Transcript
Pablo Tomasi: Welcome back. to the Telco drift the podcast about. tech telecom and sci -fi. So... ah First of all, before I start the podcast, this center there' is not going to be a video. um currently traveling and there are a little bit of restrictions in terms of bandwidth and similar so bear with me this is going to be an audio only episode so would actually be interesting to see whether there is any difference. and whether people are actually watching the videos or not but anyways let's focus on the content now
Pablo Tomasi: So. If you... Look at the future. Can you imagine a future? we're there is an AI system that needs to truly make a decision in milliseconds. Now let's ignore the sort of funny videos AI generated.
Pablo Tomasi: But think about a robot moving through a factory. or a camera detecting a potential safety incident now i know these are like a lot of use cases that we've always been discussing. but potentially. We always go back to the same idea, right?
Pablo Tomasi: that data and needs to be processed quickly of course it needs to be processed reliably and sometimes it may need to be processed locally. So sending everything back to a distant data center may not be...
Pablo Tomasi: is the best solution and And that in a nutshell. is when computing needs to move closer so it can run on enterprise it can be hosted in a telco building or it can be on the device itself now the interesting thing about edge computing is that For me, if you think about the matrix,
Pablo Tomasi: you know everyone inside the matrix is potentially an agent And... you can apply the same thought about edge computing. everything is potentially an edge computing node.
Pablo Tomasi: everything. from your phone to the Again. tower do whatever you want could potentially become a source of patch computing.
Pablo Tomasi: Now the the question of course that a lot of people were asking is um Who owns? this edge computing thing. So...
Pablo Tomasi: We know that. Talcos. have built physical infrastructure to connect people, and we also know that sometimes they have separated some of that infrastructure.
Pablo Tomasi: And that was... you know due as we saw in the Latest podcast to reduce debt. uh you know attract investors so on and so forth But maybe.
Pablo Tomasi: again because of AI this whole conversation like Where is the edge? Who owns the edge? is becoming. relevant again. And maybe we can see how telcos are trying to monetize it. Whether they are really doing something or they should be doing something.
Pablo Tomasi: But of course. the first thing that we need to think about Before we look at some case studies. out. we should try to clarify. what we mean by hedge computing.
Pablo Tomasi: And here... This is where there is. the biggest problem Because you can ask them people. and depending what they do you're gonna have 10 different versions of what edge computing is.
Pablo Tomasi: so the term is usually referring to to something very broad It can be in anything from a server in a factory to a computer. of the computing integrated in ah in a mobile network.
Pablo Tomasi: but very in a basic form. It's all about processing data closer where it's generated or where it's used. rather than sending it far away. Now, water that far away. means what we really don't know it like is your guess is good as mine.
Pablo Tomasi: And why why is that? important to find things. Um, From my point of view... If you cannot define a market, you cannot have a market.
Pablo Tomasi: so if something is too broad then there is no marker for it now There is a certain importance. in names And for those of you that have seen, we're still...
Pablo Tomasi: yes we're still in time for some you know horror of folklore Those of you who have seen movies about exorcism one of the key steps. in fighting the demon.
Pablo Tomasi: uh that it's possessing whatever poor soul is you know subject to that one of the first steps is really to know the name of the Nemo. of the demon We don't know the name of the demon you cannot Get rid of it. so clearly there is like a lot of importance in naming things and in naming demons apparently Maybe I should do a horror focused episode.
Pablo Tomasi: but Probably. let's keep it as as a plan b for now would be interesting though anyways So... ah going back to a different version of edge so you can run AI software inside a factory or you can Sort of.
Pablo Tomasi: which comes in a regional facility form. close to where city but you can be a cloud provider as well and you have your own distributed architecture. um You can integrate AI and run. as we've seen in some other episodes and all of these are Similar.
Pablo Tomasi: Ideas about Edge they're related if you want but they're not interchangeable and realistically you're not While you could have edge computing everywhere, it doesn't make sense to have it everywhere.
