Dean Bubley: [00:00] The thing which is missing in my view is the first customers for the APIs should be the internal product teams at the operators themselves, and they should be creating solutions with their own products, their own APIs, and then the second group ought to be other operators who are consuming almost, like, on a wholesale base between them.
DR: [00:20] Sure.
Dean Bubley: [00:21] And then, to some extent, the enterprise and long tail developers, they come after that, because, ultimately, if the operators themselves aren’t using, whether it’s anti-fraud, or the network APIs, why would anyone else trust them to do it?
DR: [00:32] Yeah.
Announcer: [00:38] This is Telco in 20, a podcast that helps telco execs achieve a competitive advantage with AI and the public cloud. It is hosted by Danielle Rios, also known as DR. Today we’re talking to Dean Bubley, founder and director of Disruptive Analysis.
DR: [00:57] Hi, guys. I’m DR. MWC Barcelona is right around the corner. And can you believe it’s been 20 years? 20 years of Barcelona being the industry’s biggest stage, 20 years of bold predictions from that keynote hall. Some have landed. Like, how in 2006, Richard Branson said, “Watching TV on your phone would feel as natural as watching at home.” He was totally right. Or how in 2017, Reed Hastings said, “All video would be on the internet within 10 years.” Nailed it. And I’ll take credit for my own 2021 CLOUD CITY prediction that the public cloud was coming to telco. You’re welcome. But there were others that didn’t pan out. In 2014, John Chambers told us the internet of everything would be 10 times bigger than the internet, or how in 2016, Zuckerberg said VR would change the way we live and work. So, when GSMA rolls out the themes for MWC, it’s fair to ask, “What’s real?” And, “What’s the next Metaverse?
[02:00] There are four major themes this year. “Intelligent Infrastructure” is all about autonomous networks, and 6G. “Connect-AI” focuses on network APIs and whether operators can turn their networks into platforms. “AI for Enterprise” will discuss AI factories, GPU-as-a-service, and private networks, and how operators can generate revenue from enterprise customers. And “AI Nexus, ” which will cover AI regulation, digital sovereignty, and data privacy. Who better to talk about this than someone who’s been analyzing our industry for decades? Today’s guest, telecom analyst and consultant Dean Bubley is the founder and director of Disruptive Analysis. We’re going to tackle MWC’s biggest themes and discuss how operators are adding AI to their networks, if network APIs and hosting enterprise AI workloads can actually generate meaningful revenue, and what the future holds for AI regulation in our industry. So, let’s take 20.
[03:02] Dean Bubley is founder and director of Disruptive Analysis. Hi, Dean. Welcome to Telco in 20.
Dean Bubley: [03:08] Hi, DR. Great to be here..
DR: [03:10] Yeah. I’m so excited to have you on the podcast, the first podcast of the year. Everyone knows you as Disruptive Dean, which is also your handle on X. And MWC is right around the corner. And so, for this episode, we wanted to tackle four big telco topics for the new year. You’ll give your take, and then I’ll jump in with mine, and I’ll say it right now, we probably won’t see eye-to-eye on every single topic, but I know we’ll both have something super interesting to say. And so, are you ready?
Dean Bubley: [03:39] Absolutely. Go ahead.
DR: [03:41] Awesome. Let’s start with AI. How can you not? Everyone’s talking about it, and everyone is starting to think about how their networks can become AI-driven automated value platforms. And so, it was a really big topic at last year’s TM Forum Ignite, and it’ll be a big topic at MWC in a few months. And so, how do you see people adding AI to their networks?
Dean Bubley: [04:04] I see it in multiple different dimensions. I’m not seeing one singular monolithic push to autonomous networks, but I’m seeing completely distinct movements if I look at the radio network in mobile, if I look at the access network in fixed, even submarine cables and satellite, and then spectrum policy, and spectrum management, and automated allocations. As you move towards, say, 6G and wireless, there’s enthusiasm around a concept of AI native networks. I’m a little bit wary of that, because it’s moving so fast, and it’s so ill-defined. They’re trying to standardize, and in trying things in standards when there’s both technology evolution, but there’s also regulatory divergence between different parts of the world. We end up repeating the whole problems of cloud native where we ended up with non-standalone and then standalone, and I’m wondering if we’re going to have non-AI and AI. Because there are some people that want to be able to turn the AI off.
[05:06] Europe wants explainability, China wants models to be registered, the U.S. has its own rules. I’m not sure whether we’re going to get one concept to rule the world on AI. But I do see it permeating pretty deeply across everywhere.
