MWC Barcelona 2026 | Paul Miller, Wind River

Clips
More from MWC Barcelona 2026

Paul Miller

CTO

Wind River

play_circle_outline Wind River at MWC 2026: Kubernetes-Powered vRAN/OpenRAN for Disaggregated Cloud-Native RAN

play_circle_outline Production-scale vRAN/OpenRAN deployments with Verizon, Vodafone, Telus, Japan

play_circle_outline Open Interfaces Drive Vendor Competition and Lower TCO: Wind River Optimizes Manageability and Operability for Service Providers

Info
Transcript

Paul Miller, Wind River

Paul Miller

CTO Wind River

In this interview from MWC Barcelona 2026, Paul Miller, chief technology officer of Wind River Systems, joins theCUBE's Dave Vellante to discuss how open radio access networks and edge AI are converging to reshape telecommunications and beyond. Miller details Wind River's rise to become the leading provider of cloud infrastructure for vRAN and OpenRAN, with production deployments spanning Verizon's tens of thousands of nodes in North America, Vodafone in Europe and Telus's first O-RAN rollout in Canada. The resulting business model shift is significant – serv... Read more

explore Keep Exploring

What is Wind River's role in the move to cloud-based vRAN/OpenRAN and 5G, and what are you showcasing at this event? add

What is the deployment status of vRAN and OpenRAN — how do they differ, who are the customers, and where have they been deployed? add

How does the move to OpenRAN/vRAN change the commercial business model and competitive landscape for network operators, and how can a company like Wind River differentiate itself in that environment? add

bolt Powered by CUBE AI

Paul Miller, Wind River

search

Dave Vellante

>> Hi everybody. We're here at Mobile World Congress inside the Wind River booth and stage, as they call it, here at Mobile World Congress, MWC 2026. My name is Dave Vellante. I'm here with Paul Miller, who's the chief technology officer of Wind River. Paul, good to see you.

>> Yeah, thanks for having me.

Dave Vellante

>> Thanks for making some time and thanks for hosting us.

>> Absolutely.

Dave Vellante

>> You got a great setup here in your stand. I call it a stage. They call it a stand. That's the sort of Barcelonan language, but you guys got a lot going on at the show. Tell me what it's been like for you, what you guys are doing here, what you're showing off.

>> It's actually been incredible. We've been coming year after year as this market has gone through a pretty significant evolution as they've moved to cloud-based technology in the core of the network and in the edge of the network. And Wind River's really made a name for itself in the category of vRAN and OpenRAN. So as a lot of service providers globally, companies like Verizon and Vodafone and others are deploying 5G. They're actually running that on Wind River infrastructure. And we provide a cloud technology that allows them in the radio access network to build and deploy Kubernetes-based RAN applications. And so a lot of the customers we meet with here are service providers interested in that type of stack. And it is the same stack that we're bringing into other market verticals, industrial, enterprise, medical, and others. Wind River is a company that works in many different verticals, even automotive.

Dave Vellante

>> So I recall being here in 2022, 2023, it was post-COVID, and the O-RAN discussion was really kicking in. And the narrative at the time was it's got great potential. Obviously, we need to open up our static infrastructure, but it's not ready. Now, my understanding is that your deployments are one of the few that are in production at large scale. Is that correct?

>> Yes, that's correct. So we've been in deployment for the better part of five years now. And in fact, our first and largest customer, Verizon, has tens of thousands of nodes running virtual RAN. And there's really two technologies there. One is vRAN, where you have the cloud-based disaggregation of the functions. You move from proprietary hardware to conventional servers with cloud technology with applications. And that was vRAN, and that's been deployed and proven for years now. And then OpenRAN emerged as a byproduct of that. It's kind of on the way to the AI RAN that's happening at this show. And when OpenRAN came out, that was really about open interfaces between the vendors so that you could interoperate different components that classically didn't used to interoperate. Again, running on an infrastructure like Wind River provides. And so now, we are globally deployed in North America. We're doing the first O-RAN deployment with Telus in Canada, deployed in Europe with Vodafone and in Japan and many other countries. So it's really taken off and it is in full production scale now.

Dave Vellante

>> Yeah. So the reliability is proven. And explain the real benefits that you're bringing to this industry, because for years, it was proprietary systems, very closed, really hard to change. O-RAN changes that, especially when you bring in Kubernetes and all the development capabilities on top of that. But can you explain that?