Pablo Tomasi: and and there are a very specific locations where it makes most sense the problem is that there is not a clear cut way of Really? fine tuning or like dissecting word that is actually happening So let's then bring in...
Pablo Tomasi: our dear telcos so why should they care about this I think that the the answer is really ironic. it's Isn't it ironic?
Pablo Tomasi: So... For most people, whenever I hear people talking about this, they they start to see this edge competition conversation as something that is strategically critical.
Pablo Tomasi: for telcos as like the obvious solutions for for what they need to do but the reality is that they Stumble into it. So the the way... why people started thinking and edge is big.
Pablo Tomasi: is simply because That'll go. happen. to possess assets that could be valuable in a distributed computing environment. Um...
Pablo Tomasi: So that is... luck. Hmm. like there is no bigger starting point for me rather than oh i just happen to have this very valuable land I may sell it and make a bunch of money.
Pablo Tomasi: Now how you sell it? that is what we'll define How much money will you make here? and When do you sell it? will also define how much return are you going to be able from their money that you just have you did not plan for it By the way...
Pablo Tomasi: As I'm thinking... of companies that have valuable assets and... you know. locations. Maybe it would be good to have an episode on utilities.
Pablo Tomasi: turning into talcos let me know if you think that could could make sense still sort of part of this transformation metamorphosis idea so it could really you know fit the bill but let me know i drop my message if you think that that makes a lot of sense so um effectively Because they have fiber routes uh central offices network sites we know that like power connections um the problem.
Pablo Tomasi: is that having those assets is not the same as having ah a real profitable edge business. So let's now look at a ah few recent announcements. and let's try to better understand what the records are doing how things are drifting and and whether we are just repeating the same MEC story all over again from a few years back.
Pablo Tomasi: So the first case study comes from from Europe. and in feb 2026 five major european operators Deutsche Telecom. Orange. Telefonica.
Pablo Tomasi: TIM and Vodafone they they demonstrated a a project called the european edge continuum so the idea behind the project was to federate their edge environment so an application can be sort of built one and deployed across multiple operators So the the the idea is quite simple, right? um um
Pablo Tomasi: I'm in business I want my application to run across multiple countries normally I would have to have an application that gets uh tweaked to the different countries because there are different networks but now Same application we run in Germany, Italy and Spain.
Pablo Tomasi: so this federated platform could make it easier to to manage the... deployment across different networks. of course what this matter is like Scale is critical whenever you're thinking about application and computing.
Pablo Tomasi: uh and and the the reality is that a single operator and definitely has useful infrastructure. but it's limited in in its own being like a a market player and a a developer doesn't want to build a thousand things just wants to build one and then use it forever So cooperating and and again.
Pablo Tomasi: A lot I noticed of the conversation that i'm having in this podcast day Yeah, dude. The big opportunity for tech is always about cooperation. more than anything else Maybe something to look in.
Pablo Tomasi: look into So operators could offer broader coverage without us effectively, you know merging or or anything and of course because it's europe like anything is sovereign in Europe like we have zero sovereignty whatsoever but everything is sovereign in Europe So that could be the original alternative compared to other competitors.
Pablo Tomasi: This is an interesting announcement. but but be let's be careful demonstration. doesn't equal success for business. and there is one big choking point that we always come back anytime that we think about telcos application and so on which is any platform any play needs developers And for me this is where the idea breaks for me.
Pablo Tomasi: um because there are questions on the pricing model. on everything having to work reliably, you know. so level service level commitments and so on uh but attracting and keeping those developers which are already pretty much part of other ecosystems is always going to be terrible bottle So I'm not sure.
Pablo Tomasi: whether it's anything. that these european telcos What does can do? to really create them
Pablo Tomasi: creates such ah an environment and of course because we started talking about hyperscalers there is There is another question that that brings me to the second case study.
Pablo Tomasi: which is What should techos do with hyperscalers? And for the second case study, we look at Verizon. and how the company is trying to Turn it.
