Episode 8: AI RAN, 5G Sensing and Smartphone Costs: The Next Shift in Telecom Innovation

Peter Jarich (00:02)

Hi everyone and welcome to Intelligence and Noise, the GSMA Intelligence podcast, where we try to take a look at what's going on and tell you why it's important and maybe stuff you're not paying attention to. Try and sort through some of the noise there, add a little bit of our own intelligence, hopefully point you to some of our research along the way. For a second week in a row, I am joined by Pau Castells. Hey Pau!

Pau Castells (00:32)

Hello, hi, hi Peter.

Peter Jarich (00:34)

Pau is here while Tim is trying to figure out how to get back from Belize. We haven't heard from him. I really assume that that means he's not he's not fleeing fleeing justice, but I guess we will find out. I am Peter Jarich, I'm the head of GSMA Intelligence. I should have mentioned that to you before. for those of you who don't know what we do here, one, I don't know how you found us, so congratulations for for making it here.

but but the basic idea is we we take the GSMA CEO newsletter, and if you're on that list, congratulations, it means you're important. If not, don't worry, we got you covered. We take a look at some of the news that was in that CEO newsletter, tell you why it was important, add a little bit of context to it, maybe our own spin, but then also talk about some of the things that maybe should have been on it but we weren't. So that's what we'll be doing over the next half an hour or so with

with maybe a few extra things sprinkled in. We'll probably talk a little bit of sports at the end because Pau's Pau's beaming inside. y y you you can you can't you can't you can't tell it, but but we'll we'll we'll we'll cover that off. I always begin with a little bit of context, a story. I'm gonna call it a a bit of of a self-serving Peter gets to to to tell a story time.

That hopefully sets the context for some other things we're gonna talk about. So last year, and and I usually do this or get in this mindset every time I get ready to take my my summer holiday in August, I I decided I'm gonna get a new phone. I'm gonna try something out. I'm gonna try and I'm gonna try and go for for mid low mid tiers. So if I hurt it and don't put a case on it, it's going to be fine. and so I I did that, got myself a

Perfectly serviceable Motorola Edge sixty pro. And and it's worked out well for for a year and and as I said, I would try it out without a case. And if you try it out without case, you're doing that because you know you're not gonna be losing too much if you risk it when it breaks and and of course it broke, fell out of my pocket, fell onto concrete ground. Nice, nice. I'm now one of those people you see and go, come on, dude.

Like

like really that's a that's a horrible screen. What are you doing? And how have you not hurt your fingers? So I thought, okay, time time to do exactly what I thought and look for a new phone. and right on cue the good folks at Motorola have launched their their Edge seventy Max. And what was interesting, I like, Cool, you know, it also looks mid tier, I can I can can can, you know, make that easy shift and then you start looking at things, right?

a year ago the the 60 Pro had three cameras, now the seventy Max has two. it's got a new faster SOC, it's got a it's got a cool Qualcomm chip, not quite the newest, but but serviceable. No e sim, bigger battery, supposedly brighter, but no sixteen gig of of RAM option, worse selfie camera, but it's got sort of the the the wireless charging with the magnets. And what I realized is that you you've got this really

Y you've got this interesting set of trade-offs here, right? That okay, maybe the price is the same to a little bit higher, and how do we keep the same price same and higher despite the fact that memory is has gone through the roof? Well, maybe we limit some of the higher end memory options. We we we take away one of the cameras, we downgrade one of the cameras, we maybe distract people by throwing in a a bigger battery and throwing in some some you know the the magnets in the back. And you know when I say distract I'm not I'm not

denigree and the good folks who who put these folks together. But these are the decisions that need to get made because as we know, smartphone prices are going to be getting higher and higher and higher because memory has gone through the roof. And that, my friends, leads us into the CEO newsletter. So Pau, I think I think we're gonna talk about both the stories that made it in because they're both important, but I think the one

The one leadoff one was around device costs. and and it's a bit of a it's a bit of a fun one, right? 'Cause w it kinda get to talk about all of our friends and other research firms and what they were doing. So it was it Counterpoint calling out the eleven point drop in in phone shipments, lowest shipments for the quarter since twenty thirteen due to memory costs. this is not news, right?

Pau Castells (05:07)

Yeah, no, I mean I think that that that actually and and your your Peter story intro, I think it sort of sets sets the scene pretty pretty nicely because this this actually is pretty interesting story that continues to to build on on a trend, which is that AI is making smartphones more expensive, right? And it's you know not a not the most obvious connection to begin with, but but you know.