DR: [05:20] Yeah. For sure. I mentioned TM Forum. There was a lot of talk about it. According to their standards, they have a certification process, 21 operators have gotten something, like, 35 certificates proving they’re becoming more autonomous. And, obviously, with more autonomy, you’re going to add more AI. And I think AI, like you said, is a very broad topic. There’s a lot of still unknowns, and I see a lot of operators talking about adding AI to their networks. It probably wouldn’t be the first place I would advise them to add AI in their enterprise. I’d probably tackle some of the easier concepts. There’s some real big things that you got to go through. You’ve got to go through the cultural change of adding AI to your organization. Finding the talent that know how to really do AI at scale is pretty scarce, and expensive. And so, I think they’re going to have to do a lot of training to build the talent base that they need to do this.
[06:13] And so, I might tackle some easier topics like BSS, or the back office while some of these other regulatory model shifts, definitions all settle out.
Dean Bubley: [06:24] Yeah.
DR: [06:24] So, the next big topic we saw at MWC 23 is the idea of network APIs, adding an API access layer to let developers dynamically and programmatically get access to the network. Ericsson has assembled a whole network API stack. They announced Aduna. Certainly, they bought Vonage. Nokia has its Network as Code platform that they’ve been really touting and pushing. And so, we keep talking about this in the industry. Do you think this is a big idea that can really add, let’s say, 20 percentage points of growth for operators? Is it really going to work?
Dean Bubley: [06:57] I think that actually while all the noise has come out of MWC in mobile, all the action is on fixed and enterprise. And so, it is things like, say, the Amplify Network-as-a-Service APIs for large-scale backbone and transport connectivity. That’s already there. I think it’s more difficult on mobile networks. We’re seeing early moves around things like the anti-fraud and identity APIs, which makes sense. Although, there’s multiple different ways of getting that if you’re a developer.
[07:28] I think the hard network APIs of things like quality on demand, and I’ve self-declared myself as a slicing skeptic, or occasionally slice denier, because I think there’s just some physical problems with running the wireless network. So, that makes it hard.
[07:42] And the other question is the value chain of getting the APIs to developers. So, there was something recently where it looked like you had the operators that had Aduna as an aggregator, Vonage as the CPaaS provider, and then they were having something in AWS marketplace. And I’m like, “That’s a long value chain to get.” I don’t know how much it is per API call.
[08:03] The thing which is missing, in my view, is I think the first customers for the APIs should be the internal product teams at the operators themselves. And they should be creating solutions with their own products, their own APIs, and then the second group ought to be other operators who are consuming almost, like, on a wholesale base between them. And then, to some extent, the enterprise and long tail developers come after that. Because, ultimately, if the operators themselves aren’t using, whether it’s anti-fraud or the network APIs, why would anyone else trust them to do it?
DR: [08:36] Yeah. I think that’s actually a good idea and it’s something that Jeff Bezos did at Amazon in the early days of AWS where he imposed a rule that none of the teams within AWS could access the other services without going through an API layer. And that’s what really built the products from the beginning. They “ate their own dog food.”
Dean Bubley: [08:55] Yeah.
DR: [08:56] And it was a really good way that they started to develop the services. And so, I really like that idea. But going back to something else you said, what’s really key about monetizing this idea is where are the developers? We have all this activity in terms of the technical design and getting them to just work, but you’ve got to have the developers.
Dean Bubley: [09:13] Yeah.
DR: [09:14] Ferry Grijpink over at McKinsey, those guys are saying it’s worth maybe a $300 billion market. But I just looked up some quick math. The world’s best API company is Twilio that made a ton of money across not just network APIs, but a ton of different APIs. And they have 10 million developers and $5 billion in revenue. Just five.
Dean Bubley: [09:34] Yeah.
DR: [09:36] And Vonage is at 1.6.
Dean Bubley: [09:37] Vonage is like Twilio, and they make most of their money on things like SMS and calling, and things like that. It’s not the network APIs, it’s the communications APIs. Which, at the moment, seem to be in this isolated little world alongside it. So, I’m a bit wary of that.
DR: [09:51] Yeah. We got to find the developers, and your idea of maybe starting with internal is good. If they won’t use it, how will enterprise developers use it?
Dean Bubley: [10:00] Yeah.
DR: [10:00] All right. So, the next big topic I think for 2026 is AI data centers. Some people refer to it as GPU-as-a-Service. We saw some big telcos make some bold moves last year. SK Telecom is massively expanding its AI data center footprint. Deutsche Telekom is building one of the largest AI factories in Europe with NVIDIA. And Verizon is pitching its network and data centers as infrastructure for GPU clouds and enterprise workloads. But the hyperscalers and other big tech companies like Oracle are doing this too. So, do you think the operators can really compete at this level of spend and speed of execution?
