MWC Barcelona 2026 | Stephen Rose, Render Networks

Clips
More from MWC Barcelona 2026

Stephen Rose

CEO

Render Networks

play_circle_outline AI Demand Sparks Telecom CapEx Surge: Edge Refresh, Data-Centric Networks, Disaggregated Architectures and Strained Supply Chains

play_circle_outline Stop Excel Drain: Buildable Designs and Fast Field Closeouts to Cut Rework and Protect Cash

play_circle_outline Render's Agentic AI and Guarded Micro-Agents Tackle Inspections as Off-Grid Data Centers Strain Power Grids

Info
Transcript

Stephen Rose, Render Networks

Stephen Rose

CEO Render Networks

In this interview from MWC Barcelona 2026, Stephen Rose, chief executive officer of Render Networks, joins theCUBE’s John Furrier and Dave Vellante to discuss the massive infrastructure refresh currently transforming the telecom industry. Rose highlights how carriers are evolving into AI operators, shifting from legacy, voice-optimized systems to data-centric architectures at the edge. He dives into the friction hindering this rollout – from outdated permitting processes to the reliance on manual spreadsheets – and explains why aligning the value chain betwee... Read more

explore Keep Exploring

What are the prospects and implications of a large CapEx build-out of edge telecom infrastructure by carriers to support AI and data-center demand, and how do you see that trend unfolding? add

Are current network/grid rollouts being executed in a financially responsible and professionally managed way, and how can the process (e.g., design-to-construction handoff and field work) be improved—including by using AI? add

How do your agentic AI agents work, and how does your agentic strategy help align KPIs and coordinate demand for network build-out and power management? add

bolt Powered by CUBE AI

Stephen Rose, Render Networks

search

John Furrier

>> Welcome back everyone to theCUBE's live coverage here at Mobile World Congress. MWC 2026. I'm with Dave Vellante with theCUBE Research co-founder with me. Our next guest, Stephen Rose, CEO of Render Networks. Again, the beginning of a we believe in massive refresh happening in telecom, infrastructure, carriers, new business model opportunities as the AI becomes the platform and they're going to become AI operators. Stephen, welcome to theCUBE. Thanks for coming on.

Stephen Rose

>> Yeah, it's great to be here again. Great to see you guys.

John Furrier

>> So our thesis is that we're going to see a massive CapEx build out in edge, which is the telecom infrastructure and the carriers. They got the networks, they got the wireless, they got the wire line, they got the facilities. Kind of old school kind of voice optimized, going to data centric architectures. So it kind of points to the central factories being built in AI tokens to the edge. We see a tsunami of like, we got to refresh, upgrade our infrastructure.

Stephen Rose

>> Yeah.

John Furrier

>> What's your thoughts on that? How do you see that?

Stephen Rose

>> Well, I mean, if you just think about some of the incredible statistics that are out there right now. I mean, obviously, telecom is going to spend billions over the next few years. They've been thinking about atomizing the architecture over a number of years now and thinking about disaggregating that architecture and all of that needs supplying and feeding somehow. And on top of that, of course, you've got the AI and the data center demand that's going to be actually using that disaggregated architecture. But when I also think about it, I think the velocity of the decisions that have been making in the boardroom and the private equity firms on those data and AI centers are actually going to require telecom actually all infrastructure, whether it be telecom or whether it be the grid systems or whether it be water, all of it needs to be actually working in tandem. And at the moment, the decisions around data centers and AI, the velocity around those is way quicker than the infrastructure industries are able to respond.

John Furrier

>> Talk about Render, what you guys are doing, because you mentioned power, grid. You look at a lot of these edge devices, they're different. I mean, the footprints, diversities, the spectrum is huge. You could have not a lot of power, some power.

Stephen Rose

>> Yeah.

John Furrier

>> You got grid, you got how much energy there, upgrades. What do you guys do and how do you guys solve the problem?

Stephen Rose

>> Yeah. I mean, if you think about it, again, the problem that you see right now is that there is an enormous amount of friction just in simple things like the permitting process, which isn't a very sexy part of the industry, but it's very, very ... Sorry, a lot of the industry depends on getting those decisions right. So really what you need to do is you need to make the rest of the process as frictionless as you possibly can. And that means making sure that you take a design, you move it into the field as quickly as possible. You have buildable designs that actually can then go and be closed out as fast as possible. And you want to make sure that you do that because the value chain is actually very, very cash sensitive. Whether in the construction sector or with the operators, everybody all the way through that cycle is very, very cash sensitive. So if you get it wrong, you actually don't just hamstring the rollout of the network, but you actually hamstring the way that the value chain is able to operate. And we've seen businesses go out of business simply because they've got that wrong.