I think know you you all know the story here. AI has put a lot of pressure on on memory, on on chips, and that's you know crucial component of smartphones and this sort of classic supply side shocking economics. know, I am I am an economist, so here is my my economics talk. this is actually resulting in in higher prices for for smartphones, which is resulting in or it's contributing to this.

slowdown in sales, right? So that's that's that's what Counterpoint Research was was calling out in in their news.

Peter Jarich (06:14)

And what I think is what

I think is interesting, just just just a quick jump in there, is I think I mean so that was from Counterpoint, but you know, if I look at also from you know some of the other research that's out there, you you see this just the amount of you know the amount of a phone and the bill of materials that is

down to memory is just, you know, it is just doubled in some cases. But what's particularly interesting is it's it's taken a big hit on, let's say, the the lower end of the market. Right? We're at the higher end of the market where you've got those multiple cameras, right? And maybe you've got a particularly impressive screen or everything else, there's a little bit of a room for sort of billings of that in. But we're at the low end where it is more stripped back,

the the the impact of that rise on on memory side of things is is just is just is just crazy, which I think for some of our efforts, right, I know you work with the the folks that we all do on on the device accessibility side of things has to be just a big hit.

Pau Castells (07:21)

Yeah, no, absolutely. And you know, it it it it does have, as you as you are explaining, Peter, it it does have actually pretty significant implications for the for the lower end of the market, which then gets into questions around digital inclusion, right? So is is this grading or making the challenge to connect the unconnected more difficult? And and clearly the answer has to be a resounding yes. I mean.

We we do have as a proportion of income for the lower income populations, smartphone costs are increasing a lot more than they are increasing for the for the higher end of the population, right? So this, you know, being

Peter Jarich (08:02)

Yeah.

Pau Castells (08:02)

one of the main barriers for for connectivity, and we do in our own GSMA Intelligence surveys generally find that, right? That handset affordability is one of the main barriers to connectivity for those that do not have yet a connect a connection to the internet. It

It clearly becomes a a challenge. I mean, I think one one one aspect that is probably and here I'm talking or playing a bit more, you know, cr crystal ball futuristic type thing thoughts, right? So we have clearly AI putting lots of pressure on on memory chips and that's pushing up device costs.

We also have and we saw news from Ericsson last week that they they are seeing AI impacting also the costs of equipment. And they semi announced a potential increase in prices for for the equipment they are selling. So again, if vendors are passing through these costs and pass them on to operators, then potentially the connectivity itself might become the

The service itself might become also more expensive, right? So we have this potential change in in the in in paradigm, right? Where you know we've

Peter Jarich (09:16)

that's a that's a f that's that's fair. It's a it's a yeah,

I mean it it's a really good it's a good reminder, right? Because the the the handset side of things is I mean it's in the news because it touches everyone. Right? Because we can all see that again when you you break your phone like me or or you're just looking to get a new phone, we can all see that. But it's like a reminder, and I think to your point it's not a bit of crystal ball grace as much as just

pulling out the the the the reality that memories in everything, right? Memories it's it's not just in phones, it is in everything. So it it should be impacting all different parts of the of of the digital ecosystem. So yeah, I mean I mean when do you think we'll start w

When do you think we start to see how that flows through, right? If it's impacting network costs, if that has an impact on on the actual service cost down the road. Is that a is that a next year thing two years from now?

Pau Castells (10:13)

Yeah, well let me

let me just consult with my with my crystal ball and I'll I'll get get back to you after that. no but I I I think it's it's sort of you know the we have lived for the last two decades in a world where we have almost taken for granted that connectivity costs are coming down, right? We we this has been you know the deflationary sector. You know, connectivity costs have gone down.

Devices have also come down if you compare them on a sort of apples with apples, right? So or if they have not come down in price, they have incre increased their quality. So this has been a trend and a very, very, very strong trend. what we might be now seeing is potentially a stop to that, right? We might see, especially you know, if if AI continues to push through. We might actually see for the first time, actually connectivity prices might not be coming down as quickly as they have.

been coming down before and same for devices. So yeah I'm not going to put a a timeline to that. And I'm I'm I'm hoping that you know this can be reversed because obviously you know the the supply side on on the memory chip side of things will also react to that and hopefully increase capacity, etc. But yeah I mean that's that that's one of the potential unintended effects of of the AI boom that it's making connectivity more expensive.

