A Raw MWC 2026 Recap (with Earl Lum and Marc Pous)

It’s become an annual neXt Curve tradition. A brutally honest recap and reflections on Mobile World Congress with Earl Lum of EJL Wireless Research, and Marc Pous of Edge Impulse, over amazing food, amazing wine, at a hidden, lesser-known culinary gems in the beautiful city of Barcelona - Oido.

This year, the GenAI bulldozer continues to plow through the halls of Fira Gran Via affecting every layer of the mobile industry and ecosystem. Leonard, Earl and Marc test the hype and realities of AI-RAN, network APIs, sovereign telco neocloud, and oldie but goodies, 5G monetization, NTN, and the path to 6G.

In this episode, Leonard, Earl, and Marc talk about some of the top topics of Mobile World Congress 2026. 

➡️ Humanoid robots were the biggest distraction of MWC 2026.
➡️ The evolution of "sovereign" infused with defense and national security. 
➡️ Is the mobile industry drifting rudderless?
➡️ Will MWC become a physical AI conference?
➡️ The killer AI for the telco/mobile industry!
➡️ rApps and SMO have an agentic heyday!
➡️ 3-years later: The flop that was Copilot for telco.
➡️ The value of simple and boring AI.
➡️ What is 6G? Has anyone figured it out?
➡️ The physical AI for the mobile/telco industry unveiled!
➡️ Uplink steals the 6G and 5G Advanced show at MWC 2026.
➡️ Why is 5G monetization still a mystery?
➡️ What was hot in IoT at MWC 2026?
➡️ The state of telco neocloud. Is it just another Metaverse?
➡️ The brutal reality of connecting the middle of nowhere and weapons guidance.
➡️ Is the data center in space the new "Pigs in Space"?
➡️ What do we expect will be the key themes and topics of MWC 2027?

Hit Leonard, Marc, and Earl up on LinkedIn and take part in their industry and tech insights.

Check out Earl and his research at EJL Wireless Research at www.ejlwireless.com.
Check out Marc at IoT Stars at www.iotstars.com.

Please subscribe to our podcast which will be featured on the neXt Curve YouTube Channel. Check out the audio version on BuzzSprout or find us on your favorite Podcast platform.

Also, subscribe to the neXt Curve research portal at www.next-curve.com and our Substack (https://substack.com/@nextcurve) for the tech and industry insights that matter.

NOTE: The transcript is AI-generated and will contain errors.

DISCLAIMER: This podcast is for informational purposes only.

Transcript

Leonard Lee 0:00

Hey everyone, this is Leonard Lee, executive Analyst at Next Curve, and welcome to this very special episode of the Rethink Podcast. Live here from Barcelona, Alona, Spain. And guess why we are here. We are here because of MWC 2026 and also IOT stars, even though IOT Stars made a very small presence this year. It was a great event. And I'm here with the IO OT Giant marked Post, post, post, I got it right this time post. Yeah. Yeah. Who is with, edge Impulse, which was acquired by Qualcomm,

Marc Pous 0:44

actually works for Qualcomm a year ago. Yeah, one year ago

Leonard Lee 0:47

actually. Yeah. And then of course. For Lum, the legendary lum of EGL wireless. And, what we're gonna do here is we're gonna recap. MWC 2026, I think, over, glass of vino, many glasses of vino, many glasses of vino. So, you know, if you're really interested in getting the real take, the unfiltered, brutal take from what, happened at the Fi grand. Is it the Fi grand or what do they call it? The fi

Marc Pous 1:19

Yeah, fi. Grand. Grand.

Leonard Lee 1:22

Grand. this, you're at, you're at ldo, right? Podcast. You're watching the White

Marc Pous 1:29

Well and actually

Leonard Lee 1:29

podcast.

Marc Pous 1:30

And actually I can talk as well about the talent arena, which was, the second edition this year that happened in the old Thera in classes, Espan,

Leonard Lee 1:37

Uhhuh.

Marc Pous 1:38

And there were almost about 50,000 students and developers going, there. To attend technical talks, workshops. yeah, it was super amazing. High quality content. Oh yeah. These type of communities or target audience that usually don't go to mobile wall Congress.

Leonard Lee 1:55

Yeah.

Marc Pous 1:56

And, they are trying as well to acquire that, type of audience as well. It was super, super interesting.

Leonard Lee 2:02

Wow. That's great.

Marc Pous 2:04

Yeah.

Leonard Lee 2:04

So you actually made it in

Marc Pous 2:06

Yeah, it was, three days there and so I gave really, four workshops. So there was, works for 100 people,

Leonard Lee 2:14

Uhhuh,

Marc Pous 2:14

so there were even people standing in the, sitting in the floor,

Leonard Lee 2:17

in the aisle and stuff.

Marc Pous 2:18

So it was, very well attended and, everyone following the one hour workshops and it was super, super interesting us as well.

Leonard Lee 2:26

Okay, great. So I want to clarify something for our audience. So the camera is playing tricks on you. He is not the smallest person in the room. No. He's freaking gigantic.

Earl Lum 2:42

Yes. And you see here actually pictures every year that we take and post. I will be up to maybe his waist, if I'm lucky.

Leonard Lee 2:51

Yeah. So, no, it's really wonderful to hear. And, one of the things, so it's great that they had actually had real, programs

Marc Pous 3:02

Yeah.

Leonard Lee 3:02

For some of these, graduates or, like college students who are interested in the industry. But one of the things I thought was maybe a disservice, and I posted this on LinkedIn, is all these like humanoid robots and stuff like that, right. And, it was bad enough that there were dogs, right? That were stealing the attention of everyone, making them think that, oh, if I get into Telco or if I get into iot. I'm going to be involved with building these dogs or humanoid robots, but telco has very little to do with any of that stuff, right?

Earl Lum 3:41

No, it's the connectivity of those dogs and those humanoid robots and everything

Leonard Lee 3:46

else.

Marc Pous 3:46

all of No, 5G connectivity was a thing. Pretty much two years ago, right?

Leonard Lee 3:50

Yeah. But, I, I think the connectivity become, becomes relevant for these types of devices, when you're talking about defense critical communications. Right.

Marc Pous 4:03

But don't you think robots, it's, one of the examples of h ai that means that the intelligence it's running into the robot itself takes decisions know,

Leonard Lee 4:12

But then a lot of times they're autonomous unless you're just doing orchestration or, trying to, maybe foster swarm intelligence. Of some kind.

