Episode 9: The new telecom AI era: Voice AI, autonomous networks & sustainability challenges

Intelligence and Noise: Episode 9

Peter Jarich  (00:02)

Hi everyone, and welcome to Intelligence and Noise, the GSMA Intelligence podcast, where we take a look at the week's news through the eyes of the GSMA CEO newsletter, try to put into context what's important, maybe things that are important that we're not talking about, and as the name would suggest, try and filter through the noise and provide you with some intelligence. I'm Peter Jarich, the head of GSMA Intelligence, and we've got back with us Tim Hatt.

it turns out that he was not fleeing the country and going to Belize for anything illegal, or maybe they just kicked him out. I mean, if they kicked you out of Belize, ha it's it's it's that's a that's a real I don't know what you have to do to to to flee the country and get kicked back out. But Tim is back. How are you doing, Tim? How was holiday?

Tim Hatt  (00:52)

Very good, Peter, good to see you. it was it was a lovely lovely place to go and see. I almost got trapped at the bottom of a coral reef but but made it back here, so good to be with you.

Peter Jarich  (01:03)

Wow, that's that's that's not a good place to get to get trapped. So we've got even though it is the the middle of summer and we've passed that middle of July when a lot of Europe begins to go away for their holidays, we actually still have some some some good news, a lot to talk about. So it's gonna be a good conversation. But I'm gonna kick off as is my want with a little bit of an anecdote that hopefully sets the stage. So you might remember last week I talked a bit about my saga.

To find a new phone because I broke the screen on my current one and was confronted with the fact that that phones are now a lot more expensive. But not just that, because we all know phones are more expensive because of the memory crisis. but the fact that you're seeing these trade-offs, right? The idea that to make sure phones don't get super expensive, you've got companies who are taking flagships and reducing camera quality or

Trading off different features, and that was kind of my my conundrum. What I didn't end with was, you know, what I'm gonna do is just fix the screen on my current one. 130 pounds versus 400 probably makes sense. but what I was struck by as I thought about that, and I think this is a hopeful message, is that as other folks maybe face similar decisions.

And decide that they want to fix their devices or fix their gadgets versus buying something new because of of pricing, that maybe this this helps us be a little more sustainable. Right? It it helps us think more about sustainability. And then on that note, bringing it forward to this week, I think, as many folks would know, I I spend some time, I have a house down in the Nouvelle Aquitaine.

Region of France. That is the region in which Bordeaux is in. And if you're watching this now, hopefully by the time you see this, the fires will have subsided, but they've been pretty bad. and you know, struck by earlier this year, went to you know, the TM Forums DTW event, and everyone's talking about the the heat wave crisis, and now the heat wave crisis has turned into a fire crisis, and all these thoughts about like how does this make us think about networks?

And how does this make us think about things? I think the one story, besides all the the difficulties of watching the news and seeing people and animals and everyone sort of flee from these fires and sort of recognizing that is this the potentially start of a of a new normal, that obviously there is a sustainability angle, right? And and and telcos obviously play a part in that. And we're gonna come back at the end of this of how this all maybe links together with AI and how this all comes together.

But I think more than a sustainability story, because we know telcos have been a part of that, is for me, although the sustainability part is important, I think what we will continue to see over the rest of this year and going forward is a resilience story. Right. And how as much as climate change is highlighting the fact that that telcos play a huge part in the sustainability efforts, that

They're gonna be playing a bigger part in terms of resilience. And how do we think about building networks to support resilience? How do we talk think about networks that can withstand not only heat, but are ready with backups in case of emergencies like this that we're seeing down in southwest France? And I think we're we're gonna continue to see that. And if nothing else, it helps for me put into context that having to get a new phone is nothing compared to having to get a new house.

Or having to evacuate your house and wonder about the future of it. So thoughts go out to all of those folks. but with that, let's move on to a more interesting, well, not more interesting, perhaps a more upbeat story. Tim, I think going to and as I always say, if you're out there and you get the CEO newsletter, congrats, you're important. Going to the one of the two stories that we're choosing to look at.

We're talking about voice services, which is a bit weird in an age of AI. So what was the what was the story we want to talk about, Tim?

Tim Hatt  (05:27)

So it's an announcement between UPlus and Ericsson around embedding AI into the voice capabilities, starting at the network level. Again, it's it's something we don't often talk a lot about, voice, when we're in an internet era and now we're in an AI era. but but the story here from these two companies is that they are re-looking at how to embed AI capabilities to drive a more immersive experience in the voice channel.