Peter Jarich (11:40)

Yeah, it really is. Like a and it's interesting because to your point about sort of the AI of this all, right? So you know if you look at our consumer survey and you look at our our latest data that we've got, the interesting part is, you know, when you ask folks about what they want in their devices, you know, AI capabilities is is near the bottom, right? They they want strong battery, they want a durable phone, you know, lower cost, camera quality towards the top, right? So so it does go to that

initial

trade-off discussion that you know I was sort of talking about there of of if I'm a maker how I think about this. But then if you realize that a lot of these costs are the impact of AI being driven somewhere else, knowing how people feel about that.

Right. that across societies there's growing questions about how people are are looking at AI. I've I've seen no shortage of ads all over to say, hey, people hate AI or we hate AI or we don't AI, so you're seeing this backlash. It'll be fun to see what happens if something that people seem to be more and more disposed to dislike ends up causing them causing them higher prices elsewhere. I guess it's one more reason to

One more reason to to dislike it. but

It'll all be okay, right? Right, Pau? Because if we can use AI to actually make our networks more efficient and and be better with networking, then maybe it can keep costs in check. And that's my ever so not subtle connection to the second story that that graced people's inbox in the CEO newsletter, which was which was a focus on AI ran. And this this was led in part by some work, the smaller piece of the news being being

and a bunch of of you know different launches there taking place actually with Ericsson Nokia Samsung, KT S K T to launch some some trials of of AI RAN, but probably the bigger news being Nokia's launch of an AI RAN solution. Again, we know that they've been a big supporter of AI RAN, so not too surprising there, but sort of work with with NVIDIA Silicon their own software, three different models, either a capacity plugin for their existing solutions or a standalone

Node or Total Cloud Native. Fair to say the big tagline there, hundred percent, hundred percent increase in spectral efficiency by twenty twenty-eight. I mean if that happens, right, then then then telecos gonna be saving money in the end, right?

Pau Castells (14:15)

Well, I mean I think obviously and there is there is the you know hope that that this actually materializes, but it needs to be tested in in commercial networks. And I'm sure that many operators have have heard all this before and will probably approach this with a with a degree of of caution. But yeah, with with with that that caveat or or remark aside, I mean this is this is actually, you know.

pretty pretty impressive numbers that we're talking about here, if you know not just that doubling of spectral efficiencies, but you even if just a fraction of that materializes, that's that's you know really impressive. And it's you know when when we have moved from say 3G to 4G and 4G to 5G, we have seen improvements in spectral efficiency, but they have been or they have been becoming a bit more challenging to to obtain.

obviously Mine has helped a lot recently, but but still, I mean, yeah, it is actually quite quite a quite an impressive figure.

Peter Jarich (15:22)

Yeah, I think I think, you know, the interesting part for me, I think if if I look at this and I look at these stories together, two pieces. One, we're gonna be using this term AI ran a lot.

Going to be out there, but what it means is is different for everyone, right? I mean, so so obviously again in the Nokia case, you're talking about running their software on essentially AI silicon as well as some as well as some silicon solutions that they're building third parties with with Marvell, but that's different than like what you see in Korea, which is more of just hey, here's a bunch of use cases of physical AI. It seemed to be focused more

on the use case side of things. I I know that Ericsson in the past have talked about, hey, we're using AI in the RAN, we're building AI into our radios that can do things in silicon and you don't need GPUs. I think we're going to be in for you know the the classic I shouldn't say I think. I think we all know we're going to be in for this next round of people tossing this word, this phrase around AI RAN.

And that's I think the one takeaway. The second takeaway is people mean vendors throwing it around, with an exception of of of some some clear seems like seems like some folks who aren't, which is is the Chinese, right? We're we're not seeing this from Huawei, we're not seeing this from ZTE. Again, it's it's not surprising that in South Korea it was Samsung, Nokia and Ericsson. but

You know, the fact that we aren't hearing a lot about this fr from the Chinese either says we're going to see something soon, you know, they're preparing something, or it's just reinforcing the reinforcing the idea that there's been these chip you know embargoes. There's been a lack of access to a lot of the silicon that has been required for some of this, and as a result we're not seeing the same focus. So

This is the one to watch, I think, coming up is is how the Chinese respond to the AI-ran side of things. And really, if as an industry, we can come to some sort of definition of what we mean. And we still haven't done that with virtual RAN versus Cloud RAN and all that stuff, so I'm not particularly optimistic that we're gonna have one AI-ran definition, which means it's either good for marketers or or bad for everyone.