Earl Lum 4:23

Well, there's different kind of robots, right? There are robots that deliver food in a restaurant. Here there are robots that are manufacturing robots for private networks, and then there are the dogs from Boston Dynamics or whatever that you see are scary dogs.

Leonard Lee 4:38

Yeah. Samsung.

Earl Lum 4:39

Yeah. And so whatever kind of robots you're talking about, they're each different. Some of'em, like the AGVs in the manufacturing floor, all they're doing is delivering things from point A to B or or whatever. Right? So that should be simplistic. Yeah. although you still need to have some amount of intelligence and to the full scale of let's just swarm a bunch of other things running around, out there from a military perspective is gonna need a lot more intelligence and autonomy.

Leonard Lee 5:05

Well, defense was one of the big, topics actually, and it was folded in, or at least it was a bit adjacent to the growing interest in sovereign X, Y, Z. Right. I think this year nobody would sovereign really cool buzzword. Not quite sure or how to make that happen or treat that with, mobile infrastructure or telco infrastructure or IOT infrastructure. But, especially with, what's going on right now in the Middle East. Yeah. which happened as we were flying in.

Earl Lum 5:46

Yes, thankfully. Thank you. We weren't going over that direction where a lot of people couldn't make it to MWC because of the disruptions. Right.

Leonard Lee 5:54

Yeah. And, but I think that was like a bit of a curve ball, right? Because I think the topic of sovereignty, which was really a big topic this year, it was a little bit more related to sovereignty around ai, right?

Earl Lum 6:12

Yeah.

Leonard Lee 6:12

Telco sovereignty, mobile wireless sovereignty. It's a different beast because 5G has. I really fostered this global regime around wireless technologies, standardization of technologies, standardization and commonality globally for a lot of things in the, that support this global ecosystem. Now, the introduction of Sovereign has been, I think, bit of a weird thing that actually threatens a lot of stuff that, has been achieved through, this 5G

Marc Pous 6:53

era. I have a stupid question about sovereignty.

Earl Lum 6:56

There's no stupid Is,

Marc Pous 6:58

is, is this not another word for saying this is an on-premise solution?

Earl Lum 7:04

It is in country somewhere as a solution. Okay. As opposed to distributed across many other countries. Right. For your storage of data. So if you are in France. Then whatever, customers that are in France, their data could have been previously stored all over the place. Okay. Around the world in data centers right now, the question is obviously for very sensitive data. As long as it's in France somewhere

Marc Pous 7:33

that the data cannot

Earl Lum 7:34

keep. But if it's a health situation where health data, I think, I don't even think they talk about sovereignty. That would be really more on-prem in the hospital just because of regulations and all of that.

Leonard Lee 7:44

Even then, that is still a sovereignty thing because it's, you want your citizens to be protected, guys. You want, the local privacy regulations, whether it's for a vertical like healthcare or otherwise to be maintained Right. And right. That's why this topic of privacy, right? It is such a complex thing. And so anyone who's done, like back in the day, I used to do post-merger integration work for, the pharma industry. So when you saw a lot of these large. Pharmaceutical companies basically do these multi, basic multi transactions, sometimes exchanges, dealing with the data, making sure that you handled the data correctly, you transmitted the data correctly. The regulations around that, around the globe is literally a spaghetti mess. So if you're a systems integrator, you had to deal with so many policies and you were working with legal almost all the time. And back then these policies were really difficult to institute in the migration process. Right? Today, maybe it'll be a little bit easier, but that goes to show you how. Even back then where I think the sharing and the movement of data was a little, had less friction than what we have today. Now that friction is becoming really intense because of the sensitivities, right.

Marc Pous 9:26

yeah.

Leonard Lee 9:27

the fragmentation of the geopolitical landscape.

Earl Lum 9:29

No. Well, the extreme case of sovereignty is. I am trying to put a data center space. am I trying to skirt legalities of a specific country? Because I'm not domiciled physically in a country and in space. I have no legal.

Marc Pous 9:48

Oh, that's,

Earl Lum 9:50

so if I'm a company that doesn't want anyone to know what I'm doing and I put a data center up in space, how are they gonna know? Right, so that's the extreme version of sovereignty in that definition.

Leonard Lee 10:06

Why can't you just set up a data center in the middle of the ocean, international waters, in that

Earl Lum 10:12

you could probably call it, you know, pirate cloud, right?

Leonard Lee 10:16

yeah. Pirate Cloud.

Earl Lum 10:19

Pirate Cloud.

Leonard Lee 10:19

Is that a private cloud?

Earl Lum 10:20

It's pirate Cloud, like pirate radio. You're 20 miles off silver and you're in international waters. Theoretically, no one can touch it.

Leonard Lee 10:28

Yeah,

Earl Lum 10:29

so maybe you put a data center on the ocean floor or something.

Leonard Lee 10:33

So we're back. we, took a break because we wanted to order food. And by the way, this place, which is called, what's this

Earl Lum 10:40

place called together,

Leonard Lee 10:41

Odo. Odo, OI, is amazing. And it's like a hole in the wall in the middle of some weird neighborhood.

Marc Pous 10:53

it's Gia. Is the neighborhood of Gia the favorite for tourists?

Leonard Lee 10:56

Really? Yeah.

Marc Pous 10:56

Yeah, absolutely.

Leonard Lee 10:58

What's it called again?

Marc Pous 10:59

Gia. Gia.

Leonard Lee 11:01

Gia. Oh, okay. All right. But yeah, if you see this, give them business. Oh yeah. Yeah. Fantastic, fantastic. And then when last year we were at one of their sister restaurants called Fondo, fond Peppa in Fondo, right? Yeah. It's like one of the train

Marc Pous 11:19

stop right nearby. It's like 15 minutes walking from here in the same neighborhood in front of the Bitcoin. ATM.

Leonard Lee 11:24

Yes, I remember the Bitcoin a F. So anyways, let's go around where were your big tanks and impressions and insights from MWC 2026 this year?