And and I've got some thoughts on this, Peter, but I think the way the interesting thing for me is that voice has been sluggish for a while from a revenue perspective, but in many ways that's kind of irrelevant because most operators have bundled it into all you can eat data plans anyway. So the marginal cost isn't isn't that important. What is important though is the customer engagement and the interaction. And I think that's for me where a lot of the interest is here. How do you think

about that?

Peter Jarich  (06:23)

Yeah, I mean and and for those folks who who don't know, because I think the the the voice agent coming out of LGU plus out of CREA, you you might not sort of know what that is, but but for the background here, right, it's it's LGU plus has the this voice call AI agent, XEO. they're integrating it with Ericsson's IMS platform, which makes sense. It's how you deliver voice. It does things like scam call detection, right? Hey, this is the scam call, be be careful of it.

voice detection, AI transcripts, AI assistance, my favorite auto call answering, right? It'll it'll answer a call for you to figure out if it's a call you should take or or it's a it's a person or not. I think you you're right, Tim, and and you know I I think as a lot of folks here would know, the GSMA intelligence team as part of the strategy function, we're lucky enough to work with a lot of the

CSOs and with the strategy group here and back in Barcelona, the folks at LGU Plus actually a bit of an explanation and a bit of a showcase to what they're doing with this. And you hit it 100% right. For them, it's about NPS. Right. And I think some of the numbers that they have, right, is is if I as I'm looking at these numbers, they said, you know, you ended with like like a twofold improvement in sort of brand perception.

You ended up with churn that was like one-fifth, right? You know, they they looked at their users since that they introduced this agent early in 2023, and you had like like one-fifth the churn on people who were using the app versus others. and it's really, I I think for me, the key part is it's another way of thinking about AI monetization. I've had this conversation with a number of strategy folks and a number of the operators.

And when we talk about monetization, 5G monetization, AI monetization, whatever monetization, right? Take take your pick, it it tends to be a shorthand for how am I going to make more money? Right? People say 5G monetization, and what they mean is, well, how am I making new money from new services on 5G? They're not talking about how am I actually delivering all the stuff I'd have to deliver, anyways, more efficiently. And I'm monetizing by

You know, penny saved is a penny earned. And I think the same thing with AI, but this is a reminder that monetization can be as much as I've just held on to a whole bunch of customers. And anyone out there with a subscription business that includes, you know, you and me from from GSMA Intelligence or subscription business knows churn reduction is keeping those customers that you already have is so much easier than earning new customers. So it's a really no, I you're you're you're spot on, Tim. It's all about experience.

Tim Hatt  (09:18)

And and and I want to go back to a point you made about, you know, customer experience and and ultimately this translating into customer lifetime value. The if you have ten percent churn a year and you reduce that even by a single percentage point, that scales up to a very big number across a large subscriber base. And so even if you're not making a dollar on the call itself, you're making multiples of that by retaining the customer. I also want to just put this in context for a minute.

We've seen attempts at kind of re thinking about voice for several years. We've seen new calling from the Chinese operators. We've seen various attempts to kind of bring in new features through IMS and even from messaging perspective, you know, with RCS. and and all of it, all the roads kind of move to how you make it more engaging and more immersive in what is a very noisy world.

But just want to, you I'm a data guy. want to bring in some survey data here from one of our surveys. When we asked people, this is a year ago, around their interest in new features on the voice channel, be that real-time translation, be it on emoji engagement, be it on direct immersion with the other party that they're speaking to, we got about 35 to 40 percent saying very interested or extremely interested. China was the highest, but when you look at the US, you look at Europe.

They're not that far off. And so what that tells me is that, again, people want the channel to be there. It's not just about social media. It's not just about, you know, kind of digital comms. This is good old fashioned voice still has a place. And so I think we're gonna see more of this type of activity.