Dean Bubley: [10:36] Generally, no. I think that for their own use cases, the operators will have their own capabilities and infrastructure in-house for running the network in particular. That makes sense. I think there will be specific use cases in some parts of the world if it’s sovereign AI where the government, whether it’s a regional government like the EU or whether it’s a national government has sort of anointed the telco as the super computing provider of choice, which is a bit of a special case. But even in those cases you’re seeing … From memory, the SK Telecom one in South Korea is actually in an Amazon building. And Amazon is sourcing the energy. Energy is another thing here. If you really want to start playing with the big guys on AI infrastructure as a service, there is a question of where is your next gigawatt of power coming from.
DR: [11:25] Yeah.
Dean Bubley: [11:26] When I looked at this a few months ago, there were no telcos building nuclear SMRs in a 10 year horizon whereas all of the hyperscalers do. You could argue that’s a straw man, but the point is that most of them are saying, “We’re going to reduce our energy use overall.” And they may even have legal requirements. And, at the same time, they’re like, “Oh, we’d like to have a few extra hundred megawatts of compute.” And I’m like, “That doesn’t work.”
DR: [11:47] It doesn’t match up.
Dean Bubley: [11:48] And also sometimes a gigafactory idea makes a bit better sense, if they can get sponsorship from the government, or something like that. The thing I really don’t buy is that we’re going to have GPUs in the radio network, or the fixed access network, and then we’re going to sort of Airbnb out the spare capacity. I mean, to whom and doing what?
DR: [12:07] This is that AI-ran idea that NVIDIA came out with SoftBank last year.
Dean Bubley: [12:11] Yeah. I’m sorry. Who are the developers and how do they get access? If you’re a pharmaceutical R&D company, are you going to want your protein folding algorithm running on cell site number 3284 today and another one tomorrow? No. I don’t think you are. Is there a role for telcos in AI infrastructure? I think in terms of the connectivity that’s where the obvious money is.
DR: [12:32] Sure.
Dean Bubley: [12:32] But actually in terms of running GPUs or TPUs and providing it with energy and cooling, I don’t see that they can differentiate. And I don’t buy this Edge AI inferencing. It’s the same Edge compute story we’ve had for 10 years, but with a slightly different application workload.
DR: [12:48] Yeah. I totally agree. I think the telcos are fabulous businesses, tons of capability, but we’re in the super big leagues now when you’re competing against Amazon and these guys.
Dean Bubley: [12:58] Yeah.
DR: [12:59] The chips are moving so much faster. You can’t be doing Moore’s Law math on the H200s you’re buying. Let’s just talk about just the machines and the chips. Jensen Huang coined the phrase Huang’s Law, which is chips are getting 15 times better every two years. Are you really going to be able to play that CapEx cycling game and keep up with it? And then the software needed on top of the chips to provide the service, I just don’t even know how you guys are going to compete.
Dean Bubley: [13:26] And you get back to the API game and how you deal with developers.
DR: [13:28] Yeah. I like your idea, focus on connectivity. I think there’s going to be a big need for that, let’s go in, the thing that we do very, very well.
Dean Bubley: [13:34] Yeah.
DR: [13:35] So, our final topic is Europe’s favorite, regulation. You’ve brought it up a couple of times, but let’s talk about that now. A few months ago, Deutsche Telekom and Orange announced a joint alliance calling for more regulation and advocating for a stronger focus on European cloud solutions. This will probably be a very big topic at MWC and beyond, digital sovereignty, data privacy, and just regulation of AI in general. And so, do you think AI needs to be regulated at this stage?
Dean Bubley: [14:03] I think aspects of it already are. I know that the EU has had various attempts at this, and it’s recently published what it called the omnibus, which was trying to tidy up some of the rules. I think the initiative you mentioned was, specifically, about data sovereignty, about where is data kept, and where does it go when it’s in transit, and is it on a European-controlled server or cloud? And, in particular, is it subject to extra territorial rules? So, to some extent, it was more a case of I think making sure the EU data, particularly, for things like governments, or defense, or financial institutions, or healthcare, isn’t subject to, say, U.S. or China laws for accessibility.
[14:47] And I think data sovereignty is very different to the technology sovereignty. Who owns the data centers? We have seen Amazon and Microsoft and Google set up European arms, which looks like they’re functionally distinct and sovereign. There’s data sovereignty, a legal sovereignty, and then at the far end of the scale, realistically, there’s not going to be a European NVIDIA.
[15:10] There’s not really going to be a European hyperscaler. But I think there is a lot of interest around what does sovereignty mean? Shortly after this is going live, there’s the European Digital Networks Act and that’s coming out supposedly January 20th. And, for me, one of the interesting things is whether the European AI regulation aligns with the European telecoms regulation that’s being proposed. So, for example, one of the things that the telecoms regulators have been talking about is regulating IP interconnect, and peering, which, for me, is absolutely crazy, because we should be incentivizing a dense mesh of interconnect, not trying to regulate it and allow a few telcos to almost act as gatekeepers at the IP layer. So, is it going to be regulated? Yes. Is it going to be regulation which makes sense alongside all of the other regulation? Don’t know yet.
