DBTT Episode 9: EU CADA & Telecoms Sovereignty, Telco AI, Quality of Experience & Interconnect

In this episode (recorded on June 8), Ben & Dean discuss Telecoms Sovereignty, a growing emphasis on Quality of Experience and new metrics and measurements, the new EU CADA proposals on cloud and AI especially for public-sector procurement, and various recent conferences and events in Europe, the US and Canada. We also cover Indoor Wireless and regulatory shifts towards assuring end-user and application performance, rather than just headline coverage.

01:10 - Sovereignty as a key theme at telecoms events in Europe, UK & Canada

02:30 - New proposed European cloud public-sector procurement rules could cause problems for telcos. (See Dean's LI post here)

04:40 - Benoit's thoughts on European sovereign cloud propositions. and possible re-globalisation of tech companies to map to new geopolitical realities.

13:30 - The Telecom Industry needs to work collectively, across competing subsectors, to define a proper framework for Telecom Sovereignty to avoid getting sidelined by the Cloud/AI/chips focus

14:15 - Benoit talks about policy-meets-data, a recent webinar he did with Ookla (link), and another study about user-experienced quality vs. Internet Exchange location.

21:10 - Dean and Ben discuss Ofcom's new position paper on mobile quality, and its use of OpenSignal crowdsourced data and methodologies. Also network speeds and latencies (Dean will be reviewing that paper soon on his Substack)

29:30 - What's next? Dean will be at conferences on NaaS, spectrum, satellite D2D and then DTW (events list here) and will be running his own workshop on Indoor Wireless policy & tech trends on July 16 (details + earlybird here). Benoit  will be chasing the dead EU "fair share" network-fees zombie as it tries to reanimate itself around the world  

30:10 – Book Ben & Dean for your event! Get us on stage at your conference, or facilitating at your private meeting or executive offsite. Connect & message us on Linked.