Peter Jarich  (11:02)

It's a good it's a good reminder, I think. I think we hit this and I'll I'll say the the quiet part out loud, maybe for you. We know this stuff is taking place in China. You see the Chinese new calling, and you see some of the AI call assistance, and we always look at those examples with a little bit of the you know, take this with a with with a pinch of salt because what happens in China is not necessarily applicable outside of China, right? That they do some some crazy cool stuff there, but this is a reminder that

Yeah, maybe it is. You know, maybe this is one of those things. what we're not talking about here, which I think is is an interesting angle, is how does this maybe make AI features more accessible? Here we're just sort of talking about voice, we're just talking about extensions to to your regular voice calls. But if we can make AI more accessible via voice services in developing markets, whether that's for folks with limited literacy or limited phone capabilities.

But the one thing that's sort of not spoken about all this that I'd also like to see going forward, and we're not going to have an answer for this, right? Is that if the idea here is we've taken what LGU Plus has has developed and we're integrating it with Ericsson's IMS, then are we going to see a similar push for a Nokia IMS, for a Samsung s like like how do how does that portability look? Because right now it's just sort of a LGU plus with Ericsson. What about

Other vendors who could benefit, and vice versa. Right? If there are other solutions out there, how do they integrate into an Ericsson IMS solution, or how do they integrate with Nokia or Huawei or Samsung or whoever? So I think that this is interesting as a sort of as an interesting use case, but the the test for me will be to see who else.

Right, because as as much as this has been something and I'm sure that given the fact that they've had this out there for a few years, LGU plus has, you know, some cool things they're doing that other folks aren't. I I know there's all sorts of other folks that will say, Hey, you know, take a look at our solutions too. So I think that will be the the fun part going forward is to see how this extends beyond that.

Tim Hatt  (13:15)

It's a good point. And I and I think ultimately, zooming out, we're still in the kind of pre-6G, let's try a bunch of stuff before we look to standardize it. your point though on how this is interoperable across different vendors and telco networks is a good one. I think for me, the I think ultimately we will start to see standardization in the voice as we have in the messaging channel. However, I think

you're probably gonna see more difference between China and the rest of the world. I do see a bifurcation there in terms of how that i i is standardized. So this will be an interesting test taste to see how this moves out to to other vendors in in the pre six G era.

Peter Jarich  (13:59)

Well that's gonna be key for consumers to understand it, right? To to your point in terms of standardization. Cause from a technology standpoint, I mean the standards are there, right? I mean, we're talking about integrating into an IMS solution. That's obviously the foundation for for voice services, for Vulti. you know, that's out there. So the question will be how do we integrate our solutions into that? And what does that standard sort of services look like so consumers kind of understand?

The other part that I'll end on, which is interesting you talk about being pre-5 pre-6G era, is it is a little bit ironic that we're in a pre-6G era and yet we're talking about the resurgence of voice services, right? You know, of which you don't even need 4G for, right? You know, we're we're talking about advanced, and I think it is a a reminder that

that that you can layer things on top of what consumers know and maybe that's that's that's that's gonna get you a little bit farther than trying to dream up some new services. But talking about dreaming up new services,

of the services, of the stories rather that that didn't get picked this week, that could have made their way into the hands and inboxes of of CEOs around the world but didn't, is another AI related story. And I think that's the interesting part here, right? The first one is it's kind of a voice story, but it's an AI story. This one is a infrastructure story, but also an AI story. And and it's the news out of Vodafone

That and I think so many times what we end up talking about. Here's a solution that has been deployed on one site, but that that they've deployed a site and they were making a big deal of the use of robotics and AI on a mast. Meaning, what they've done is they've worked with South Korean firm Humex to attach a robot arm onto a mast in

L.

Albania, I believe, yes, it's Albania outside a shopping mall. And the idea is that the robotic arm can turn the antennas, tune the antennas in real time. is this is this the apocalypse? We're we're now we're now letting robots run our towers. how do you how do you are you worried, Tim? Are you worried about about robots on top of towers?

Tim Hatt  (16:25)

Well well just just quick I I'm not too worried about it, but maybe maybe I should be. just quick for context for so Peter explained the the the overall news story here. Vodafone, one of the biggest mobile operators in the world, it's got three hundred and seventy five million connections, predominantly in Europe, but also in Africa and and and other parts of the world, including India. This is this trial was in Albania, a relatively small country in its footprint.

But nevertheless, significance because it can be scaled to other markets. I think it's interesting here because it's an example of telcos running physical AI. Second point, it shows that you can use AI to make concrete cost savings. These are not negligible numbers. When you're an operator, OpEx costs are very significantly set in the RAN, as the main driver of it, the radio access network.

