MWC Barcelona 2026 | Sarbjeet Johal, Stackpane & CUBE Collective

Clips
More from MWC Barcelona 2026

Sarbjeet Johal

Founder Stackpane, Analyst CUBE Collective

Stackpane

play_circle_outline MWC26 Barcelona: AI-Telco Convergence, Edge Compute Surge, 6G Demos and the Monetization Challenge

play_circle_outline Edge orchestration emerging; static network rules insufficient for disaggregated devices

play_circle_outline NVIDIA, Axelera, Groq: Interconnects and Specialized Inference Hardware Driving Soaring AI Compute Costs

play_circle_outline AI-RAN and RIC Spotlight: Nokia vs Ericsson — Vendor Control Limits Telco Innovation

play_circle_outline Sovereignty, regulation, and local markets present telco opportunities and constraints

Info
Transcript

Sarbjeet Johal, Stackpane & CUBE Collective

Sarbjeet Johal

Founder Stackpane, Analyst CUBE Collective Stackpane

In this interview from MWC Barcelona 2026, Sarbjeet Johal, founder and chief executive officer of Stackpane Ltd., joins theCUBE's John Furrier to discuss how AI is reshaping the telecom industry from radio access networks to the far edge. Johal highlights the surge in compute demand driven by AI workloads across both consumer and enterprise sectors, noting that AI-RAN has become the defining theme of this year's show. He explores the coming convergence of 6G, quantum computing and classical compute – a trifecta he believes will push the industry to its next i... Read more

explore Keep Exploring

What are the implications of AI, 6G (and upcoming quantum computing) for telecom operators in terms of compute and bandwidth demand, network management, service quality, and monetization? add

How are telecoms (for example AT&T) using AI and new models to manage network quality, traffic routing, and radio access networks (AI‑RAN), and why are static rules insufficient for highly disaggregated wireless infrastructure? add

Do you agree that networking/interconnect and specialized inference compute (not just raw edge CPUs/GPUs) are key drivers of AI hardware success, making inference increasingly compute‑hungry and expensive? add

What was your takeaway from the show, and what was your biggest "aha" moment? add

What is the current outlook from industry leaders on the telecom and AI landscape—key trends, competitive dynamics (e.g., Nokia vs. Ericsson), and the impact of sovereignty and scale? add

bolt Powered by CUBE AI

Sarbjeet Johal, Stackpane & CUBE Collective

search

John Furrier

>> Welcome back everyone to theCUBE's live coverage here in Barcelona, Spain for MWC26. Formerly Mobile Role Congress, I guess they changed it to MWC maybe many years ago, but we call it MWC. I'm John Furrier, host of theCUBE. Sarbjeet Johal's here. He's the CEO of Stackpane and also a distinguished CUBE Collective research analyst with theCUBE. Also, well known for his Analyst Notebook segments at these marquee events, of course, Mobile World Congress here. Sarbjeet, great to see you again. Been following your tweet streams. You're covering the show like a blanket, looking really good, the keynotes. Thanks for coming on. I guess the first question before we get into some of the notes is we had Carolina on earlier. We've got Bob O'Donnell coming on. Dave and I have been unpacking it, breaking down everything, putting it back together. It's a complex show this year. There's a lot going on.

Sarbjeet Johal

>> Yeah, it is.

John Furrier

>> That's the general consensus among the top analysts. What is your take? What does your Notebook tell us? You've been getting all the stories, you've been briefed on all the information from the vendors. You're talking to the telcos, customers, and the enterprise. What's jumping out at you here?

Sarbjeet Johal

>> At highest level, we need more compute, because the data volumes are going up and AI is actually, in many ways, data factory itself. It produces a lot more data on its own because all the stuff we are producing, it's going somewhere, it's traversing over the wires. And it's a consumer play in a big way, as well as B2B. So on the consumer side, that's where the telcos are coming to start with. So they will get more data volumes, to process that they need more compute.

John Furrier

>> Huge demand on the compute side.

Sarbjeet Johal

>> Huge demand on the compute side. And then customers are looking to telcos to reduce the fraud and create the better service levels and then give them better connectivity everywhere. So AI is playing a huge role. So if you go around the expo, the AI is written on every booth. So AI-RAN is the big thing. And actually, I said that in dad joke, I said, "AI-RAN, where did it go?" So then actually this is AI show to start with. And the 6G is being introduced, it will land in about three years or so, we are told. And it will land at the same time when we are expecting quantum to hit the market. So AI plus quantum, plus our classical computing, it will create this trifecta of factors which will take us to the next level. It will be a thirst of going to the next level and then-