Dave Vellante

>> So talk more about that dynamic, because you've got a situation where operators are scrambling for capacity and they need, obviously they need connectivity, they need GPUs, they need water, they need power, and they're throwing money at the problem. I would imagine that there's a lot of capital that just sits there wasted. How can organizations and operators sort of better align and better utilize their capital? How should they think about that? Because I think they're thinking about it as one big chunk and an outcome, but that's not the right way to think about it, is

Stephen Rose

>> No, it isn't. And I think actually the decision for how you want to roll out these networks and are they being rolled out in a financially responsible way is very important. If I turned around it and I said to you, "Look, we're going to run our billing system on the back of an Excel spreadsheet, you'd think I was mad." But actually, we see many, many networks being rolled out or grid systems being rolled out on the back of an Excel spreadsheet. It almost seems impossible to believe that you'd have billions of dollars of capital being actually managed in that way. So I think there's a number of opportunities that we can look at. A, you can look at how quickly you can get the design into the construction phase in a seamless way because you want the closeout and you want the original design delta to be less than 5%, let's say. At the moment, you're probably talking about 50%. You want field teams to be able to do the work qualitatively and you want to make sure that they do it in such a way that they're not able to repeat the work. And you can use a lot of AI, whether it be visual inspection or whether it be audible signatures or be speech to text. There's a whole bunch of different types of AI that you can use concurrently to be able to create agentic systems. And those agentic systems are what inform the way the work gets done.

John Furrier

>> Talk about your methodology you guys have to help deploy efficiently the infrastructure build out, because you're seeing the central big data centers we read about in the news, Texas, or all of it, they're kind of behind schedule.

Stephen Rose

>> Yeah, yeah.

John Furrier

>> Can't find the labor. I'm looking at some of our research on theCUBE research. Let's just take San Jose, for instance, I won't say the name of the provider. In San Jose, they've got 95 gigawatts project, 85 projects queued for 95 gigawatts, moderate competition, some have high intense competition. There's so much work out there. That's just San Jose. I mean, North America is massive, billions of dollars. What's going to go faster? What's the bottleneck? Why is it taking so slow to upgrade these traditional voice optimized architectures to data centric? Can you share your thoughts on the state of the market?

Stephen Rose

>> Yeah. I mean, again, I think the real problem that you've got is that strategies are not necessarily consistent or cohesive between the interdependent sectors. So I think that's a big problem. And I'll give you a working example. I was at one of a similar event. I was at Metro Connect last week. I'm talking to a small little operator in Utah and what they have got is they've got a wonderful little network over there. It's been going great for a number of years and suddenly they're going to find that Meta and various other hyperscalers are going to be turning up on their doorstep with absolutely huge demands for connectivity and power. Nobody can meet that demand because they're not geared up for it. So that's where you see the problem.

John Furrier

>> cultures. Meta, Amazon, Google, coming to the-

Stephen Rose

>> Exactly....

John Furrier

>> small little-

Stephen Rose

>> And a small little municipal organization that's actually ... And actually, hasn't been under a lot of competitive dynamics over a number of years as well. So they're organizationally not set up. They're digitally not set up. Their field teams haven't worked to the competitive intensity that they've had .

John Furrier

>> They know their facility.

Stephen Rose

>> They know their facility.

John Furrier

>> And they have the power and the footprint.

Stephen Rose

>> Well, they haven't yet got the power. That's also part of the problem, but they have got the connectivity footprint. That's where they can get ahead, but I think it is actually trying to make sure that there is a link through the entire value chain, whether that be in the people that are actually allowing the permits to go through, or whether it be the organizations that are supplying the workforce, or whether it's the materials that are coming through, getting that value chain aligned. And then the key thing is once you've got all of those component elements there, there's an assembly of them. It is putting a workflow through them, an AI intensive workflow through them that is really able to make sure that things don't get repeated. Sorry, things is minimized in terms of wasted resource and time.

John Furrier

>> Yeah, Stephen, one of the things that's jumping out at me, just riffing in real time here in theCUBE is the demand for AI to be infused into the infrastructure, Meta, Nvidia, they're bringing AI and that culture into a kind of an old ... I won't say old school, but a slower moving velocity, but the build up pressure is massive.

Stephen Rose

>> It is.

John Furrier

>> What's the magic secret sauce to kind of make it go faster if you had to move it along like, "Okay, let's go faster."

Stephen Rose

>> I mean, there's a few things that I would do. I mean, I would really think about the contracts that I've actually got and think about the way that you want the behavior in those contracts to work. It sounds like an odd thing, but actually, punitive contracts don't really work. You really want people to work together. And actually, even between players in the construction industry, you may want a coopetition model. I think leadership and change management. If you don't have change managers and you don't have leaders that are really equipped for digital change, that's going to be a big problem, because you're going to get resistance from the bottom up in the organization. Field teams must adopt digital ways of working, otherwise they're going to be very slow.

John Furrier

>> Sorry, Dave. Do you see the monetization conversation kick up higher now that there's an enablement, accelerated enablement going on?