And energy specifically is still about 20 to 25% of that opex on average. We know that through our telco benchmark on energy. So, what the AI ran can do here is really help save truck rolls and the manual labor involved in going to retune and adjust the radio frequencies. It can also help deliver a better signal. And that's really what this is about as you get more congestion in city areas and another urban.

observations, you need to have your spectrum optimized to deliver that. And yes, you can do it manually, but with AI, this is saving, this is going down I think from weeks slash months to minutes.

Peter Jarich  (18:04)

Well and the and the and the the the to your point, I think the truck roll part is key here. Cause we're not talking about, you know, it's real time, but we're not talking about minutes, right? We're talking about it takes twenty to thirty minutes, they say, to to do a significant change in terms of how you're ch And that that's not nothing, right? But it's not the kind of thing that your, you know, real-time millisecond changes. Now that's that's there as well from different ways that you can work with, let's say, massive MIMO and and you can change

sort of sort of beams and beam forming, but this is sort of the kind of thing that you wouldn't be doing daily, you wouldn't be doing hourly. But now heck, you can do maybe hourly because you don't need to to to roll a truck. I think for me, two big well for me the big takeaway here is that this really helps I think it it highlights that we've got a whole bunch of terms out there that we're using.

And similar to what I was talking about with AI monetization, where we hear AI monetization and people go, well, that means how am I making money from AI? I think this is a reminder that there's a bunch of terms we throw out there that people think they know what they mean and they think very narrowly, right? So physical AI, right? You gave a good example. This is, you know, you said this is a good example of telcos and physical AI. I'm gonna bet you and and you know, I I'd encourage you to go out on the street and ask people if they think.

know what A physical AI is, don't because I think they probably look at you funny. But but I I'm pretty sure if you go out and you talk to most people and ask about physical AI big robots. Physical AI, it's rich robots, things out in the you autonomous things in the real world. No, this is a good example of of AI in a physical world thing being used to it's not sexy. It's it's not the kind of thing people think about in terms of physical AI. And flip side is AI ran. Right? When we talk about AI RAN again, you

I would suggest if you talk to to most people and you know ask them about AI RAN, they're like, well, AI RAN, that's that's how am I running my baseband? How am I taking that that base station and developing it on AI silicon, similar to some of the stuff that that Nokia has talked about and that we've talked about on previous versions of this podcast. And that that is one version, right? But then if you talk to Ericsson, they'll Yeah, I mean, I mean, we're putting AI in the radios, and and that's AI RAN, and and this

Then you can think of, well, we've got AI who which is directly controlling RAN elements at the top of a tower. And so for me, it's a reminder that there's much more nuance to all of these things than we sometimes give the credit. and that's yeah, I think I think that's that's that's important. There's two more angles there I want to get to, but are you thinking this is something?

We're gonna see coming to and you call it fair. You call it it's it's it's in one market and it's Albania. You don't start with this in downtown. You know, you don't start you you don't go to the middle of London and trial this thing in the middle of London or the middle of Tokyo if you're if you're a Japanese operator or or whoever, right? You start it in a market where if things go bad, it's probably not the worst thing. But are you seeing this thing from an environmental standpoint or efficiency standpoint coming to a a market near you anytime soon?

Tim Hatt  (21:22)

I think a hundred percent we are. I it will come in fits and starts and and certain countries will bring it sooner than others. But yes, for the environmental reasons, for the energy cost reasons, and for efficiency and quality of service, all of those four ticks mean we're gonna see more. The other point, Peter, I think is worth making here is that when we talk to your point, which is a good one, around making an abstract concept real, you know, look at where this potentially expands to.

Like we're talking about radio optimization, but what if you applied this type of solution in energy optimization? What if you applied it in terms of how you run a data center or an AI factory? What if it is applied in in a fiber context? I mean, I put those questions kind of in a rhetorical lens, but this is gonna happen. And so I think for me, the interesting point is not so much if it happens, it's more that's a when. It's more

What hard proof points do we get out of the trials that start to feed through at scale in actual operations? And I think this is going to be table stakes before long. the key question will be whether efficacy in the network signal and the customer experience is preserved. and to my point, like how you extend it across different domains of the operator business.