Earl Lum 11:37

A lot of the companies that I talked to were drifting in the ocean of telecom, not knowing which direction to go because they were not selling anything in 5G. They didn't know what was six G, and they're just, there was little hope is what I saw from a lot of companies in terms of where they were going to be, which was way different than the prior generations of 4G, et cetera, where you really knew what was going on. But with China done, and most equipments, mostly Europe and everything else, if you're a hardware guy, what can you sell and where is now the issue and the problem?

Leonard Lee 12:22

Wow, that's a deep insight.

Earl Lum 12:25

So, and again, I came here with Zero Hope.

Marc Pous 12:28

So as I mentioned before, I was at the Italian arena with developers and students, and one of the takeaways is that all of them are super interested on ai,

Leonard Lee 12:38

AI

Marc Pous 12:38

from NJI to ai. There were robots everywhere and yeah, I think, uh. It's gonna be, that's gonna be fast.

Leonard Lee 12:50

Okay.

Marc Pous 12:50

Past adoption basically of these technologies. Right. All earth still. So, dunno if it's, scary or not. Still need to decide.

Leonard Lee 12:59

Well, what, does that have anything to do with MWC? Do you think m wcs then going to change as a conference and the attention's gonna move toward AI and physical ai? I mean, everyone's trying to create a physical AI story, right?

Marc Pous 13:15

It's weird. Telcos selling physical AI and I don't know, it's,

Leonard Lee 13:20

yeah,

Marc Pous 13:21

but I'm not sure how AI is applied, for example, on all the telcom know all of these. Last year we were talking about this AI run.

Leonard Lee 13:29

Yeah.

Marc Pous 13:30

I'm not sure how these evolved in the last year.

Leonard Lee 13:32

actually evolved pretty well, but not in the way that, Nvidia would like.

Marc Pous 13:36

Okay.

Leonard Lee 13:37

one of the things that I think is a real bright spot for ai, so the killer AI is not generative ai. It's not ai, it's called machine learning. Yes. Ml Hot. Hot. And like this is one of the things that I've mentioned on IOT Coffee Talk several times. One of the big benefits of generative ai, all this investment, all this. Hardware, accelerators, the Silicon Technologies interconnect lot is gonna benefit machine learning, right? It drives the cost down, makes it much more e economical, allows, machine learning in particular to propagate out toward the edge. That's why all of a sudden we're seeing all this excitement about Edge ai. Yeah. Right. So shout out to Pete Bernard, our brother. Yeah. And IT and Edge AI Foundation. But I mean, that, that is really that catalyst that's making, machine learning in particular very exciting. And, you know, even if you look at. I, Pete might argue, uh, up to the contrary, but for the most part, you look at edge ai, it's machine learning, right? ML ops, ML tooling, all these things have gotten much better. The libraries are massive now, right? Think about it. So, absolutely. You know, model builders, AI solution providers can have tap into this vast resource. Now that wasn't there before, what we're seeing with the operators with ai, it's very, very difficult to get like transformer based or some of these more robust, AI models, reasoning models of well close to the radio. In fact, you know, one of the hot areas. And this is something I wrote about a long time ago, is, the R apps.

Marc Pous 15:40

okay.

Leonard Lee 15:40

You know, part what they are basically these wrappers for, automation. Okay. It can be AI based, it could be, programmatic scripts that gripe automation at an orchestration layer, which is called the SMO. It's part of the whole open ran framework, right. O ran framework. Mm-hmm. And that's becoming really hot now. And the ecosystem is starting to build, but it is at a higher level. Right. And above that, you're seeing some company trying to do some, agent stuff. Like, Ericsson announced that they have a partnership with AWS, to offer what's called, our apps as a service. The whole idea of it is you're using Bedrock Strand, a lot of the AWS to, a. Log frameworks above the SMO so that you could do, I don't know, intent driven management, orchestration of the, the network, but also Qualcomm, the guys at Qualcomm with edgewise, which is Qualcomm's, SMO. They also have a very similar thing that they actually announced this at MWC 2006 as well. The stuff is, it's early, but the thing that actually is making impact and where like operatives are actually seeing value, right?

Earl Lum 17:06

Yep.

Leonard Lee 17:08

It's like frigging ml and you know, here's

Earl Lum 17:10

is operations, the back end. Yeah. Maintenance monitoring of radio links. Yeah. Are they degrading? Can you see when they're gonna die? Because most of the time you don't wanna wait for the fault alarm on the network before you do something'cause it's too late. So how do you do that predictive analysis over the last 14 days? This node has been degrading and it's probably gonna have something bad happen to it. So how do you go out there and fix it before it actually gets or it's failed? Those are not sexy things. No. Yeah. But those are the things that save the operator money.

Leonard Lee 17:46

Yeah,

Marc Pous 17:46

absolutely.

Leonard Lee 17:47

And so like all these people that are in the conference looking at these frigging humanoid robots and stuff and thinking, oh, that's so cool. But I, I posted if you thought that this is the coolest thing. Here you number one, you completely missed a boat, and you're focusing on the most trivial, stupid thing at Mobile world. Any mobile world, any

Marc Pous 18:13

conference, the shiny object, right?

Leonard Lee 18:14

Yeah. Right. And so anyone who, by the way, anyone who has written that humanoid robotics was the thing at MWC next year, don't come back. Right. You know what? And throw away your laptop.

Earl Lum 18:29

And one of the cool things that you might've missed at the Fujitsu booth was the ability to be Godzilla and make your own Godzilla film by acting out God acting things.

Leonard Lee 18:38

Well, no, no, but but see, the reason why I say this is we've seen this in IOT. The thing that pisses me off or in anything is like jokers that go around saying that human robotics are the biggest thing at a mobile wireless conference. And then having guys that are working on our apps actually. Doing things that are helping, operators get absolutely no attention. Do you know what I'm saying? And they got zero attention. These are the companies that are doing something that's real with machine learning that's real, right? And they get zero. That's the crime of height and why, you know, these guys that go around looking at the shiny objects and making a big deal out of it, distracting all the attention away from people that holds the industry back. We saw that with Metaverse. We're seeing that with agent ai. We saw that with copilot. By the way, copilot was a flop.

Marc Pous 19:33

What? What do you mean that

Leonard Lee 19:35

copilot? Copilot, you know, three years ago when they tried to make copilot everything in front air, everything with like chat, GPT flop, I don't think I've spoken to a single ISV that went headfirst with the copilot thing three years ago. None of'em have made that revolution. Right? But then you have all these guys that were doing simple, boring things like intelligently shutting off a radio or a sector or whatever, right? Saving a operator, 2015, 20% on their opex for energy

Earl Lum 20:18

or improving field tickets and all those other things that are boring, but are critically important for your network operations, how much you spend on that going forward.