Transcript

0:01

Hello, and welcome to the ninth edition of Dean and Ben Talk Tech, a podcast between, uh, two analysts on telecoms policy, broadband, and related topics. How you doing, Ben Mark. Hi, Dean. Yeah, I'm good. I'm good. Uh, we're, we're past, uh, high heats, so it's quite pleasant, uh, here in France today. Yeah. Back to normal in London as well, at least for now. So, uh, yeah. But you, you haven't been in London much though, have you? No, I'd say, um, the last month or so since we recorded the last podcast, um, I've been to North America. I can't remember if it was like two or three times I went to, you know, connect X in Florida, which was sort of, you know, infrastructure and towers and small cells, Canadian Telecom Summit in Toronto, spectrum a and AI thing in Washington. And then a couple of, um, other things on, um, AI and voice, um, you know, vendor specific things. I went to a, a moderator thing for Amoc in Madrid and, uh, net axis I spoke at in, in Italy on, uh, telecom sovereignty. Come back to that in a second. And I was at the, uh, cloud Communications Alliance last week in Amsterdam. And a real key theme that's come out of all of this is sovereignty. And I know we can also talk about what the European Commission announced last week, um, with its, uh, sovereignty package. And there's a couple of things that I'm seeing. Um, one is that operators are looking at the opportunities from this requirement for sovereignty and what that means in different geographies. Yeah. Aligning with policy and regulatory trends. And there's different ones in Europe and in Canada and UK and that, that really drives down to or under potential to offer, um, sovereign infrastructure services. Like you've got companies like Telus and Canada and BT building data centres. There's a lot of interest in, um, localised AI factories, either for government or academia to get access to, uh, secure, uh, sovereign resources. And then there's point services like unified comms and voice and video conferencing. But there's a second side of it that I suddenly realised last week, um, looking at some of the European, um, cloud and AI Development Act details, um, particularly relating to public procurement of cloud services in Europe. Um, and essentially what is happening is that, um, not just in telecom, but generally, um, there's a big push to classifying cloud services based on four tier model for sovereignty. And that's going into the procurement rules, particularly for public sector organisations. And there's sort of differing levels of sovereign ness, um, how much is, is run in Europe, how much is owned by Europe, how much is deployed and, and, and operated by European citizens. And whether data can be used ex extra territorially or not. And I mean, a lot of this is, is driven by. You know, concerns about, uh, EU sovereignty on, you know, hyperscaler versus hyperscalers and cloud and AI models. Um, and there's a parallel stream around chips as well. But I think that, that some of the telecoms operators may get caught the crossfire here, especially if they're, um, offering services to government. Um, because even the lowest tier of the proposals would have real implications. Even if you want to provide unified communications for a local council or something like that, um, you know, people collecting the bins or, or collecting local taxes or, or whatever, you would still potentially be ca captured by the new, uh, public sector procurement rules. Plus, and this I don't think is getting discussed. Telecom's networks these days are cloud-based, you know, whether it's a 5G core network, whether it's a virtualized, um, BNG in the fixed network, whether it's the OSS and BSS stack, all that's cloudified. And presumably that means that if you are an operator offering any services to public sector in eu, part of your stack is cloud-based. And some of that will be based on technology or companies that are not EU based. You know, whether that's for the core network, whether that's the support desk via your, you know, radio network or whatever. Well, are, are you seeing anything like that Benoit? Because it, to me that, that that's under explored well. I agree with you. Now, it, in my mind it ties with something that I found, uh, quite interesting, uh, also quite contentious. Um, France, as you may or may not know for a number of years now, has had a rule saying that any cloud service provided to public entities has to be at least 50% owned by, I dunno if it's a French or a European company. I think it's European because they wouldn't be allowed to, uh, to, to have national preference. Um, so for example, uh, I know of a number of, of joint offers between TELUS and various US companies to offer what they call sovereign cloud. Now, I haven't looked in detail at what those contracts look for, look like, et cetera, but I think initially this was primarily a way to circumvent the kind of us extraterritoriality on, on data. Um, but I'm wondering if potentially there isn't a thread there as well. Um, you know, for maybe one of the large telcos to actually take a leading role in saying, well, we need to have a similar cloud approach across Europe and, you know, take a strong position in the cloud services provided to telcos in order to avoid that. Now, I, I'm not a legal guy in any way, shape or form, so I don't know how far that would go. I know that France has been pushing to make this a European role and until recently there was a lot of pushback. Um, but I am wondering, in fact, it ties in with something I've been thinking about