Stephen Rose

>> There is. And actually, what I'd love to see is I'd love to see a better understanding of how those commercial contracts would work. How would you actually incentivize an energies provider and a telco and somebody that's going to be using in the data center? How do you actually get those guys to work in concert? And what's the commercial construct that would actually get them to behave concurrently? I think that would be a really interesting opportunity for us to explore. Again, we don't typically think like that.

Dave Vellante

>> So explain, Stephen, more about your agents and your agentic strategy. There was an article in the Wall Street Journal yesterday or today how some major data centers in Virginia pulled off the grid and it almost shut the grid down, which is surprising, right? You think coming on the grid would be the problem, but actually, it's got to be a balance. It's okay. That's kind of amazing. So how do your agents work? What do they do? How do they help sort of align the demand for building out networks and managing the power and the like? Tell us all about that.

Stephen Rose

>> Again, the most important thing is agentic AI. The number one rule around agentic AI is you get an alignment between the KPI that you're looking and the data that you're using. And you want that actually fairly small and fairly contained, because that way you can control it, you can put guardrails on it. And then actually, the way we think about that is it's almost like, again, an atomic or an atomized version of LLMs. Yeah. So what you do is you take one small agent, you introduce that, you put guardrails around it, you make sure that it works, it's relative to very specific KPIs, then you're introduced in a second and a third. So the working example of that is the one that we've just announced, which is Clearway, is we have an assurance one, and the assurance one is effectively taking various types of inputs, whether it be visual inspection data or would be log data from a bore machine or whatever it happens to be. You bring those systems together and you allow the decision to allow the agent to make a decision. If it makes that decision and it's comfortable with that and it hits a certain tolerance, then you go into the approval. And the approval is effectively where the work gets completed against a commercial contract or against a defined scope of work, acceptance criteria, and then an invoice can get raised. Now, that sounds very trivial, but the one thing you don't want to hold up is an invoice in the system because then everybody keeps their businesses fluid. So those two agents will work together. And as you go further back into the life cycle, you'll see more agents, more autonomy, but controlled autonomy.

Dave Vellante

>> So I'm imagining you have sort of a top down series of KPIs and objectives, which what you described as their constraints for individual agents. I'm curious as to how the agents work together. So one agent might be to optimize the efficiency of X or whatever the spend around electricity. Another agent might have security built into it, et cetera. How do the agents work together to adjudicate according to those top down metrics? And where does the human fit into the loop?

Stephen Rose

>> Yeah. So the humans come into the loop immediately, as soon as the tolerance of risk is not acceptable. So if there is a certain type of work and it's not done to a particular standard, it drops below a certain tolerance, that's exactly where the anomaly comes in. When the anomaly comes in, that's when the human comes into the loop. And then there may be context for an approval or not approval, but there's going to be a green or a red light in that particular workflow. But the most important thing is that the system is able to work with agency within those tolerances. So again, further upstream, for example, if work is not being done, you might reroute resources. If you reroute resources and you don't need a human in the loop, then the beauty of that is actually you catch problems early. And I'll give you a working example. If you're doing a QA inspection, it might be six weeks after the work is done, before the QA is actually done by the acceptance company. Well, by that time, the crew is left and they're 60, 100 kilometers away, they're somewhere else. They've got to go back, they've got to do rework. It's a lot of wasted time. So that's how the agents work is that they are continuously assessing what is the next best action, and then they're giving that to leadership.

Dave Vellante

>> So the agents will take action on a low risk, high probability outcome, and they'll what, develop a plan or some alternatives that they present to the human on-

Stephen Rose

>> They'll present alternatives to the human if they're below a certain threshold. If they're within a certain threshold, they'll just take the next best action. So it's a very, very clever, very sophisticated system, but it's hyper controlled.

John Furrier

>> Stephen, great to have you on theCUBE. Again, the buildouts are happening. The agents are going to fill the void there for labor. Final question for you. Agents are ushering in the autonomy. I know it's bounded and there's some issues to manage there, but it's also ushering in the robotics era. So your thoughts on physical AI, because agents will just give us some workflow advantages. I'm imagining robotics will be a big part of the future.

Stephen Rose

>> Yeah. I mean, I think it's a massive potential. I mean, again, if you just look at the combination of robotic visual systems, the amount of command that you can actually do that. Actually, being able to use speech to text to be able to deal with those systems remotely, through your coordinated systems, there's still going to be a lot of humans in the loop, but there is a really exciting period of robotics ahead of us. And in the deployment of telco technologies and grid systems, I think that's a great start.

John Furrier

>> You're bullish in.

Stephen Rose

>> Oh yeah, absolutely.

John Furrier

>> Yeah, it's great. And hopefully go faster and get the build out going. Thanks for coming on. Appreciate it.

Stephen Rose

>> All right. Love it. Thanks again.

John Furrier

>> I'm John Furrier with Dave Vellante. Mobile World Congress, MWC 2026, live coverage with theCUBE. Check it out. We'll be right back after this short break. Wall to wall coverage three straight days. We'll be right back.