oriented platform, then the options open up, so to speak. The flexibility is there, but getting to that point for the majority of operators is a really tough and capital constrained challenge, and that's one of the reasons why many of them gag at the mention of six G. Certainly when it's positioned as another generational leap, even beyond 5G gets veins throbbing.

Guy Daniels, TelecomTV (19:07):
Thanks Ray. And what a way to conclude our annual next gen Telco infra summit and breathe. We are back next month with our two in-person events, telcos and AI and the great Telco debate both held in London, both available to watch here on telecom tv,

Ray Le Maistre, TelecomTV (19:28):
And this is when we end the year with a bang. Telcos and AI was new last year and a great success with its slightly informal and very interactive format helped of course by great speakers and the great telco debate is something else altogether. There's a reason why people come back year after year and why they get so involved. It's a fantastic format for an actual debate. No holds barred

Guy Daniels, TelecomTV (19:55):
And guaranteed to get the veins throbbing because yes, there will be mention of six G. Now don't forget to catch up on all of our programs, links to the on-Demand versions are all listed below this video. Until next month from Ray and me, thank you for watching and goodbye.