Leonard Lee 20:28

yeah,

Earl Lum 20:29

obviously getting rid of people, right?

Leonard Lee 20:31

Yeah. At some point. Intelligent link adaptation, all these things that are, that enhance the user experience, right? These are like the things that are important. Not like a freaking, robotic dog.

Marc Pous 20:44

Yeah.

Leonard Lee 20:44

Right. So anyway,

Marc Pous 20:46

And what do you think about the six G has been a lot of noise about six G.

Leonard Lee 20:50

I know. Did you guys, do you guys hear anything?

Earl Lum 20:54

I know what, I know what the term is or I think I know what the term is supposedly. That it's a sensing network of some type. Some of the people that I talked to, are we just creating a bunch of radars on a macro network for six G?

Leonard Lee 21:12

Pretty much,

Earl Lum 21:13

right? Let's just call it what it is. It's a radar.

Leonard Lee 21:16

radar. It is,

Earl Lum 21:17

right?

Leonard Lee 21:18

It's a radar. I talked to John SME about it

Earl Lum 21:22

radar,

Leonard Lee 21:22

and I said, yeah, it's basically a radar. I talked to the guys and a lot of people may disagree

Earl Lum 21:27

with that, but you know what, it is a radar man.

Leonard Lee 21:29

It is all part of the infrastructure becoming cyber physical. So my key top saying for MWC 2006 is, you know how like everyone's talking about physical AI and they get the robotic dogs and the frigging humanoid, the network is the telco industry's physical ai, right? Because that's what you have to care about. Not the robotic dog, not the humanoid, not all this other stuff. It's the network, right? The air interface, it's the UE plus the network and allowing them to work better together, be more adaptable, right. And then take things to the next level. But then we have all these people that somehow think physical AI is all the stuff that particular company

Marc Pous 22:20

Yeah. Is

Leonard Lee 22:21

defining it as, and it's not. Right. But actually it's a rehashing of stuff that's already been there before. It's just people are discovering, oh, it's people who don't know that are gonna discover physical AI is stuff has been there before for a really long time. Yeah. Right. That's the,

Earl Lum 22:39

but going back to your question about 16, is it gonna be late? Probably. To what degree do people care about it?

Leonard Lee 22:49

Yeah,

Earl Lum 22:49

maybe little. And that scares everyone at this show because without that G driving some demand and hype. What are they gonna sell?

Leonard Lee 22:58

I did have a conversation with Eric Putin, who is the CTO of Ericsson, as well as talk to me about this. And, you know, the, the odd thing is they don't wanna suggest, they wanna suggest that it's a software upgrade.

Marc Pous 23:14

Right.

Leonard Lee 23:15

But you need just because hardware,

Marc Pous 23:17

NBI ot, thing

Leonard Lee 23:19

yeah. Sort of, it's like just a, a software thing, right? It's not like a physical hardware thing, but I don't,

Earl Lum 23:28

but it kind is. Today's base stations are not radars.

Leonard Lee 23:34

Yeah. And then

Earl Lum 23:35

so you need that radar, which is a physical asset you have to buy, which is at CapEx cost. For the operator to go deploy that. And I've been telling people that whatever hopes and dreams you may have for six G, it'll be half the number of volumes of units you had for 5G because people are gonna have to do ran sharing. A lot of network sharing's gonna have to happen for six G just because of the economic cost of deploying that technology and all those radars.

Leonard Lee 24:07

But one thing that a lot of people have been talking about related to six G, which actually is also related to 5G advance, is uplink. So there's a lot of conversation about uplink because you know, people scratch their heads wondering why, cellular iot didn't take off well. Number one, it was a differentiated right. Versus Laura and all these other. Lower cost where the module cost was much lower right? Then, all the reasons why. BIT really didn't take off outside of right. We've had endless conversations we

Marc Pous 24:43

I, I think I already forgot, forgot we talked too many times about this.

Leonard Lee 24:50

Yeah. Okay. They, I don't, I, quite frankly, I think they have fully thought through what the hardware changes need to be. They think that, oh, it's just gonna be a flip or switch. But I think in the next couple of years, there's gonna be, as you get closer to the actual timeframe where. Thinking around, six Gs sure there's gonna be a realization you're going to be, need new hardware or some new hardware. It might not be a complete replacement, but you need something to support these new features. And it's not just a software upgrade.

Marc Pous 25:29

Ericson, Huawei will be really sad if they, if they kind of sell infrastructure.

Leonard Lee 25:36

Yeah. Well, you know, but there's that. But then there's also, how do you achieve some of these new features that. Move the ball in terms of evolving mobile wireless technologies,

Earl Lum 25:51

right.

Leonard Lee 25:51

Infrastructure as well as computing. Right. Because mobile computing, when we look at all these new car, classes of devices, like with Qualcomms, Snapdragon wear a elite.

Marc Pous 26:04

Yeah.

Leonard Lee 26:04

You know, they're trying to usher in all these ai well, stuff doesn't really run all that great on today's, 5G NSA primarily 5G NSA networks, right? Some people argue that, oh, it's okay, but you need to, in many instances, probably use like a red cap, right. For the, you know, the link budget as well as the power budget that you have to, deal with these mm-hmm. These form factors. So.

Earl Lum 26:38

I think the whole issue about six G is if, will it make less money and 5G is made. And is there going to be this issue? Where do I, do I have to do it? Which is probably yes, but how much of that do I do knowing that I didn't make any money on the last one? This new one's probably not gonna make money. And so we're still, we still haven't answered the question of how we make money.

Marc Pous 27:10

Yeah. Well this has been on the table, this topic since 4G, right?

Earl Lum 27:16

Right. 4G made money because of the data and all this. But did it cap out?'cause we didn't pay any more for 5G. We get more data. Do we need 10 gigabit per second for six G? No.

Leonard Lee 27:31

Yeah.

Earl Lum 27:31

Not me. I don't wanna pay anymore because what is it that I'm gonna critically need as an application that I don't already have today that I don't know I need in the future?