Pau Castells (17:44)

Mm-hmm. Yeah, I mean an actually in a way the AI for RAN developments, they they are not just one thing, right? So they involve, you know, quite a quite a lot of, you know, small improvements or small gains for the I guess, you know, the the cyclist fans there. That that was a a theory that that that took took up a few years ago.

the the marginal gains theory, right? That you know, if you improve these and these and this and these and these very, very, very small things, then actually you start if you add them up, you start having, you know, pretty significant improvements, right? And that's that's and that's at least my understanding of of what many of these AI RAN improvements actually play out, right? They they're just kind of making little things more efficient, but then once you add them up, you know, you get to these big numbers.

Peter Jarich (18:34)

Yeah, and if we're talking about hundred percent gains in in by twenty twenty eight, right, that gives you time to add some of those things up. I I would remind everyone that Ineos is no longer the the sole title sponsor of their cycling team, so those marginal so so the story of those those marginal gains that they were they were so fond of driving,

Yeah,

we we I'm sure it has nothing to do with that. the story that didn't get picked that we want to talk about so we can keep moving, is

The one that actually I want to talk about, and I think this was the more interesting one, and in part it's because I also tend to focus a lot on this, was a story around ISAC. you talked a bit about their sort of the the Ericsson of it all with what they were talking about in terms of the memory side of things. This was some news from Ericsson alongside ATT, where they talked about doing a demo of ISAC over 5G. I appreciate that.

They called out that this was done at ATT Stadium outside Dallas during a major international sporting event, which will remain nameless. I guess I just do do you have to pay royalties? Like do you have to do you have to put an enormous amount of money just if you use the the name of a FIFA and World Cup in your announcement? Maybe I I don't know. But that that's my assumption, at least, given that the announcement was in July.

Lie. And it was all about drones, about drone detection, tracking drones, and really the big news there is that we were doing it with with 5G. We we talk a lot about ISAC as a 6G technology that seems to be a big focus of it, one of the big use cases for 6G, but the story here was right, hey, we're we're doing this right now. I've got my own thoughts on on why I think this was particularly important, but your your your view.

Yeah.

Pau Castells (20:41)

yeah, no, so so I mean the the fact that that that it was actually already you know tested with the existing technology that that that was a bit of a surprise to me and I'd be keen to hear your thoughts on that, Peter, which I'm I'm sure you do have. that that aside, I mean I guess what what makes this even it's you know with 5G advanced or with six G more more probably we will

start seeing some of these technologies being available. I mean, there is the potential, I guess, you know, advantages that you know having or being able to reuse an existing network to actually develop sensing technologies, right? That that actually seems like, and these announcement here with 5G already, you know, further tugs into that, right? You know, we we're talking about you know just finding or trying to develop new applications and technologies that that can have a commercial viability for

Telco. So I think that that that's interesting from a I guess more business perspective. the question as to whether there are plenty of b business opportunities out there or demand for this sort of sensing use cases, it's TVD, right? So I mean we can think all of of security applications, maybe transport as well. but how much is are are people actually

prepared to pay and even if they're prepared to pay, is that going to be sort of enough for, you know, the quite likely big investments that telcos need to put into in place, right? For for for this to to work. So yeah, I mean lots of lots of questions, I'm afraid. but I'm I'm sure you've got some answers.

Peter Jarich (22:22)

Yeah, no, I think I think the drones of it all

I think no, I think I think as you said, the drones of it all is you know, undeniable. So I think clearly we can get the idea there's there's a war going on where drones are the the primary

Battle vehicle that's taking place against that backdrop with continuing worries around terrorism, especially around major events, it makes sense, right? But can we go beyond drones? You're right, that's number one question. I think for me, the interesting part here is two things. One, and this also took place at the same time as Ericsson got listed as a supplier to the UK Tactical Commons Systems Framework for public sector procurement. So this does point to this sort of overlap between

Telecoms, public sector, telecoms, military and defense spending, how we actually see those networks. And if we look at it, we talk to operators and we ask them what, and this is from our operators' survey, where we ask them about their B2B services, and we ask them what's the number one vertical that you're selling into. It is public sector, right? So this is a space they know. Defense is the other one. I think this is just another highlight, right? Ericsson, ATT getting there saying, hey, here's a cool

Use case where yeah, drone detection clearly a clearly a focus for for defense, clearly the idea that if you're ATT and can say exactly to your point, right? If you can leverage the network, I think they said the announcement, hey, if you can expand this across your whole 5G footprint, pretty cool, pretty quick what you can do there. And the quick part, that's the last part for me, which I think was interesting and why this was a bit of a important piece of news where it's maybe not as clear, is that if you look back a few weeks.