>> Yeah, that's absolutely correct. There's kind of two pieces to that. The first piece is really commercial, where in the legacy approach, you'd have competition at the service provider for their business at the beginning of the network build. And then a classical vendor, a telecom equipment manufacturer typically would get selected and build the network. And for the next 10 or 15 years, that service provider is tied to that vendor. Now, you have a massive change in the business model. With OpenRAN and vRAN, now you've got hardware servers from multiple vendors, vendors like Dell and HPE. And if one of those vendors disappoints, the customer can switch to the other three, four years into the deployment. If I disappoint, they can switch to my competitor. If the application disappoints, they can push a button and re-orchestrate a new application. So now, we've moved to a business model where this competition for the life of the network, that drives TCO down for the operator. It's a massive business transformation. The second piece of it, at least for us is then, all right, and now what I've just described as a highly competitive landscape. How do I differentiate myself? And we've done a lot of things to optimize the performance of the system and make it manageable and operable for the service provider. And that's where Wind River has come from a relative unknown to be the number one provider of that technology.

Dave Vellante

>> Interesting. So it's like open systems comes to the telco industry, but the risk there is obviously you get no differentiation, but you're bringing in value on top of that.

>> That's right. That's right.

Dave Vellante

>> Okay. I want to ask you something. Let's kind of move to the shift from data center AI to edge. Something I was reading, edge AI represents a fundamental shift from data center AI. It's not simply cloud AI that's closer to the user, but it's a different class of system altogether. What does that actually mean?

>> Yeah, there's a couple of things. The first is you want to think of generative AI and digital AI. When you'd normally, and many of us have used things like ChatGPT where you type in a question, it gives you an answer, that's off a static model. As you move more towards the edge, you're moving more towards systems that interact with the physical world. And when you interact with the physical world, edge AI and physical AI start to emerge. Instead of training large language model based functions, it's now inference. It's the execution of an AI model in a way that interacts with a human being. So for example, our parent company is an industry leader in self-driving cars. They use AI with radar and camera processing to make decisions about where objects are and drive the car. We work with industrial manufacturing and robotics, which is using AI camera recognition to move things on assembly lines. So as you get into the physical world where you're sensing something in the physical world and then taking an action in the physical world, you're now in the land of edge AI.

Dave Vellante

>> So you guys have this kind of cool vehicle down here.

>> Yes.

Dave Vellante

>> It's got an active license plate and of course Wind River is the software layer on top of that, connecting all these sensors and devices. One of the first autonomous vehicle interviews I ever did, I texted a friend of mine who's an expert in the field and I knew nothing about it. And he said, "Ask him about Byzantine fault tolerance." Which is this concept of in a military concept, if the general says, "This is what we're doing," and you're in the fog of war and things change, it's very hard to communicate. But in an autonomous vehicle situation, you have to have that ability to communicate across these disparate vehicles. Is that something that you guys can actually enable to drive safety from a technical standpoint?

>> Actually, that's exactly what the demonstration's about. So you think about it over time, the automobile has moved into now, as we sit here in 2026, a software defined vehicle. They're self-driving. They have radar and camera image processing. They have all kinds of software for infotainment in the vehicle. There's a huge amount of software in the vehicle. It's become a computer on wheels. And so naturally, you think once it's a computer on wheels, these computers interact with each other. And historically up to date, these self-driving capabilities and safety functions have really been thought of with the vehicle by itself. But what if we could bring awareness of the environment from other vehicles to that vehicle? And that's what we're showing here today. It's called Cellular V2X or cellular vehicle to anything. We have one car that's sitting there sensing the environments around it, all the people walking by in the show, sending that data up to the Verizon 5G network through a Verizon ETX service, and then sending that down to a destination vehicle. That destination vehicle, it may be behind a building or behind a wall, and now it can see the person walking on the other side because it's getting that information from the first vehicle. So effectively, we're sharing sensor data and that destination car sees that data coming in real-time because of the performance of the 5G network as if it were just new sensors that it had. So it can see beyond its horizon now. And now in the demonstration we give, instead of hitting a pedestrian, it'll actually break before it even sees the pedestrian, because it's getting the information from another car. So safety, comfort, and convenience, there's hundreds of applications that leverage this technology and we're at the forefront of it.