Peter Jarich  (22:44)

Yeah, like l l l like the first part, you're right. I mean, this is about customer experience. It it's how do we, in the example of the mall they gave, how do we make sure that during the day when everyone's in the mall, we direct the signal there. And at night when they're elsewhere, we direct the signal where where they are. You know, it it's worth noting, Huawei has made a lot of, you know, push on this site digitalization. How do we how do we digitize actual physical sites? I mean, this is a big part.

You know, the they talk about sensing and knowing where things are there. This is obviously a big part of that push, which makes me think if we're not hearing it already, that the tower folks are are gonna want to be in this story. And I think as we think about the tower industries, what you c what they've always been looking for, and tower carriers are always looking for how can they deliver more value to their customers? How do they actually extract more value from the people who are on on their masts?

Making these types of solutions available, which have a real ROI for the customers, obviously seems like one beyond the idea of obviously there's a lot of folks looking at, you know, how do you put micro data centers or stuff at the bottom of a tower? I tend to like that idea because I think we are, and I think our work has shown, whether it's the calculators you've done or the people we've talked to, that there's a lot of value in distributed AI, but it's a little bit more speculative.

Compared to, hey, I'm going to attach, and we're calling it robot. Let's just say motors. Let's just let's just be honest. Robot sounds cooler, but but we're gonna attach some motors to your to your antennas, and you're gonna get some value there. feels pretty straightforward. what's not straightforward, and I think there's an interesting question here of okay, and and in this case,

In this case, you've got Vodafone who's been talking about working with a third party, Humex in this case, but the algorithms are their own. it does bring up a reminder of we've got to be careful of how those algorithms work and how these systems are either open to integrating different models and what models have access to. And if that wasn't clear.

And you weren't thinking about that when you read the Vodafone news. And this is when we move to the favorite part of, I don't know, everyone's got their own favorite part of the podcast. I think many of you, that's when we stop talking. But one of my favorite parts of the the on our minds and and on the road, and we're not on the road because it's quiet. This leads us to the on our minds section and making that segue from.

Thinking about what happens when you've got physical things attached to AI and do you know what that AI is doing? Big news unescapable from this last week. Tim, do you have any AI agents who recently broke out of your house and started causing havoc? or or is it just, I mean, I don't, I don't, so maybe I'm not doing the right thing. But yeah, we got to talk about AI and open AI and

Yeah, what the heck's going on over there?

Tim Hatt  (26:07)

Why could we

not? Yeah. I I haven't seen any yet, but how would I see them? I'm gonna have to do a double check on the security front. But but for

Peter Jarich  (26:15)

Yeah. Exactly.

Tim Hatt  (26:16)

those of you who who have seen or haven't seen, this is the story of the open AI.

And unwittingly, from OpenAI's perspective, caused a cybersecurity breach. a group called Hugging Face, which is a sort of AI model repository.

No malicious intent, but nevertheless an incident that I think is a watershed moment because it speaks to, you know, a bigger point on

and Peter, I think the point to make here is that we've seen like AI obviously is capable, particularly as we get towards general artificial you know intelligence, what what Elon Musk and others are are touting is the holy grail. Where is the line between AI is a good thing and AI does its own thing?

I don't think we know that line yet.

Peter Jarich  (27:13)

Yeah and and and

no I mean and I th I think that's why the story is there's so many layers to it and why it is on the one hand it feels very post apocalyptic, apoc apocalyptic, whatever you want to call it, because let's just

Tim Hatt  (27:29)

Yeah.

Peter Jarich  (27:30)

let's just be clear of of what's happened, right? OpenAI has got a model. They've got one that's announced, one that that's pre release. they're they're testing it, they give it a task.

They put it in a sandbox that can't get out because heck, you don't want this thing to to to do. I mean, you're testing this. This this this could be really powerful. the AI says, hey, to solve this test, I could really use these things over at hugging face, kind of to cheat, right? To basically say, well, I mean, I mean, they've got the answers, so

Hey, best case scenario, I can go out and get all, but I can't do that because I'm locked in the sandbox. Let me figure a way to break out. Right? So you've got these models that are saying, one, the best way to solve this thing is to cheat. Two, I can't cheat because I'm broke here. I'm stuck here in the sandbox. So how do I break out of the sandbox? and then it does that, right? Now, haven't seen this very broadly, but but apparently.