DR: [15:57] Yeah. My take here is I get it. You need to be safe, you need to protect your country’s assets and infrastructure and data. But telco, in general, always misses the boat on the technology shifts. And our instinct is to be like, “But, wait, regulation” instead of racing to plant our flag. And there’s a balance between do you really want to make a lot of money and be constantly complaining that you’re not as successful as the tech companies but at the same time your instinct is to go fight on the regulation angle first. And so, I’m a little bit on the commercial side, “Let’s go use it. Let’s see if we can actually get this working for us.” And then maybe go fight all of this regulation stuff, and that’s my take.
Dean Bubley: [16:39] I agree. And I think it also comes back to what we were talking about before about AI for the networks, because as soon as you have AI running the network, it puts it into the high-risk threshold, because, increasingly, the networks are themselves viewed as critical national infrastructure. And so, there’s a bit of a paradox here. If you talk about AI regulation, well, the first bit of AI that gets regulated is the stuff that runs in the network.
DR: [17:01] Again, I’m a little bit more don’t run and ask for regulations first while everyone else goes on a land grab. You’ve got to really plant your flag and figure out if it works. But speaking of planting your flags, we both travel a lot. In telco, you have to. You’ve traveled to over 100 countries. And so, my question for you is which country has the coolest passport stamp?
Dean Bubley: [17:25] I think it’s my latest country that I visited a few months ago, which is the Micronesian nation of Palau in the West Pacific, and the passport stamp, itself, is quite cute, because it looks like jellyfish. But along with the stamp, they also put in the Palau pledge, they ask you to sign at immigration. It’s, essentially, leave no more than footprints. It’s one of the few countries in the world where they have a legal requirement that you need to use reef-safe sunscreen and things like this. And it’s, basically, this pledge document, which is, like, a half page in your passport, and they ask you to sign it. But the actual stamp, itself, looks like a jellyfish.
DR: [18:03] Yeah. So, I do not have a fabulous stamp like that. My favorite stamp that I received is one from Peru. It’s a fuchsia pink. You see this pink everywhere you go when you’re in the country around alpaca, outfits, and pom poms. And so, it was a very colorful, pretty stamp.
Dean Bubley: [18:20] That’s excellent.
DR: [18:21] Well, Dean, we covered a lot of ground talking about a lot of different topics, and I look forward to 2026 and seeing what happens in telco and in AI.
Dean Bubley: [18:29] Definitely. And I’m sure we’ll be bumping into each other, maybe even at one of the fringe events in Barcelona even if I don’t go to MWC itself.
DR: [18:36] For sure. Well, Dean, thanks so much for coming on the podcast. It was great to talk to you.
Dean Bubley: [18:39] Thank you.
DR: [18:40] Awesome. Stick around. We end each podcast with a Telco in 20 takeaway. I’ve got two minutes to tell you something you need to know.
[18:56] Dean and I just talked about how telcos show up late to every party, whether it’s messaging, cloud, or streaming, our industry is famous for waiting while everyone else builds. For example, WhatsApp ate your SMS lunch. Yum, yum, yum. You totally missed the boat on cloud, and Netflix built a $400 billion lunch on pipes you own. This makes me think about Benedict Evans’ latest report, “AI Eats The World.” He makes the point that unlike PCs, mainframes in the web, or even smartphones, we genuinely don’t know how much better LLMs will get. That’s why hyperscalers are betting big, $400 billion in 2025 alone. They’re racing to win, and they’ll wait to figure out the details later. The report quotes Google CEO Sundar Pichai, who believes the risk of under-investing in AI is dramatically greater than over-investing
[19:54] And Meta’s Mark Zuckerberg, who says the worst case scenario for early AI investment is that we’ve just pre-built infrastructure for a couple of years. These guys are betting big, because they understand something telcos don’t, in moments of uncertainty, fast beats perfect. But telco is treating AI like we have time. We don’t. So, stop waiting. Invest, build, and experiment, because the actual worst case, you’re still figuring out your AI strategy while someone else takes another slice of your pie. Want to see what moving fast with AI actually looks like? Come find me at MWC 26 in Barcelona running March 2nd through the 5th. Team Totogi will be there showing off BSS Magic, a way to transform your telco without having to actually transform your systems. Connect with me on LinkedIn or X @TelcoDR to set up a time to meet. Until then, tune in to more Telco in 20 episodes, like and follow, and leave us a five star review.
[20:55] Don’t forget to sign up for my awesome email newsletter on TelcoDR.com. And join 160,000 followers on our YouTube channel by hitting that subscribe button. Later, nerds.