John Furrier

>> The 6G is interesting. Ericsson's got a pretty damn good demo. We see 6G being pervasive devices because they have shorter distance, but with terabit speeds coming out, we had Cole Crawford and Allen Salmasi from Veea. Allen invented TDMA Technology co-founder, Qualcomm. He knows a little bit about spectrum and wireless. And they're seeing this orchestration model emerging. They're seeing the carriers and operators having to move from legacy thinking of selling connectivity, which by the way, we need more connectivity. We've done that with 5G, it's faster, latency's better, but there's no monetization, Sarbjeet. So you're seeing everyone scratching their head saying, "Wait a minute, the edge and all these things happening, we need more bandwidth." But when you bring AI intelligence to the network, good stuff happens. We had AT&T on talking about how they moved from analytics to putting up 30 new models. You're starting to hear things like, "I could create quality of service and new policies around how networks are configured, how traffic's managed and routed." You can't have static rules when you have so many disaggregated devices.

Sarbjeet Johal

>> Exactly.

John Furrier

>> You can't have a KV cache across radios. You have to have the radios talking to each other.

Sarbjeet Johal

>> Exactly, yeah.

John Furrier

>> So I think AI-RAN, the radio access networks, are going to be a big conversation. I mean, here we're in our world, cloud, we talk about pipes, and intelligence, and connectivity, but there's also the wireless piece, the far edge. So again, the show's got a lot of moving parts because you got infrastructure, legacy, towers for cell towers, cellular towers, you got central offices for traditional telecom connectivity. Then you got the enterprise, they want their own sovereignty too. So you got all these moving parts, sovereignty, AI injected into the network. How do you make sense of it? Is there a common pattern that you're seeing here, or everyone's just like mind blown on the architectural shift that's happening?

Sarbjeet Johal

>> Yeah, actually it's everybody's dream goal to hop onto the latest, whatever the latest happens to be. And telcos are no exception to that. The telcos actually, historically, you will agree with me, when the cloud was hot, telcos wanted to do cloud. AT&T was doing that, and BT, and we are cloud providers. I was doing dot-com days working at B2B e-commerce, they wanted to set up the exchanges. Every telco was our customer, I was flying-

John Furrier

>> We want to do everything.

Sarbjeet Johal

>> I was flying around the world talking to all these telcos. So they want to do everything. That's a problem. And they actually play decent role in SMB space, smaller and medium businesses, they want to provide services to them and all that stuff. But the problem is scale. Right? Because they don't have the scale, because of that, they can't compete with Silicon Valley. And I actually, this year more so than any other year, I'm more pumped about being in Silicon Valley for 33 years. And being from there means what we do from Silicon Valley, why people can't compete with us? Because we are not regulated, we break the rules, we have all the money in the world, whatever we need, when we can sell the story. So telcos can't do that.

John Furrier

>> Yeah.

Sarbjeet Johal

>> But having said that-

John Furrier

>> Well, no, but I want to just pause you real quick because I wanted to make a comment. Because in the past five years when we've been here at MWC, I missed last year, but I've been here the past four years with theCUBE, since 2010. The telcos, I called them glaciers, because they move like glaciers. Now, and they balked at the cloud, they tried it. But DR, at telco DR, Danielle Royston, she's been pushing cloud native, and she's been now successful, but took her two years. She's grinded hard, she's been pioneering, they now get it. But the force of change is on their doorstep. So to your point about Silicon Valley, I'll just call it the tech sector.

Sarbjeet Johal

>> Yes. Yeah, let's call it that.

John Furrier

>> It's on the doorstep, Silicon Valley, also the tech sector.

Sarbjeet Johal

>> Yeah. We don't want to piss off the rest of the US.

John Furrier

>> New York. Yeah, I'm New York. No, but in general, but it's mainly Silicon Valley.

Sarbjeet Johal

>> You moved to New York and I kind of don't like that.

John Furrier

>> It's innovation. No, I'm coming back. No. Well, we'll see. No, but the point is that they can't ignore the tech. It's in their face.

Sarbjeet Johal

>> Yeah.

John Furrier

>> And so they have to move. Now, I've also been very bullish this year by saying, "They got the power, they got the data, they got the networking." If you look at NVIDIA, their success is because of their networking. The way they did InfiniBand, they brought their systems together. InfiniBand was a critical milestone for NVIDIA. Now they got interconnect, Celestial AI just got acquired. So interconnect's a hot business. Axelera, right next to us here, they're booming because they're building great inference chips. You got Groq now at NVIDIA. Inference is a compute heavy, not heavy, but required. So you got... It's not just throwing compute at the edge. "Hey, we'll just throw some compute at the edge." No, no, no. you got to have specialty compute feeding the inference engines and quasi... It has to be software defined. Your reaction to that?

Sarbjeet Johal

>> Yeah.

John Furrier

>> Am I crazy?