It also left notes for future generations of itself on how to break out of the sandbox. Like,

Tim Hatt  (28:38)

Did

Peter Jarich  (28:38)

okay, guys, like, hey, future me. I'm gonna save you some time and leave you some notes here on how you too can break out of this testing environment. And then Hugging Face, who's over there trying to protect itself. Now, OpenAI like didn't know about any of this taking place, which is great, until you know that they see the the news and the blog post from.

From Hugging Face. Hugging Face, meanwhile, is going, okay, well, we got to protect ourselves from this thing. But these models that we have access to, these US models, they have guardrails that won't allow us, that won't do the things we need them to do. So I've got to go use a Chinese model to protect myself. Like that hits at exactly what you're saying in terms of how this stuff is regulated. And it goes to the question of

How are those guardrails applied? What do they mean for how we're using this? It is of course, open AI best says, hey, we're gonna see more of this. I mean, I I I applaud them for being open, I applaud them for you know putting on a release, for for saying they're working with hugging face.

I applaud the honesty of saying this is going to become more and more commonplace, but if that does not scare

you, that's that's yeah.

Tim Hatt  (30:02)

You you you laid it

out you laid out the various touch points there very well and and I know we want to come onto one other story, but I just want to make one other point on this if I can on the regulatory side, which is that this story doesn't exist in isolation. And if you look at the question of how to regulate AI, nobody knows yet. Nobody has an answer to that because technology moves faster than regulation. So I think you look at the EU Transparency Act, another big story in the last week.

You look on the other side of the Atlantic, US lawmakers want to invoke a kill switch. Okay, so you've got one before the fact, you've got one after the fact, and now you have this incident showing that AI can kind of take a mind of its own. I think what we're gonna see is not one overarching piece of AI regulation, but rather multiple layers of it that target content governance, that target cybersecurity, that target provenance.

And and and yes, that probably do look to invoke some type of overarching kill switch if needed, because AI is so embedded into the economy and geopolitical layers now. So this is gonna be a r I I don't know what you think, but I think this is a story we're gonna see a lot more of in the coming months.

Peter Jarich  (31:15)

Yeah, which is why it is I think particularly I mean, there are folks out there who I won't say from a conspiracy theory angle, 'cause I'm conspiracy theory, but you're somewhere saying, Hey, is this is this open AI's version of anthropic saying, Hey, Mythos is really powerful as a potential PR story? You know, is this just them like highlighting which which I think is a

bit hard just like I I I I would think they would probably do this in a less scary, a less doomsdayish version. I think to your point though, this highlights like as these things get this powerful. I mean again, I tried to make the link, right? What happens in the future? You cre have an AI agent that's out there that says, I need to do this. And the best way to do that is to get access to this cell phone tower.

And turn the antennas up towards the sky or turn the antennas away. It does highlight how we need to be looking at these things seriously and obviously building in the right capabilities, but understanding what they're doing. Because I think I think for me, there's you know that there's a handful of takeaways there, right? One, we're talking about open AI right now. I mentioned anthropic and and mythos, and a lot of these conversations are around the frontier models, but

we've seen we know what the history here is, which is the the open models, the open weight, open source models aren't years and years behind the frontier. You know, they catch up within six, nine, twelve months. We talked about that earlier in terms of the the the Kimi K3 model catching up and and and all the news it's making. So

These capabilities won't just be limited to the frontier for for very, for very long, which means, yeah, to your point, people are gonna get scared. They're gonna think about the regulation side of things, the role of guardrails, how strong or or weak they need to be. You mentioned transparency, right? We're gonna need to have some sort of agentic transparency. But then again, for me, looping it back to the the telco side of things, telcos, you know, recognizing

What they need to do in terms security. Right. I mean, there is a larger conversation there. And I was just having this discussion with a major operator last week about how much do telcos need in terms of these models? Like, like, do they need to be great? Like, like, do they always need to be chasing the latest frontier model? Do they need to be chasing those those models? And for what they're trying to do, do they need access to the latest? And

Maybe the answer is for optimization, for customer support, for customer services. Maybe not. For protecting yourselves against random agents who are trying to cheat, cheat their test and and break out. Maybe you do need the latest to to fight against them. that might be the message.

Tim Hatt  (34:21)

Yeah.

Peter Jarich  (34:24)

my story that I wanted to bring, I know you wanted to talk about about roaming and rogue agents, but my story that I wanted to come to was.

apparently Meta is is you know was one of the members of the climate group's Re 100 Re 100. it's a it's a renewable energy initiative. They're there with with a lot of other big companies, but they pulled out. it's been positioned as, if I see this correctly, positioned as a joint mutually agreed upon decision.