ever since. Uh, Trump came to power and, you know, uh, Trump two came to power and, uh, you know, started, um, kind of antagonising Europe in a, in a much more forceful way than, than even during Trump one. Um, and, and one of the things I wondered about is, you know, is there a point, is there a breaking point where these big us, uh, tech companies start thinking, we need to be less associated with the us We need to be seen as, you know, global companies that have capital from all around the world, rather than being owned by US interests? Well, that's, I, I don't know, but, but it might, it might make sense for some of them to actually consider this as geopolitics become more and more, uh, of a factor in the equation. Uh, I mean, I, I, I think geopolitics is a huge factor here, but at the same time, I think there's, there's double edged swords for everyone. Um, so I think, I mean, we we're recording this on the day that supposedly, you know, Trump is about to, um, announce his investments in AI companies, probably pre IPO. We've got the open AI and ANTHROPIC IPOs scheduled for the next few weeks. So, you know, if anything the US government is, well, it's already taken stakes in Intel and a number of others. So you, there, there's a, there's increasing difficulty to decouple and you see u um, European technology companies increasing their US operations to avoid the risk of sanctions in Europe. I think there's, there's a double-edged sword for the operators. I think the operators and the cloud companies in Europe, the local, you know, sort of cloud and hosting providers see a big opportunity for offering sovereign services. But at the same time, they themselves are dependent on non sovereign capabilities and, you know, ingredients and not just in their network, but in their IT stack, which can't be replaced. And I know there's things like, there's, you know, project Silver with a, with silver, with a y, um, which is this sort of open source, um, telco cloud initiative, which would, you know, have, have a high sort of European content, but you can't just rip and replace existing cloud native core networks that are still being deployed for 5G or your O-S-S-B-S-S stack. You can't just switch it over to a, um, a sovereign alternative next Wednesday. Yeah, that's a multi-year, perhaps even sometimes decades long project. If you, if you look at the huge transformation challenges that even existing cloud moves take, if you then, then say, well go back to the drawing board again and add sovereignty. That's basically takes out multiple years of development. So it has to be whatever it is, it has to be done in progression. And I think that the, the, the executives of the operators and and suppliers need to balance their desire for sort of external opportunity and selling cloud services and sovereignty. So, and sovereignty, uh, sovereign data centres with the looking internally and go, well, how much does this actually hurt us? At the same time. Yeah. Yeah. Well, I, and to be honest, I don't even think they have seen that angle at all. Um, or, or if they have, they've been keeping very quiet about this. Uh, I mean, this is, this is why I've been talking so much recently at events about telecom sovereignty as distinct from tech sovereignty or cloud sovereignty. I think that the telecoms industry, whether it's in Europe or globally, needs to ban together to say, look, these are the touch points of the sovereignty debate for the network, for telecom services, for our, our own IT and cloud stack, um, and other dependencies on supply chain. Uh, and I think that there's a, a real tension as well between the approach in Europe, um, which is very much about strategic autonomy for cloud and data residency and AI and chips. And what I see more in Canada in the us, which is more about more security and resilience first, where to some extent it's like, we don't really mind where the thing you are buying is from as long as it passes these certifications. We've got these test labs, you know, there might be exceptions for Chinese or Russian vendors, but, you know, there's high risk, but we recognise that, that there's a bunch of this stuff that's gonna come from the US or Taiwan or Japan or wherever. And I think that whether Europe can balance the, the sort of, you know, the, the drive for, you know, strategic autonomy with the near term requirement for resilience and securities is a important one. Yeah, no, I agree. And, and then, you know, the flip side of that is if you take the citizens' view, so just to give you a, an example again from France, uh, there's a report that was released last week by the French Parliament on the use of AI and parliamentary work. And it turns out that self-declared, so it might even be understated, but self-declared. There's 30 to 40% of parliamentary basically put every amendment through an ai, um, you know, engine and, and all of these, not all of these, but you know, massively, this is US AI engines. Yeah. And, and so, I mean, this is not even about extraterritoriality because basically the minute you push this into an AI model, it becomes data they can use for whatever they want. And, and it raises huge kind of, this is the heart of government in theory. Um, and so, I mean, obviously as a citizen, right? There's one part of this is, is the, the kind of shortsightedness and callousness of politicians who are just going, oh, here's a shiny new tool. Let's not think about it for a second and just use it. But, but also more generally, you know, it raises really important questions about how