Sarbjeet Johal

>> No, you're not. Actually, you make a lot of sense and you move at 90 miles an hour with your thoughts and I'm trying to catch up. So the inference is compute hungry, there's no doubt about that. And it will be even more compute hungry going forward because we are throwing policy at these agentic frameworks. Right? Because we have to reduce the hallucinations. We have to program our systems of record using the AI constructs. And for that, we need more control over the flow of logic, if you will. So most of the logic in our applications is sitting in the code, and now we're trying to move that to AI tier, that will consume a lot of compute. And just a side note, some of the developers in Silicon Valley who are like, they're trying to do a one person billion dollar company kind of concept, it's a stretch, but they are spending like more than 100 K a year on just AI. The AI bill is so high, because AI is not free. That $19 tier or the $200 tier is just the beginning. That gives you the taste. It's like a drug.

John Furrier

>> First hits free. Now in this case, $19 a month.

Sarbjeet Johal

>> Yeah.

John Furrier

>> Talk about the impact of AI the way you're seeing it, because you mentioned Silicon Valley, they speak velocity, roadmap, architecture, telecom speaks, risk management, durability, returns. And if you look at the investment cycle of telecom, and this is consistent with data centers as well, the hardcore physical side of the data centers is the horizons on the investments are 10 years. And the build out, just on refresh alone, AT&T, Verizon, all the carrier operators are going to be upgrading, retrofitting their existing infrastructure. We think, and we're proving it out here, to have AI factories at the edge. And when you have mini AI factories connected to a centralized big factory, NVIDIA factory or whatever it is, they talk to each other. That's distributed computing.

Sarbjeet Johal

>> Yeah.

John Furrier

>> Okay. If you believe that, which is true, you then say, "Okay, now you have intelligence between factories and it's sistered up against all this infrastructure." That's edge. And then you got the far edge with the radio. So there's some speculation of what use cases might emerge. Services, multiple sets of services. Verizon, AT&T could power agents that are being bought by users in marketplaces like Apple.

Sarbjeet Johal

>> Yeah. I saw an interesting use case being shown at Google Cloud pavilion. So they were showing how... They always actually do this, like how telcos can monetize using AI, monetize their network using AI. So they're saying, "Okay, now you can use third party APIs to map your identity, user's identity, to the SIM identity, and then you can enable the commerce. You can press the button, it knows it's you, you don't have to enter the credit card or anything else. The identity is taken care of." So that's just one use case. But monetization is again for... Okay. Coming back to having software chops or having the deep technical chops, let's say... They don't have it. Okay. Just a side note, like Jim and I did a video about the proximity factor. The proximity factor, I gave an example of like why cloud was concentrated in Seattle. And why when I was at Oracle, we couldn't succeed with Gen 1 Cloud by having the teams in the Bay Area. Why Google Cloud was number three, and still is number three, and moving towards number two, I believe, strongly believe. So the reason is that concentration of know-how was in Seattle, and now in AI space, the concentration of AI is in the Silicon Valley, back in Silicon. We are on top again. So because I'm from there, I always-

John Furrier

>> Of course, you're on top.

Sarbjeet Johal

>> Yes. We're on top again. So if now anybody wants to do AI in a meaningful way, they need to have a lab there, they need to have presence there. AT&T needs to have presence there. They have it, but they need to make it bigger.

John Furrier

>> They've got to grow their ecosystem.

Sarbjeet Johal

>> Yes, exactly.

John Furrier

>> And open source is booming, so it won't be hard. All they got to do is jump in.

Sarbjeet Johal

>> Yeah. So yeah, actually another note I made was that from the day one of the show, the main keynote was that AT&T CEO came to Mobile World Congress after like six or seven years. He never attended. And he was spot on actually, I thought his thinking frames were so clear and they're one of the more advanced provider. May not be the biggest, but they're highly regarded from the technology point of view. He said something very profound, I think. He said, "We move with the core product, our strength," which is fiber for them. He said, "We'll lay down the fiber and everything will hang from there. The commerce, the edge, and working with likes of NVIDIA on this side, and likes of other smaller telcos, or having interoperability on the other side." So I think that is what's needed. And the reason why he could say that, because they have scale. They spent, how many billion dollars? Actually, it was in my note. $14 billion along with Ericsson to bring the O-RAN, to make the O-RAN reality. So I mean, they can throw that much kind of money-

John Furrier

>> The upgrade cycles are going to be massive. So what's your takeaway from the show? What's your big aha moment here?

Sarbjeet Johal

>> The aha moment is, I think AI is democratizing the intelligence. And if the telcos can grab some of that intelligence and weave that into their operations, that would be great. So the RIC, it's a RAN Intelligent Controller, which is proprietary mostly right now, even in the O-RAN world. Likes of Nokia and Ericsson, they're keeping the whole grip on that. If they can release that grip and let the innovation happen, and/or if the VC community can jump in and take a stab at the RIC, RIC part of the O-RAN, I think that will go a long way and that will free these telcos off of this stronghold of a couple of vendors.