Leonard Lee 27:45

but the thing is this is where I found The mindset for, or the, so the new thinking of how you innovate on top of, new infrastructure, new capabilities. It's not there. Like for instance, with network APIs, everyone's still scratching their heads wondering, how do you monetize network APIs? Well, first off, you're probably not gonna monetize network APIs. It's gonna be whatever kind of service or new capability that you're able to deliver to. The end user.

Marc Pous 28:22

Yeah.

Leonard Lee 28:22

Right. And then how well is that whole stack understood from the network capabilities that are being continually expanded, enhanced, through like these 5G upgrades transitioned to 5G sa and then broader deployments, right? Coverage, availability, these are all gonna continue to expand, but who's thought through that whole stack? And actually there's not a lot of people who've done that, and that's why monetization continues to be a question mark because you need to have a service.

Speaker 4 28:59

Yeah.

Leonard Lee 28:59

That's how you monetize. But then if you don't have a architecture for defining services, differentiating services that capitalize on what's different. Then you're never gonna be able to monetize beyond what you are familiar with. And that's the problem people are thinking in terms of monetization based on their legacy thinking versus, and then also referencing a bunch of use cases that have been proven well initially were not very well thought out. There was just like people just like guessing and people continuing to work off the same PowerPoints that we saw five years ago. I mean, honestly, it's like kind of a joke. Wow. Yeah.

Speaker 4 29:48

Anyway. Have

Leonard Lee 29:50

we had enough to drink? Did you try this? no, I have not, not should give that a try. Yes. This restaurant is very, very good. Very good. So what else did you think was cool? Any IOT stuff? Mr. Iot Giant.

Marc Pous 30:10

Think about that. Oh, one thing that, well, it's all, ai actually what I am, what I've seen, but something interesting was, I had a meet and greet with the, the dj, the famous dj. Oh,

Leonard Lee 30:24

yeah,

Marc Pous 30:24

I explain how. He's starting to introduce AI in his creativity

Leonard Lee 30:31

Yeah.

Marc Pous 30:31

Concept, how he's embracing ai. He says that is still not good enough to, to, he said something like, to accept the AI generated music into his catalog of songs. However, it starts to be like a good tool to, to get ideas or to get the specific, yeah. Moments of song or, or, or techniques to, to apply on his new music.

Leonard Lee 30:58

Yeah.

Marc Pous 30:59

the funny story is that he was basically in spite like, and the singularity concept from 15 years ago.

Leonard Lee 31:09

Oh really?

Marc Pous 31:09

Right. So it's, still, being inspired by that. What else? Well, I have been a lot with the Arduino people who, who has a really, really interesting single our computer and, very low cost and yeah, and people love it. So it's a, it's a very pie with this microcontroller, very well integrated to have like real time operations that allows you to do, that. You can do with a, with the usual art know microcontroller, but with this dual brain of having a Linux Dian device on the other side where you can do more high level running, connect all reach with a protocol. And that was super interesting to see the engagement of the people. Um, again, for the Arduino, for the Arduino grant.

Leonard Lee 31:58

Yeah.

Marc Pous 31:59

and honestly, I've been only. One hour at all Congress. I haven't seen, match.

Leonard Lee 32:06

yeah, that's one, one hour more than Dean Bubbly.

Marc Pous 32:10

Oh, come on Leonard.

Leonard Lee 32:12

I saw him at the Amdocs party, so Yeah,

Earl Lum 32:16

he's gonna hate you for that.

Leonard Lee 32:18

No, he'll love me for it.

Earl Lum 32:19

He,

Leonard Lee 32:20

he knows, he's proud that he, he doesn't go to, he doesn't go to the fea, so, but, yeah, it's, sovereignty. We talked about sovereignty. We talked about AI six G. yeah, I, I think they, they tried to make this about Telco, Neo Cloud. I don't know how much it,

Marc Pous 32:41

What is

Leonard Lee 32:42

tele Neo Cloud Cloud? Neo Cloud, like GPUs and trying to create like data, sovereign data centers.

Marc Pous 32:48

Oh yeah. We were talking about sovereign data centers before,

Leonard Lee 32:52

right? Yeah. But yeah, I mean, you know, they, a lot of the companies that are doing it feel like the resource constraint, you know, not being able to get GPUs. Then when you ask them, what are you doing with this stuff? They don't really have a good answer. Well, we're building it and hopefully somebody can scale out on the application once they figure out the value of it. And so it's like, you know, it's, and you know, the thing that worries me is that a lot of the companies that are doing this are, like the South Korean companies,

Marc Pous 33:29

okay?

Leonard Lee 33:29

And the one thing that bugs me is this, during the Metaverse type cycle, I'm speaking to Korean companies. They held onto Metaverse even though it collapsed, two years before they were the only ones still sticking around with startups and companies going, going and setting up metaverse booths. And it was like a, it was a freaking joke. And so my fear is that they're gonna hang on in this very expensive game. Of, the Neo cloud game, right? and the reason why I say this is, could be very unfortunate, is like everyone said, everyone likes to say, oh, there was nothing. You know, this is real, right? AI is real, and AI infrastructure, all this investment, it's real. And chat GT really does little things. Well, guess what? So did Metaverse. It did, right? Yeah. I mean, the technology was there. You can get immersed in some goofy cartoon, cyberspace space, and people were dropping Snoop Dogg. Millions of dollars on digital, what do you call real estate? Yeah, yeah. Think about that, right? I mean, a grown ass adults dropping hundreds of thousand dollars digital real estate. Oh my God, I, I'm sorry. I have to laugh, isn't it? I mean, what do you think now? Okay. In retrospect, now that Metaverse is over, NFTs are a joke. What do you think, in retrospect, but this is the kind of stuff that this industry has taken seriously at that time. Yeah. It was the headline. We took it. I'm not, I didn't take it seriously. I called it BS from the beginning, but

Marc Pous 35:31

yeah,

Leonard Lee 35:31

the industry took it 1000%. Seriously.

Marc Pous 35:36

Yeah. But don't you think that the industry at that moment thought that their adopt, that they could accelerate the adoption of technologies? because I mean, it's not complete different technology. Probably not for the target audience was not the generation. That was actually, no, the target audience,

Leonard Lee 35:55

they thought so, but I mean, it's like

Marc Pous 35:57

probably they will need one or two gener new generations to, they wanted to go so fast, I guess with this, not,

Leonard Lee 36:04

no, but I think Metaverse also, people didn't really know what Metaverse was.