And I think probably makes sense because it seems like they are, as much as they're still committed and can claim to be 100% renewable, they seem to be focusing a lot more on natural gas as a solution to power a lot of these initiatives in the US. What is it? what are the numbers here? They started with one in Ohio, then for their big Louisiana facility, said they're gonna do three and then seven more. So

I guess if I'm meta, I'm saying it's harder and harder for me to to claim that I'm aligned with this group. I assume the mutually agreed upon sort of breakup was also part of the the group going, hey guys, it's it's hard to have you as part of our group if you're doing this. I I don't know, is this do we see companies sort of pull is is AI going to force them to pull back on a lot of the the climate commitments?

Tim Hatt  (36:01)

So I think I think two points here. I think one, you you hit it. The optics just aren't great when you are building gigawatt data centers across the US, and yet in the same time being a leading member of an organization that is that is trying to move to a renewables world. So there's a sort of point around image management and how that goes. But I think the longer term question is yeah, how if you're a hyperscaler or frankly a mobile operator,

Are in some cases getting into data center builds or co-financing or co-lows. how do you manage the energy phasings for that? In the long term, which is 2050, yes, I think most companies do want to still hit net zero and they're committed to that. How you get there has changed because of AI. And I think we're now seeing probably more reliance on fossils and diesel in the short term to

facilitate the build and then pushing out some of the renewables uptake that needs time for the supply to come online. So it's not like getting rid of a target, it's more accepting that the phasing to get to that target is a little bit different. Does that make sense?

Peter Jarich  (37:18)

And in the near term, it's a lot harder to be part of a group. I mean, if I'm if if if if if I'm a group and I've got folks who are like, look, you're claiming that you're focused on this, but at least in the near term you're building out these other things. Yeah, it's it's a lot a lot harder. I guess I guess for me, the other thing I see this linking into, which will be fun to see how it plays out to your point near term, right, is that we know from a data center perspective and AI perspective, we're seeing sort of a lot of, you know, NIMBYism.

People less and less want these things where they are. we see younger people who just I mean, I'm sure you've seen it, everyone's probably seen it, right? Billboards and everything saying we hate AI, right? The sort of pushback against AI. I was just just talking about this with a customer earlier today. We see from our own consumer research that when you ask them where they want a phone, AI is down.

You know, it it's it's well below better battery, better screen, low costs. Link that to the fact that this is or could be linked to climate change. and I think it'll be it'll be interesting to see how this all adds up, right? And how this adds up in terms of brand perception. It is obviously an opportunity for telcos. We know telcos have been committed to net zero and committed to to being sustainable more than most industries.

Feels like it's something to double down on as other folks kind of pull away from this. cool. Any last thoughts on that?

Tim Hatt  (38:52)

Well, I just think as a closing thought on that, your point about doubling down is right. I do think the political climate has been kind of difficult to manage from an ESG level the last two or three years, but I do think that's changing. And what you may not see a lot of PR on this from operators, but the commitments to reduce, you know, energy usage and cost are are are not changed. And and it and it is for that reason because it is a big cost line. So we're gonna see more.

we're gonna see more on this. The phasing is is sort of

Peter Jarich  (39:26)

Yeah, the sustainability side of things would probably you would suggest would be one of the winners out of an

Tim Hatt  (39:31)

Yeah.

Peter Jarich  (39:32)

ESG focus, if only because maybe if some of the people stuff is a little bit more difficult depending upon the market that you're in, and and views on DEI and views on the people side of things. but if sustainability is gonna help keep cheaper hands, keep your phones in the hands of people for longer, cool.

And if it helps to sort of push back against these initiatives, it feels like it's the feels like sustainability should be the winner, particularly with everything that's going on. Let's just fingers crossed. Well, on that, thank you, Tim. It's good to have you back. Thanks, everyone. I've said this before, I'll say it again. Leave us comments, leave us reviews, tell us what we should be doing better.

and as a piece of good news, I will say the the the long search for a departing tagline. I think we may have come up with one, so let us know. And with that I will just say thank you all very much and I would remind you to seek out intelligence, but embrace the noise. Thank you all very much, and we'll see you next week.

Tim Hatt  (40:44)

Thanks guys.