Pablo Tomasi: footprint or part of it into AI infrastructure. So, of course, as we're talking about the US now. and in July 2026 Verizon disclosed a dark fiber agreement. worth more than 1 billion dollars to connect Google's data center companies also is sharing The discussing plans to retrofit selected central offices as a smaller data center for AI inference.
Pablo Tomasi: So there are two different approaches here to modernization. The first one is fiber, and that is sort of the... the standard play that I've seen. Many telcos are pursuing now. So AI infrastructure depends on connected between data center.
Pablo Tomasi: the more computing capacity expand the more expensive the need for high capacity connection between facilities. and that's when the tech can monetize its fiber networks So again cables in any from wayward is not sexy but it does bring money so maybe i should explore that in another episode So you you are, you know.
Pablo Tomasi: Connecting. you know locations and that's that's the business the second part with It's a... the upgrading of some some central offices.
Pablo Tomasi: I think it really depends and I'm not then export on it some may be better suited than others for distributed computing. so the suitability is not guaranteed a but you would have to think that what browsing is currently really thinking and evaluating is the suitability of those locations um because they are hardware can require.
Pablo Tomasi: additional actual capacity advanced cooling of course there needs to be some physical security for the site and and so on But it's it's interesting to see. What? what is happening here you have like a standard monetization of your connectivity and i saw to part of that broader ai edge computing development and i'm potentially trying to do something more because that would be the second option could be more differentiate or differentiation
Pablo Tomasi: Uh... And... it will be sort of launching their own b2b potential managed solutions for edge computing and so on there is also one one additional thing that is worth mentioning While, you know, talking about 1 billion DLs, of course it's it's good money.
Pablo Tomasi: i'm not really sure how many years though is the deal for And so... it's it's good money but it's not gonna sort of Change.
Pablo Tomasi: You know. the the revenue profile of of someone as big as Verizon. So... interesting. But I would say it's probably...
Pablo Tomasi: in their conservative play. Now let's move on a third case study. and we are jumping back to Europe, specifically Spain.
Pablo Tomasi: um yes you jumping around regions may sound a little bit odd but they're subtle. there is a certain rationale into it so now we're moving from connectivity to more industrial intelligence, so Telefonica Had a couple of announcements and particularly announced the commercial edge services.
Pablo Tomasi: For. B2B and public administration. So. it started to roll out with the announcement in feb 2026 and then it completed the plan a lot of 17 nodes I believe I was in June.
Pablo Tomasi: If I remember correctly. and one of the applications involves an array manufacturer again if I remember correctly so the idea is that the company is combining different assets from 5g to fiber to edge computing And... they are serving more of a solution right rather than just sticking to to connectivity
Pablo Tomasi: and because of these edge nodes. the um the application effectively can run in in different more ideal location and it's closer to where the yeah where the where the data is generated. So the basics of edge computing, right?
Pablo Tomasi: and what is interesting from this is that we see the movement from connectivity tool. and more. integrated offering but also we see that there is no need to to have edge everywhere I would imagine 17 nodes are sufficient for the vast majority of use case and potential customer the telephone has identified so again it doesn't have to be everywhere it just has to be strategically well placed
Pablo Tomasi: placed so that is very good now we go back though to some of the challenges that we saw in the in the first case study. Uh... so we'll roll out and you know like me a pilot a showcase it's all well and good but uh that doesn't mean this is a successful business so it doesn't yet mean there customers will be willing to
Pablo Tomasi: embrace this model know pay revenues and so on and so forth so it's an attempt we will keep an eye and see how well that Evolve. over the next few years.
Pablo Tomasi: now of course i need to mention the AI-RAN a little bit i think we've already talked a lot about it so you know suffice to say nvidia is working with a lot of telcos started with TMo us and of course has a big partnership with nokia and somehow there's like physical ai application using this distributed edge infrastructure but we've seen recently uh quite a few tiles announcing different might sound different um
Pablo Tomasi: you know engagement within. this development so T-Mobile completed a world first ai run call I'm going to turn it off.
Pablo Tomasi: BT, Deutsche Telekom and ELISA, they all announced AI RAM pilots. on live net talks. Using Nokia's good accelerated any run software on NVIDIA area run computer a computer.