Marc Pous 36:09

I don't know.

Leonard Lee 36:10

nobody knows. A lot of people think they know ai. They don't know ai. They think that they know ent, ai. They don't know ai. And so I don't know what's different. Right. What, and you know, what's not different is that. The people who are actually materially delivering value.

Marc Pous 36:32

Yeah.

Leonard Lee 36:32

And they are not recognized. And I think that's the big problem with the industry.

Marc Pous 36:39

Yeah.

Leonard Lee 36:39

You should be challenging investments to those companies. You should be paying attention and putting a spotlight on those companies. Right. Like in the iot there were a lot of great companies, a lot of great people doing great things. Got none of the attention because some platform guy was trying to sell how platforms are gonna revolutionize everything. You can scale all this stuff. None of that happened. Right. That's just my point.

Earl Lum 37:08

I think there are a lot of companies that are doing the mundane technologies and hardware out there that are critical for all of this stuff to work on a day-to-day basis that we're probably in the noise here at the show. Because they were just blown away by the volume of hype around AI and everything else that the big guys have been pushing, because they have to have that narrative.

Leonard Lee 37:37

Yeah.

Earl Lum 37:38

But again, all of these are concepts that may or may not get adopted at some point. Further down into the, into the future that don't address today's current problems that we talked about last year, about why are you not fixing the problems we have today? Yeah. Trying to fix the problems we don't know we have for the future.

Leonard Lee 38:00

Yeah,

Earl Lum 38:01

that still is the same problem here.

Leonard Lee 38:03

At

Earl Lum 38:03

this show,

Leonard Lee 38:06

and the biggest joke, quite frankly, is that when you go and you watch the keynote and they get up on stage and say, the industry, you know, like revenue industry wide revenue globally is only growing at 4%, which actually is not that bad considering how big mobile, wireless industry and telco industry is. So they never tell you that 4% increase in, communications., In absolute terms is bigger than how quickly, as big, maybe slightly less than how big, how fast, cloud computing infrastructure services is growing. And I put out a chart about that just to remind the, yeah. People look, learn your numbers, right? There's a thing called law of small numbers. There's a lot of big numbers, but, yeah, the joke is when you get up there, you malign your own industry and, but you're hyping nonsense that distracts from the efforts that are actually moving the ball, right? That could drive monetization through modernization. And, you know what I'm saying? It, it's like, like all the stuff that we've been talking about on iot Coffee Talk for. Going on seven years now, you know?

Marc Pous 39:23

Yeah.

Leonard Lee 39:24

I think the other hype N TNS

Earl Lum 39:26

satellites.

Marc Pous 39:27

Oh, oh yes.

Earl Lum 39:28

Talking to, you know, the little communicator here. Right. When, when do we, what We really, really use that? Or is that really military oriented? Or Leo?

Leonard Lee 39:42

Yeah, pretty much. I mean, you know, it's like, it's, you know, uh, what, what, we've said it before, but it's like, what is, you know, how big is the business of connecting the middle of nowhere and somebody living in the middle of nowhere

Earl Lum 39:58

very small,

Marc Pous 39:59

looks like a SpaceX, found that, not sweet spot with, people with no activities.

Leonard Lee 40:05

Yeah. You, you'd like to think that, but I think they've found a bigger one. Helping milit, militaries, orchestrate and guide their ordinance.

Earl Lum 40:14

With the, all of those satellites that

Leonard Lee 40:16

no one know

Earl Lum 40:17

is, is up there. They've been launched, they're cornering a bunch of stuff that we're not, that we've seen over the last several weeks. Somewhere out there in the world. Right? Yeah. And they, I

Leonard Lee 40:30

mean, they have, they have, there's limits to the capacity, so this idea that, terrestrial networks are gonna go away, never go away, and then, I've heard some really ridiculous arguments that, to the effect that terrestrial networks are gonna go away, right?

Earl Lum 40:47

no. Any, anyone out there that thinks that is too, should not ever come back to this show because you're too stupid.

Leonard Lee 40:55

Oh

Earl Lum 40:55

my, my God. If my phone is talking to a base station that's 200 feet away, why would it want to talk to a base station that's 400 miles? Up in the sky.

Leonard Lee 41:05

Oh, in, in some case, 30,000.

Earl Lum 41:08

Yes.

Leonard Lee 41:09

Kilometers away.

Earl Lum 41:10

The network will choose the best signal, which is Terres Lee, every single day. Unless you're in the middle of nowhere and you're gonna die, then you buy.

Leonard Lee 41:19

one of the things I heard about N nt N it, that was pretty common across the board was it's niche.

Earl Lum 41:27

Yeah,

Leonard Lee 41:28

it's niche.

Earl Lum 41:29

Just like iridium. Yeah. What is niche? Global star. Yeah. Maritime stuff. Yeah. You are on a boat and you want connectivity and all these other things.

Leonard Lee 41:38

I mean, it's absolutely true. we're not gonna name names, we're not gonna give details. You want to talk to any of us? Give us a call for your client.

Marc Pous 41:47

So what you want. It's very cheap to launch. Leo?

Earl Lum 41:51

No, it is extremely expensive.

Marc Pous 41:53

Yeah.

Earl Lum 41:53

cheaper, but it is only cheap for Starling because it can watch a rocket every day. But if you're like the Chinese

Marc Pous 42:03

deal, right?

Earl Lum 42:03

If, if you're like the Chinese, they're trying to do a 30,000 satellite constellation, it's going to be extremely expensive because they don't have reusable rockets today.

Leonard Lee 42:14

Yeah.

Earl Lum 42:14

And

Leonard Lee 42:15

Not yet. Not

Earl Lum 42:16

yet. They will have it eventually. But the issue is when you try to put cellular connectivity, you're putting a base station on the satellite. It's not just A-K-U-K-A band link. That has always been, so all of that stuff adds weight, which means these satellites are heavier and heavier than bigger solar panels to do the power consumption of not GPUs in space.

Leonard Lee 42:39

Not yet. Ouch. Not yet. Or.

Earl Lum 42:43

I don't know if you're gonna need it, but weight is everything when you're launching a satellite. So the question becomes these, the lifetime of a LEO satellite now I've heard is not even the five years originally, anticipated, we're down to three. So every three years you're launching everything that died three years previously that you launched. So it is a great market if you make components for the satellites because they'll just die every three years.