Weeks ago. There was an announcement from Cohere Technologies with the US Department of Defense's Future G program. That's that's the Tom Rondeau program looking to drive new technologies, leverage commercial technologies in support of mission first agendas and priorities. And they they announced this deal with Cohere to help develop some ISEC capabilities. And it's very clear if you look at the announcement, the announcement says, Look, we need something that that works not.

Way into the future, but but something that we can start taking advantage of sooner

Pau Castells (24:42)

Mm-hmm.

Peter Jarich (24:43)

rather than later. It was all about giving money to an innovator to help develop solutions that they can leverage on existing networks with their multi-g technology, as they call it. And I remembered when we talked about this when the news came out a few weeks ago in the podcast, and and I said I

I get, right? This makes sense, they're trying to invest, but is this also a bit of a message to everyone else to say, hey guys, we want this, we want this soon. Don't sort of rest on your laurels, think about developing something. It's hard not to see this from an ATT and an Ericsson to go, yeah, a couple of weeks ago you guys said you needed something very quick and you spent twenty-eight million to help develop that. here you go. We got it. Here's a demo for you. Now

I think a couple weeks apart from each other and and it's hard not to see that connection. What I don't think we know yet is, you know, what is six G ISEC gonna deliver then that five G ISEC can't? I hope it's you know, we're gonna need to see more explanation of that. What is a technology like Cohere Developing going to do better than what you can do on an existing 5G network? We're gonna need to hear that. But I think it all comes back to your point, which is beyond drones.

Beyond

tracking drones, which maybe you can also use cameras for and radar and other things, what are the use cases that are gonna drive this? Because it can't just be it can't just be that single drone use case and we're gonna need to see more of what that is. So that's my that's why I think this is particularly interesting, because there's so many different pieces there.

Pau Castells (26:13)

Yeah. Absolutely.

Peter Jarich (26:16)

What else? On the road. We're now at that last part for for On the Road and On Our Minds. Neither of us are on the road anytime soon because summer holidays are coming up and much of Europe shuts down. The World Cup is over. congratulations, I would say.

I would be careful because you d the the a FIFA bylaw is not well known. There's a week for the host country to to decide if they want to overturn the results. So, so so I would I would

Pau Castells (26:46)

Ha ha ha.

Peter Jarich (26:49)

warn I would warn all of your your compatriots to not start celebrating. I will say that if you saw the halftime show,

Shakira looked in in I I trust Shakira's future much more than Madonna's. Madonna, I don't think could even walk. She was on. I mean, sh I think she's famously said she no longer has knees or ankles or something for dancing in high heel. Shakira was out there without shoes. She's protecting her future. so so good for your fellow Barcelonan. fair to say it was actually last last piece for me. You have also cemented

Soccer, football as something that Americans will not like because this was the classic it went a hundred and how many like I I there put probably tons of people around that wait, I had to sit for two hours and and one person scored one person scored once over the course of two hours. And we're all we're all excited for that. That is a it is the antithesis of what Americans like as a but congratulations. It was a well yeah, well well well thought.

Pau Castells (27:52)

Yeah, no. and yeah, it's it it's been it's been and it continues to be a complicated road to to make football mainstream American sport. but but I do I do have the feeling that this time around, you know, there have been actually some some further successes there. I mean I think seeing seeing your president actually already deciding that the next World Cup will not take take place in Spain, Portugal.

And Morocco as it's planned, but it it might actually take place again in the US because the US can do it better than anyone else. That that that was actually pretty telling sign for for me.