secure the political process is from, from, you know, at least some sort of prying, if not interference, because obviously AI is also recommendation engine. So a and, and well, and, and that this is before we get into, you know, the, the last month or two, it's hot topic of agen AI and intent based and, you know, leaving your personal agents to get on with whatever it is that they were doing. Yeah. Um, and, and, and, and, and or yeah. And, and it's whether it's your personal agent optimising your personal workflow, or it's an agent AI system inside an operator or, or a public sector organisation that's doing stuff based on intent. Um, and you get into a whole new realm of guardrails and all the rest of it. Some, some which is, is outside of my direct focus, but I'm, I'm increasingly aware that the sort of agen orchestration and, you know, do you know where your agents are and what they're doing, uh, is a, is is another emergent question. Yeah, absolutely. Right. So long story short, we, this is not the last time we're talking about sovereignty. No, no, no. But I, but again, I'll reiterate, we really need to have previously warring factions of the telecoms industry. So that might mean, you know, connect Europe and the competitive telecoms association and the tower companies, and the MVNOs and the CPAs players, and the MSPs all getting together and going, Hey, you know, telecom sovereignty is important for all of us. Yeah. We know that the politicians are focused on hyperscale, cloud and AI and data centres and energy and power and water and absolutely defence and everything else. But let's at least have a framework which doesn't add in telecoms as an afterthought, because at the moment, it very much is. Uh, I absolutely agree with you. Great. Um, so flip me over to you. What, what, what, what have you been up to? Have you, you know, so I've not been travelling at all, believe it or not, it's the first time in ages. Uh, I've been at home for at least a month, but I have, uh, I have done quite a, a bit of, uh, exploratory work and collaborations on, uh, the intersection of policy and, uh, real, real life data, let's call it that. Um, so a couple of weeks back, I participated in a panel that was organised by ler, where they showed some of their latest findings, um, and, and asked me to comment on, on some of these. Um, essentially this was focused on fixed, and it was about basically, uh, uh, three topics. The first one was, um, how despite adoption of fibre, the, there are wild differences in actual quality of experience in, in different markets, uh, because of all the other parts of that chain that are wildly different. And in particular, adoption of latest, uh, wifi, uh, protocols makes a massive difference to, uh, the end user quality of experience. Uh, the second one, which I think is to me the most interesting one, was looking at how, um. Despite, again, being on fibre, uh, distance from the nearest IXs or the nearest, uh, content servers, um, actually had a huge impact on latency. So, uh, looking at, you know, other types of discrimination that are not based on your access technology, but actually on like physical geography and, and stuff like that. Um, and then the third one was on satellite, which is in and of itself a really, uh, interesting topic. But if I can come back to the second one for a second, what, what I find fascinating is that to me, to me, that's an absolute blind spot of policy. Um, in other words, I don't know of any policy in Europe where a regulator would look into where internet exchange points are and how closely that maps to the national, um, geography. But now we have measurable elements that show that it has an impact on quality of experience. Um, so to put another way, speeds are similar, but latency is wildly, uh, degraded If you're far away from an high speed, now, you know, who does this affect right now? Probably gamers and a few kind of power users. But down the line, if particularly if we say as, as the EU has that latency is an important metric in terms of, you know, delivering future applications, then this is something that people are gonna have to look into. Uh, and, and, and this, this is what annoys me about, about this whole discussion about, you know, edge, which we've had for the last decade at least, and is coming back again with physical AI and edge inferencing for, for AI is actually, it's edge interconnect is at least as important as edge compute. And that everyone's so busy to say, oh, we want to put GPUs at the cell cell site, or in the local exchange, or whatever. Well, actually you wanna have a much denser grid of interconnections, whether it's a traditional peering point or some other forms of transit, or even between mobile operators where they've got dedicated interconnect networks. But, you know, as we move, whether it's towards agen AI or even just workloads, which are chatty Yeah. Where you've got the moment, a lot of requests going out to the rest of the web. You've got content delivery networks. Um, even also things like checking for, uh, cybersecurity risks. You need to have access to everything else. And instead there's, we're still focused on this idea of can we have, you know, workloads running as close to the user as possible in the network, the network edge. Well, that's fine for things that you can put wholly at that local location, but most of what we do, I mean, what we're doing now is, you know, with this call is cloud plus interconnect in various ways. Um, yeah. And that's such a blind spot. I agree. It, it, and also you can, you