Marc Pous 43:12

And the, and,

Earl Lum 43:13

but as a service provider, maintaining that network, can you imagine a mobile network dying every three years and you had to replace everything

Leonard Lee 43:22

G six

Marc Pous 43:22

Q

Leonard Lee 43:22

AI supercomputer.

Earl Lum 43:24

Yes, so it's gonna be really expensive for a satellite operator maintain its network in the sky that has a churn of every three years.

Marc Pous 43:35

I'm talking about non-terrestrial, satellites in data centers

Earl Lum 43:40

in space,

Marc Pous 43:41

Spain in space. It's gonna be Leo or, or more far away.

Earl Lum 43:46

It really depends on the, what they're trying to do. Right. And I think if it's gonna be, it could be a neo, it could be a geo, depending on what is it that you're storing and computing up there.

Leonard Lee 43:59

yeah,

Earl Lum 44:00

if you're doing it in the leo, the problem is if you're meshing that many data centers at a low earth orbit and you kill a data center every three years because the whole satellite dies, it's probably not gonna be in Leo or should not be in Leo just because of the finance. Right. Of having to kill all your data centers every three years and then where are you transferring all the data? Is it going eventually back down to earth to some little island that's sovereign in the middle of the Pacific Ocean where you don't want anyone to know what data it was that you were running?

Marc Pous 44:31

Yeah,

Earl Lum 44:31

Up in space Or where do you store all of this information? Because if it's stored on the satellite and the satellite dies.

Leonard Lee 44:37

Yeah.

Earl Lum 44:38

Where's the backup? Is it another satellite that just is dying a year later?

Leonard Lee 44:42

Yeah,

Earl Lum 44:42

everything gets pushed off.

Leonard Lee 44:44

the problem is, you still have to answer the question of why all this build out for AI supercomputing data centers? Why when, the monetization in the end market is not there. Um, the demonstrable revolution that everybody had theorized it's not materialized, maybe outside of coding. Coding, definitely, like we were talking about before, benefits, does it wipe out developers? It shouldn't. It should enable developers and it, oh, by the way, little hint, for those of you who got laid off from a big software company, because they told you, they laid you off because the ai, why don't you use that very same AI to build a company and software to take out the company that fired you? Do that right? And you could name that company. Revenge.

Marc Pous 45:48

Yeah.

Leonard Lee 45:50

Payback. Payback.

Marc Pous 45:52

That's a huge method. Yeah,

Leonard Lee 45:53

I mean, so don't worry you have AI on your side. If it's really, really that great, you know, I mean, think about it. you should think of it as an opportunity, not a tragedy in your career. So channel that anger.

Marc Pous 46:10

No, but going back to the why to have, data centers in space, I think it doesn't doubles, Elon Musk explaining that the full stories about the energy, right?

Earl Lum 46:22

Right. It's space in a vacuum saves on cooling. Right, because it's cold

Marc Pous 46:29

in the vacuum, unlimited solar energy because it always, the sun can power the satellite.

Leonard Lee 46:37

Sometimes there's a, there's a thing called the dark side of the earth,

Earl Lum 46:40

But the other, the other thing that you have to worry about is this every 10 year solar flare cycle that happens in the sun, that pops everything out,

Leonard Lee 46:48

there's a thing called cosmic.

Earl Lum 46:50

So if the cosmic gray

Leonard Lee 46:52

actually

Earl Lum 46:53

orientation, you can't really shield it a hundred percent. So you have this ultra expensive data center that has massive, because you don't have to have water to cool it and all these other things because space is so cold. But then the solar flare comes and takes out your whole system. Then what do you, what again? Where is the backup?

Marc Pous 47:12

No, no. There is a complete, organization, I don't know if it's United Nations or what, that basically monitors this and just contact, so it's

Earl Lum 47:21

Right. All of the junk that's up there.

Marc Pous 47:23

Yeah, yeah, yeah. Is

Earl Lum 47:24

tracking everything.

Marc Pous 47:25

Yeah, exactly. So they, you as a company who has satellite space, you need to follow the roads of this organization. And move your satellites. When they ask, theoretically turn them off, or

Earl Lum 47:37

you're supposed to do that, do they do that? Not all the time.

Marc Pous 47:41

Yeah.

Earl Lum 47:42

But I guess the whole point of the data center space, one of the points is we don't have to worry about water resources on planet Earth to cool it and the maintain the energy. But think about how many solar panels you would need for a gigawatt in space and how efficient so solar panels are today to enable that. The reason why all of the starlink satellites were direct to cellular get bigger and bigger is because they need more solar panels to power the base stations that are on there.

Marc Pous 48:15

Yeah.

Earl Lum 48:15

And none of'em even have a data center yet. So if you're going to even talk about a hundred megawatt data center in space, are you going to need 10,000 square meters of solar panels to create that amount of power to power that, that data center?

Leonard Lee 48:32

Yeah, that's gonna be an issue before you even think about all that stuff, it is the question of why.

Earl Lum 48:38

why is it

48:38

up

Leonard Lee 48:38

why do we need to build all this stuff out? Then what is it for

Marc Pous 48:43

just to save resource in the planet? This what he said, you can invest this energy to do something,

Leonard Lee 48:49

but why even could create that problem to begin with is the question.

Marc Pous 48:55

Yeah.

Leonard Lee 48:55

Right. is it so that, people can have cyber girlfriends and boyfriends? is that, or to do the silly whatever thing that the video, little videos,

Marc Pous 49:05

Act, actually this is interesting. We were talking before about this agent ai, the development thing that says that everyone can build stuff 10 x speed. So that means that we will do, in theory, 10 x more services to sell. So are we gonna be. are this is gonna be demand for those services. how you monetize that? I don't think people, it's gonna, we're not talking about layouts or that we are talking about like, if you have 10 x more products in the market, how do you sell of them?

Earl Lum 49:43

how do you get those products to be visible? Just, we have all these products a day, how many of'em of the available products that the three of us actually consume?

Leonard Lee 49:56

Yeah.

Earl Lum 49:56

That are available on the market today.

Leonard Lee 49:58

Yeah.