John Furrier

>> Any signal coming out of the noise around enterprise, what's your takeaway on the enterprise? Because a lot of enterprise here, all the vendors serving the enterprise are here too. I mean, they have their own needs. 6G is going to be pretty hot for them.

Sarbjeet Johal

>> Yeah. So enterprise actually... Okay, we are here... Okay, by the way, they call these telcos CSPs. In our world where I come from, we call CSPs a cloud service provider. CSP here is communication service provider. So all these telcos are CSPs. And then at the highest level, it's all about communication. So it's B2B, in B2B world, and when we create a purchase order, it goes to another entity. It's communication. Right? So we have standards from EDI, to , and whatnot, in e-commerce, and we have standards in healthcare in US, and there are some global standards but not many. So what I'm trying to say is that the enterprise is a super set of what communication is. So this telecommunication or the communication in general is what they use to conduct their business. They have their own, as we have talked about this in many, many times here, from here and from other venues, that big companies own their own, they have their own networks, they own their own fiber. Right?

John Furrier

>> I noticed you had a-

Sarbjeet Johal

>> That's where the-

John Furrier

>> I want to ask you about NVIDIA and Nokia, because at GTC in Washington DC, Jensen was on stage with Nokia, they announced a strategic relationship around 6G. I think there's a lot more going on there, I'm digging into it. I haven't reported anything yet, but I can see action happening. Certainly edge. Justin Hotard is their, CUBE alumni, he's now running the show. Nokia bought Bell Labs, and they bought Alcatel-Lucent, so not just a finished ex phone company. They're an infrastructure provider. Okay. So they're going to be a big piece. Ericsson and Nokia are two big players. The third big player, Huawei, but they're China, so we'll put them inside for a second.

Sarbjeet Johal

>> There's some Samsung, Samsung too, but that's number four.

John Furrier

>> Samsung, number four. Okay, but top three, put Samsung in the mix. But let's just take Ericsson and Nokia. Nokia is mostly American based, with Alcatel-Lucent and Bell Labs. Ericsson's a little bit more international. So you got sovereignty going on with AI, there's just a cloud conversation. So cloud and AI have intersected. What's your take on Nokia? What's the view from the top there? What are you hearing?

Sarbjeet Johal

>> Yeah, Nokia, by the way, having so many patents and having access to US money, actually, I think they will do good. And yes, Ericsson is leader, but the leader has more to lose than the lagger. Right? So, I think Nokia will do good. Having said that, we haven't actually touched upon the sovereignty piece, which is very prevalent in the AI area. The geopolitics, the war is going on in the Middle East, and there's less trust in US from the rest of the world with Europe having-

John Furrier

>> I think that's one factor, I don't think that's it. Because AI has service opportunity. If AI is going to be accelerating business model, the carriers in these countries have a sovereign opportunity to serve their citizens locally, regionally, nationally, in the country. So I think there's a lot more value on the table with sovereignty, not just data. We know the data, the data, want to keep it in the country for the governance side. There's the geopolitical, "I want to control my own destiny." But think about what a carrier could do in a country or in their sovereign region, they could really tailor services to small, medium-sized businesses, large enterprises, consumers, and use the AI intelligence.

Sarbjeet Johal

>> There are cultural aspects, there are economic aspects. I had a great discussion with Karen Dillon from Harvard University, she wrote a book on how the developing economies can prosper and all that stuff. So every country is different, some countries are more different than others. So for example, India, it's still big rural population. Urbanization has not hit India like it has hit China, for example. So their needs are different, and their economics is different, how much you pay per gigabyte and for this connectivity. The-

John Furrier

>> It's outrageous....

Sarbjeet Johal

>> appetite is different. So definitely, yeah, that global... Okay, this is actually one of the crux of the problem for telcos. Most of the telcos serve locally, only few telcos go across borders. And for that reason, their scale is small. For the scale part and the sovereignty part, because they are tied to the... They're highly regulated by the government, local governments, what you can do and what you can't do. So that keeps them at bay from innovating. So innovation, actually scale is the best friend of innovation. So if you have scale, you can do economies of scale, of course, but also division of labor, you can divide and conquer.

John Furrier

>> AI works.

Sarbjeet Johal

>> Yes.

John Furrier

>> Sarbjeet, great to have you on. As always, phenomenal work. Thanks for coming on. Appreciate sharing your notes on the show. Enjoy the rest of the show. Thanks for coming on.

Sarbjeet Johal

>> Thanks, John. I appreciate it.

John Furrier

>> All right. I'm John Furrier, host of theCUBE. David Vellante is out on assignment, he's out on the show floor getting the data. We'll be right back with more live coverage after this short break.