can kind of flip that around, right? There's a reason that these ips are not there very locally is that, that there's a scale issue in making it work financially, right? And so the thing that has always puzzled me about, you know, edge, whatever we wanna call it, um, is that, you know, where's, where's the actual economic equation, right? I it's not a matter of, you know, having some space and putting a data centre in there and, and it's a matter of is that actually gonna be profitable to anyone? And, and this is where again, you know, the, and it's a tricky question for policy makers because as always, we know that there is a digital divide and it's a legitimate goal of policy to try and minimise that. Or, I mean, I don't think it'll be fully eliminated ever, but, uh, it so it, there, it's legitimate to do that, but it requires money to come from somewhere and at some point the state cannot fund everything. So I, I, I, I think it's a really interesting topic because, um, and this is where having access to kind of real life data measurements is really useful because you can see how big. The problem is, and then decide whether that's an acceptable problem to have or not, and in which circumstances you should intervene as opposed to some kind of blanket policy, uh, just on the principle of differences existing. Yeah. Yeah. And I, I think that, that there's a number of initiatives and you mentioned yeah, for open signal and crowdsourcing, but there's a number of other initiatives around improving data collection and metrics and KPIs. Uh, and I see that, yeah, there's, there was, remember being in a session in, in f last year, the sort of open source tele telecom infrastructure project talking about quality of experience metrics and how we can get them what QOE means. Um, and I think that's a really important area of focus. And, and I think the problem is that it's intention with some of the other policy angles about competition, about sovereignty, actually. I mean, neither of the companies you just mentioned are European based, and so presumably they, they've got a, they've gotta process their data locally. Is there a European model? I don't know. Um, and you know, so that's a good question actually. So, and, and yeah. Um, it's interesting and, uh, so I've been talking to Open Signal a lot lately as well, and uh, I'm sure you've seen that Ofcom last week released a new position paper on, on quality of experience for mobile. Yes. Um, and that's based on open signals, uh, measurement philosophy, I guess not just the numbers, but actually the, the kind of approach as well. Yeah. Which I think is, is quite interesting to see a regulator very openly saying, okay, this is something we have to outsource and we have to outsource it to people who know what they're doing. Yeah. Um, but one of the things that struck me, and I think, we'll, we'll probably have more time to discuss this once it moves to the next step, and we get some feedback from operators and, and, and other people in the industry. But one of the things that struck me is how, how small the speeds are that are considered, uh, what, what, um, what open signal calls excellent, consistent quality. So we're talking five megabits per second download speeds on mobile, and I think it's 1.5 up. Um, and. This is so far removed from the headline speeds that I think it also prompts a question as to whether policy in general, and I'm particularly thinking about the European Union, when I say this, should be more focused on, you know, kind of metrics that are relevant to user experience rather than headline speeds that actually say very little apart from the technical capability of the network technology. Yeah. Yeah. Um, and, and, and it, it struck me as I was skimming through the soft com document that, you know, the, the what's currently in the DNA and in the, uh, you know, Europe, uh, targets, which are not specifically in the DNA, but there's a, there's a tie in between the two, the digital decade stuff. Um. Is actually very superficial and very artificial. It doesn't relate very well to real world experiences. And it's, I mean, I, I I, I talk about we need to have a focus on good metrics, not easy metrics. And too often we focus on the stats, which we can get out the back of the network. You know, there's a nice easy reporting dashboard, but that's not good enough. And, and frankly, you know, with crowdsourcing and Frank or also even AI estimation of some of this, we ought to get better to the numbers, which mean things. I think it's actually worse than that because it's also the lowest common denominator of 27 countries that talk to their own telcos and what the telcos tell them they can get. So not only are we, you know, getting the easiest metrics outta the networks, but we're also getting the easiest metrics out of the least kind of modern networks. Yeah. And, and it, it, and there's pushback and there's pushback from the operating. Oh, it's a burden to collect the data. I'm sorry, but you've all got nice shiny new AI systems. You keep talking about. You ought to be able to have a nice interface that goes into all your, now, now, you know, collated data saying, give me the answer to these questions that should be getting easier based on what we know of, of automation and ai. Yeah. And if there's one thing where we know AI functions relatively well, it's collecting data and, and not having to interpret it in any way, just showing it in a way that is relevant to whoever wants to see it. No, I agree with you. I agree with you. So I think, uh, I think