Earl Lum 50:00

And if you exponentially increase the amount of stuff that's out there, then the question becomes how do you differ? How do you determine what is it that you wanna buy as the end person? Right? They all look the same supposedly, but maybe one of'em is a better product, quality of wine or whatever How do we do that filtering to figure out what it is that you actually wanna spend your money or time or anything else on?

Leonard Lee 50:25

So I just realized that I think we're, this is gonna be an hour and 30 minutes.

Earl Lum 50:30

It will get,

Leonard Lee 50:30

this is like a No, it won't.

Earl Lum 50:32

No,

Leonard Lee 50:32

it's a like Freeman. Right style marathon. So anyways, we're gonna, we're gonna, adjourn for the evening.

Earl Lum 50:42

So

Leonard Lee 50:43

on this podcast,

Earl Lum 50:43

why don't we just do a quick couple of blank stuff of all the stuff for the last hour, what that we've done?

Marc Pous 50:49

And then one, and did we do, like, like a batch for next year? Oh, okay. We could do that.

Leonard Lee 50:55

No, no, we did,

Marc Pous 50:55

can do that. That's gonna be fine.

Earl Lum 50:57

Alright, so my summary of six six G is not gonna be about massive BIO. 60 G will be about boring stuff, like passive antennas, macro radios. It is a radar. And I don't know what an AI radio is, so if someone can explain that to me, give me a shout out because I have no freaking clue what that is. Yeah. Alright. I pass.

Leonard Lee 51:23

Okay.

Earl Lum 51:23

And I will have less

Leonard Lee 51:24

I think next year there's gonna be a realization that, the killer AI is truly ml, the whole agent thing will play out to dire disappointment and then people will lean into machine learning, because they need to have. Something coming out of all of this AI hype, but that that'll be a, yeah, that'll be a tangible thing. And, I think this podcast is going to influence the industry to have a better lens on what's important next year.

Marc Pous 52:03

Actually, all the model monitoring thing, it's, it's something super hard, but then you want to have, not all the phrases of why not the model took that decision during that moment. Now why that happened? So all these small one makes a lot of sense. And you did your homework.

Earl Lum 52:21

I will add one LA I will add one last thing. I saw a case study of AI determining if there was retail theft in a store, but the camera clearly showed the person stealing the item. So I'm trying to understand what is the AI adding on top of what I visually saw with the camera showing me that they put something into their pocket. Is the AI determining, was it an expensive item that person stole? And why would I care what they did in the aisle of the store when there's a choke point of a door where they have to exit, where if I'm gonna stick anything, why would it not be at the door to detect what that person stole and put into their pocket? Why not scan the, to see if there's 16 bottles in the jacket of this person? Why would I care what they took off the shelf in the store when they still have to leave the store somehow

Leonard Lee 53:13

so you can sell more cameras?

Earl Lum 53:15

Exactly, so the camera companies, kudos to you. Yeah, good. And I still don't understand the whole

Leonard Lee 53:20

good

Earl Lum 53:20

job, anti-fat use of ai.

Marc Pous 53:23

All right, mark and my, close

Leonard Lee 53:25

it out.

Marc Pous 53:25

And, and just to go against your bet, I'm gonna say that next year it's all about agent phones. It's gonna be all about agentic phones. You're gonna have your agentic phone with your own agent running all of your agenda, your emails and everything. And you're gonna be super proud of it.

Leonard Lee 53:45

Super proud or super creeped out. No, I I, that's a good one. That's it. Yeah. It's, I could, I could buy that.

Marc Pous 53:53

Yeah.

Leonard Lee 53:54

It's just

Marc Pous 53:55

not like that. The

Leonard Lee 53:56

I pc not, not, not, not it. It's gonna be modest though. I think which is fine, as long as it's safe. Yeah. That's all that matters.

Marc Pous 54:05

Well, we'll need, multiple, which is really tough to understand if it's yeah, if it's safe or not.

Leonard Lee 54:10

Yeah.

Marc Pous 54:10

What are you gonna prove it's safe or not?

Leonard Lee 54:12

you don't lose all your money.

Marc Pous 54:14

account. Check every second.

Earl Lum 54:19

So it is AI ran going to be the next open. Ran.

Leonard Lee 54:24

No, it's gonna be the next, ai. AI ran is gonna be the next open Ran quid

Earl Lum 54:28

game, quid game.

Leonard Lee 54:29

Yes.

Earl Lum 54:30

So we're gonna start a new chapter of Squid game.

Leonard Lee 54:32

Game it's called. And the AI ran,

Earl Lum 54:34

AI ran in on top over, over whatever.

Leonard Lee 54:38

Yeah. So we're going overboard now. And then also Sovereign is going to be Sovereign plus defense.

Marc Pous 54:46

Yeah. That's, uh,

Leonard Lee 54:49

military is going to get in big with 5G technology. Okay. Which is different from standards, which is different from what the industry implements. And you need to know the dis distinction. Yeah. Just because you use 5G technology doesn't mean that you're using 5G.'cause you can use 5G technologies for all kinds of things. Outside of the objectives of what the industry is trying to do.

Earl Lum 55:20

Right. Commercial communications and other things like that.

Leonard Lee 55:23

hey, thanks for listening to this very, very long ridiculously yeah, ridiculously long podcast. But this is what you get. This is our tradition. This is how we end MWC, every MWC with a little bit of vino, some incredible food, and a beautiful hole in the wall restaurant

Earl Lum 55:48

that you should go to.

Leonard Lee 55:49

You, you should go. Food

Earl Lum 55:51

is fantastic.

Leonard Lee 55:52

Yeah. If you don't listen to anything that we say, at least you should

Earl Lum 55:55

come here

Leonard Lee 55:56

This restaurant, come here. This recommendation, which was made by the iot giant is Yeah, deso. So, remember to fall next, correct? www.next-curve.com Also. Lum at, EJ Wireless Research or EJ wireless.com, right?

Earl Lum 56:15

Yeah. EJ wireless.com.

Leonard Lee 56:16

Yeah. And then, the iot Giant, he's at Edge and Pulse, just call Qualcomm.

Marc Pous 56:23

Yeah. And IT stars and I'll see you next year.

Leonard Lee 56:26

Yeah. We'll see you next year. different place, different bad time, but we will see you next year To recap, MWC 2027.

Earl Lum 56:38

Oh my God.

Leonard Lee 56:38

Right. All right. See you guys.