Peter Jarich (28:32)

I hope I mean I I hope Morocco hasn't gotten too far with its high speed rail between high speed rail connecting all the big cities. So that will be that that's that's like I said, you gotta give it at least a week to see if it remains remains in Spain. Last last piece of news for me, and I will bet you this is something that w this is not an AI not an AI podcast, but I think

like so much of everything that everyone does nowadays, everything is is AI related. And so I think the last piece of news for me is sort of on my mind versus versus on the road is something that and I'm guessing, Pau, this is something that that you're not even not even on your radar and it probably shouldn't be, but new Chinese

AI new Chinese model, Kimi K3. and I will say I'll be honest, when I first heard of Kimi K three, it sounded like a Love Island character or or something similar of a reality TV show contestant, but it's technically I should say technically it it's essentially, let's say, eighteen months after the Deep Sea Gar One model came up and and we all

Started to freak out and the world was like, whoa, whoa, whoa, whoa, whoa, look what's going on in China and and at a fraction of the cost. Feels like we're we're going through that again with a new model from a Chinese company called Moonshot. open source not necessarily much lower cost, at least as I understand it, but performance is up there with some of the

frontier models at least per approval who check on these things. And the interesting thing is, yeah, so it's it's 18 months after we we've gone through this whole thing with Deep Seek. Obviously in the last 18 months, trade wars have continued, geopolitics have not gotten any less any less heated. So it does put that sort of you know it's a reminder of how this things how these things progress. For me, I think the interesting part is, and this is something that I don't really

know

that there's a good I don't know there's a good answer to but it's probably something we're gonna need to start looking into and I've heard this from some of some of our our operator customers is how much do you actually need particularly from a telco perspective to do what we want to do? So let's go back to you know we're talking about the AI RAN side of things and how you may be able to improve some network performance with these these technologies

Do they need to be continuing to evolve at an exponential level month after month after month? And do I need the latest and greatest and fastest and best or at some point is good enough good enough? And I think that's for me, I think an interesting question there of as you've got the open AIs and you've got the anthropics and you've got Googles and you've got everyone else there who's competing and rolling out new models that are better than the last week after week after week.

at at what point. It it's interesting because it's it's not like it's not like when we talk about sort of quantum. We're on the quantum side of things, we go, well we're here but we need to get to here. Right? You know, we're at a point where we're we have some basic capabilities, but we know to really do what we want to do. We've got to get here. With with AI, it seems like we're we're doing really cool things right now with these models. And next week we're gonna have a new model that's gonna even be a lot better and the week after that, but for

i for me it puts into perspective and again I've had these conversations with some of our our operator members going d how much do we need to get caught up in that? I don't know. How would we is there a way to evaluate that from your perspective? Are we just gonna get caught up in this? Or is there a way to sort of break out of that from from our perspective and go, this is what we need and and enough is enough?

Pau Castells (32:24)

Well it it's it's really down. I mean it's a it's a very almost philosophical question, right? as in and and we do get asked the same question as well when it comes to the quality of telecom networks. So it's not just an AI question as in how good AI models need to be. we have you know similar discussions, say in the European context, for example, around well actually but you know, we have good quality networks already. Yes, they're not as good as they are in other parts of the world, but

Hey, you know, it's not it's not so bad, right? So I think it's a question of what do you want these technologies to to support and unlock. And you know, many times it's it's not clear, right? Because it's we're we're we're we're not talking about supporting or the the underlying infrastructure and capabilities that support innovation in things that we cannot define.

And if you take a sort of a too short term ish look at things, say, well actually, you know, that that's you know what we have is good enough. but that might be more a question of you know one being unimaginative rather than than than not you know thinking or or or not thinking through sufficiently through into into the future. So I know I'm not giving you a straight answer, but it is

Peter Jarich (33:45)

No, no, no. It's a it's it's it's that

there probably is no there probably is no good answer to it. It's just it's interesting to see if I'm in fintech or I'm in something where having the greatest model is going to give me a competitive advantage because

you know, it's marginally better, but I can translate those marginal improvements into a a real bottom line result, I get it. I think if I am an operator where I am looking at AI as a tool to help drive efficiencies or roll out services more efficiently, it's a little bit of a different model and I don't think we're I don't think anyone in the industry, like I don't think I don't think anyone's quite figured out where that is, but the getting caught up in in all of this. And again, to the point of we haven't figured this out, if you haven't seen it yet

Yet

we just released last week our latest piece of research on sort of the AI token token economy, tokenomics, call it what you want, about some of the the barriers and obstacles to think about and the fact that you know we're still working through some of these basic use cases, there is no good answer, I don't think. So it's a we just need to keep asking the questions.

well, on that note, we've got no answers for you. will will Spain continue to be the World Cup champion next week? Will we be back in the US in

Four more years for for World Cup twenty thirty. will any of these AI models really really be necessary? we don't have answers, but tune in to to watch us discuss more of them next week. thank you, Pau. Appreciate all of your your input on this, and I'm sure we'll have you back again. Thanks to everyone for joining us, and tune in again next week.