that's another, for me, it's another kind of long-term, uh, line of thought is, you know, this intersection of, uh, metric collection and, and, uh, policy and, and basically how does it allow us to answer some of the, either the blind spots or the questions that have been very hard to answer until now, and, and how do we slice the data in a way. That allows us to have some insight as to how this could be addressed from a policy point of view. I think that's a really, really interesting question. I mean, I agree. And the thing I, and I, I'm, I'm gonna do a review of the, the off comm, it's called connectivity. You can count on, I've got a half written review that I'll put on my substack, but the, they break it down between, uh, dense urban areas, rural areas, uh, indoors and on trains. Uh, UK has a particular problem with train connectivity, and, and I'm, I'm a big, you know, I'm, I'm with trains in general. Yeah. I would say, I mean, in indoor wireless is, is longer a preoccupation mine. I'm, I'm running a workshop on indoor wireless, which cover some of these on July 16th. So quick plug for that. And early bird tickets. Look at, look at my LinkedIn and I, I, yeah. Show and the show notes. Um, but a key thing here for indoor is, you know, actually you can start to both measure it with crowdsourcing much better than the traditional drive tests, which are outdoor coverage. Um, and you also, there are new technologies available to enhance quality experience for indoors, whether that's neutral host in public settings, which you gonna offer, you know, better design and capability where the building owner potentially pays for that because they're either they're mandated to do it, um, or because they need to offer good services to their, their visitors or employees or tenants. Um, and, and so I think that there's definitely a role for QOE monitoring and measurement, but it has to be in context, not this sort of nationwide aggregates or, you know, overall we do this number of petabytes month. That's, that's no use. We wanna have speeds where people are, oh, it's very limited use then. Yeah. Yes. Yeah. No, no, I agree. I, I agree with you. Yeah. And, and by the way, do you think that HS two is gonna put as much money into, uh, getting good connectivity as they put into getting bats across the, the hills, or that's whether it's Frank, frankly, I mean, is is actually HS two is, is the, the least of the problems is, is normal sort of long distance trains and commuter trains and tunnels, but Yeah. Oh, don't, don't get, I suspect comm Yeah, I suspect commuter trains is a big, big issue. There's, I mean, there's a whole, I i, I, I wrote something about the train connectivity. It, it is one of those things that takes forever. Um, the UK actually is star. The, the government has actually got a policy now. They're gonna be part funding putting, um, I'm not sure if it's specifically Starling, but satellite backhaul for better on train wifi. And there's this debate about should you have wifi on trains or should you just use cellular either through the windows or with repeaters? And I think you need both. And so having starlin collected, um, wifi on trains would be fantastic. I've been on a few aircraft recently with fast satellite collectivity, and it's, uh, yeah, you, you can actually do proper work rather than the old sort of, you know, maybe if you're in over the right part of the ocean in the right place. Um, yeah, yeah. So, so if that's replicated on trades, that'd be a very good thing. I agree. Right. Alright, so, um, so what, what, what are your plans, like very quickly? Yeah. What are your plans for the coming few weeks? Right. Um, this week I'm heading to Lisbon for Amplify, which is the sort of network as a service, uh, industry association. I'm in Brussels next week for, uh, a Spectrum event and also something about D 2D satellite policy. And then I'll be at the TM from DTW in Copenhagen. And then as, as I mentioned on July 16th, I've got my own indoor wireless workshop in London. Um, yourself? Um, I don't have a lot of travelling planned, but, uh, I think, uh, one of the big topics in the coming few weeks is, uh, that the kind of network fees debate is rearing its ugly head in non-European geographies. Mm-hmm. Uh, and so I'm gonna be focused, I think a lot on, um, I guess kind of educating other policy people on, you know, the, the generally bad ideas Yeah. In, in this space. Uh, but it is, it is tricky because it, it clearly, you know, the, the European stance has, uh, emboldened a lot of telcos around the world to say, Hey, this is a new source for revenue for us. We should, we should fight for this. Yeah. And I know GSMA and Connect Europe and others are sort of going around the world stalling it. Uh, I've heard things from India and I think bits of latam and Africa on that. Um, so yeah, so one level it's the, the usage charging. And the other hot topic, which maybe come back to another one is, uh, lots of operators, especially in the US and bits of, uh, developed Asia, talking about token charging and can we sell AI tokens? Yeah. Um, which, uh, that's a discussion for another day, I think. Yeah. Alright, well, excellent. Uh, that was a, another good one. Yeah. Uh, and, uh, I'll speak to you, well, I'm sure I'll speak to you before a month, but, uh, in public, I'll speak to you in a month. Excellent. And, um, to everyone else, uh, please, uh, like and subscribe and reach out to either of us, um, whether if you want further questions or you want us to come and talk to you about things. Thank you. Bye-Bye.