Pablo Tomasi: probably and then sort of 9 EMEA Telecoms operator are building and expanding NVIDIA powered infrastructure across centralized and distributed locations so a lot happening there but we covered a little bit some some other time so i think that's that's enough but again all the rich computing that is happening let's move down i think on the Lust.
Pablo Tomasi: Okay, so what did I have? And that is about SK Telecom. So, SK Telecom very recently, in in August it is split um it's wholly owned subsidiary SK Broadband.
Pablo Tomasi: in two parts. One remains SK Broadband and another part is established as SK Horizon. So as part of the plan, there's going to be also investment from a couple of investors.
Pablo Tomasi: into SK Horizon KKR and I think the other one is
Pablo Tomasi: Um... IMM. M. investment. Stonebridge Consortium, looking at my notes. so the SK Broadband remaining company will focus on innovation the fixed line in the enterprise business while the new company the Horizon
Pablo Tomasi: Uh, we'll be looking to expand AI data center infrastructure.
Pablo Tomasi: So the idea is like separating the company um so once so you have capital intensive infrastructure you're bringing some of that investment and your enable it a couple to have a more focused role in really driving that a very specific infrastructure play
Pablo Tomasi: and um SK Telecom remains sort of. in charge. of the whole of the whole circus as to say so they should be able to again drive direction author.
Pablo Tomasi: that edge computing AI data center. play whatever way they they want to go
Pablo Tomasi: So very interesting. um And again ah I believe. Korea is quite a unique market when it comes to what they are actually doing in Edge and AI. so we'll be very good to see you again over the few years over the next few years.
Pablo Tomasi: how well is that working out and whether it would be possible for other countries as well. to take a similar approach when it comes to driving the growth of AI application and
Pablo Tomasi: the whole lot.
Pablo Tomasi: So, Um... Hopefully. these a few examples were like showing a little bit of different approaches to monetize what is a continuum of services and opportunities and if we then need to Okay, let's try to be a little bit Eve.
Pablo Tomasi: Let's try to look at some possible futures. how how this market can can develop. On one side you you can try to be optimistic, right? So there's a bit of a telco renaissance.
Pablo Tomasi: Um... and And that would be sort of the scenario for telcos the user physical infrastructure network took expertise relationship and they build differentiated edge business so they don't stop the connectivity but they build on top of it so that's sort of where Telefonica is trying to go How realistic is that?
Pablo Tomasi: Yeah. I think we're going to be in the dystopian future but Never say never. Then there is a second option. the They rent?
Pablo Tomasi: locations. you know they they ran infrastructure. and i think this is likely in most cases gonna be the main option so it's not going to be a glamorous play but they have some assets they can rent them out yes they're not going to capture the big value of the vi but it's a a solid business Then there's gonna be of course some sort of hybrid model were.
Pablo Tomasi: it's going to be half the way between okay really trying to push forward with the in edge play and really just rent this stuff out So let's try to close this Close this down then.
Pablo Tomasi: Um... Yeah, it has brought new capabilities new possibilities and like ah renaissance of the Edge.
Pablo Tomasi: um Frenzy. but the the underlying problem is about the edge will be small zero. hasn't really has really changed I think telcos are is struggling with that.
Pablo Tomasi: Not only that, because I want to see a struggle with the definition of edge. and without having the definition cleared up there is very little too. that can be possibly done. from a business model perspective.
Pablo Tomasi: but there's still a lot of question open. on the execution of displays, the economics, but our customers really want that. and who's going to control the different parts of this of this play um However, the one thing that we can say uh which is again it's pretty obvious so not a great way of concluding the podcast but hey It is what it is.
Pablo Tomasi: the the to rear rims. of the connectivity and the computing they're getting closer and closer together So... Definitely.
Pablo Tomasi: something will be boiling up from from this connection more and more over the coming years so we will see who's gonna be ending up on top Now that was everything I wanted to share with you today So thank you very much for listening. um see you or actually yeah no see you don't even have a video but talk to you next time