Dave Vellante

>> This is exciting, because the promise of autonomous vehicles is it will be much, much safer than human vehicles. I know, for instance, with a lot of drivers, sometimes you see red lights up ahead and people don't slow down. They keep going. What you just described is the system would sense that even maybe before it's visual. So that's very powerful. From a technologist perspective, what are the constraints that you have to deal with at the edge, whether it's connectivity or latency or determinism? Can you take us through that?

>> The primary one is latency, right? Because obviously, if you think about that safety critical application we just gave an example of, you can't take three seconds for the traffic to go up to a public cloud and back. So you have to deploy the application on the edge of the network. So in the case of that Verizon example I just gave, in their multi-access edge complete, their edge cloud, we deploy the application so that it has a low latency connection to the vehicle in tens of milliseconds. This allows the data to transit the network and enter the second vehicle in time to affect its decisions. If you can't achieve that latency and performance, you can't build those types of applications. So from a technology perspective and safety critical applications, low latency is very important.

Dave Vellante

>> I want to ask you about reasoning and how much reasoning is going on at the edge. We're learning as users of LLMs about reasoning, and reasoning takes time, but do you reason at the edge? How do you reason at the edge and how do you do so in real time?

>> Yes. And so I'll give one example. In the automobile that we're just talking about here today, there's a concept called sensor fusion. And what's happening is all the cameras in the car, the radars, the LiDARs, all the things that are bringing environmental data in are fused in the computer into a single view of the world. That's called sensor fusion. It then uses that as an AI to recognize object. Is that a car? Is that a person? Which direction are they walking? Is that a cat walking across the road? At what velocity is it moving? How do I avoid it? And it's making AI and machine learning based decisions about how to take actions to avoid that. Now, what we're doing is connecting edge AI across systems through the network. So now, you're collaborating between systems operating in this AI manner.

Dave Vellante

>> Okay. But you have to do that in real time and you don't have the time and you watch the LLM say, "Well, I'm now thinking more and I'm going back and I'm checking my math, et cetera." You have to do that in real time. And what's the secret sauce? Is it because you are able to build that world model, but there's also some technology behind that.

>> Well, and that's actually Wind River's original reason for existence. So we've been in business for 45 years, purely as a software company. The primary product that we built there was called VxWorks. It's a real-time operating system. And so the capabilities in that real-time operating system enable people to build what we call deterministic systems with deterministic behavior. So the exact problem that you're describing is really, I need to perform some computational problem within a fixed period of time or it's not going to be sufficient for the application. That requires determination and high performance. And that determinism and high performance can only be realized if running on a real-time operating system.

Dave Vellante

>> So we talked about autonomous vehicles. What are some of the other use cases? I'm sure there are military use cases there. I'm sure there are factory use cases. Take us through some of those.

>> Yeah. Industrial manufacturing, we have a robot downstairs that's doing NVIDIA-based image recognition and moving objects, medical device systems, remote operating systems, aerospace systems. They're all using these technologies. We're seeing AI enter all of these edge systems today. And we really see this happening through a model we call sense, think, act, where you've got sensor input, the thinking process, and then an action taking place.

Dave Vellante

>> What about the economics at the edge? Inference, you talk about, do people debate, do we need GPUs at the edge? What's the state of the economics, the tokenomics some people like to say? And project out, put on your CTO hat and tell us where it's headed.

>> Well, I think we're coming from a place where AI was fairly nascent, right? And the technology and platforms from companies like NVIDIA were very costly. They're still costly and they consume a decent amount of power. Now, as you move towards actually solving the customer problem, you validate whether there's a business model there that can accept those things. So for example, if we look in the core of the network, we have another demonstration running downstairs where we're using agentic AI to help automate environments. That takes weeks of work and turns it into a minute of a question. So the funding for the computational infrastructure and the AI software is coming from the operational cost savings from removing that timeframe and the work and labor from individuals. So there are ways to fund and monetize these things so long as you're solving an actual customer problem.

Dave Vellante

>> So I know you're super busy, you got to run, but give us the one takeaway that you want people to remember from MWC 2026 about Wind River.

>> Well, certainly, I think edge AI is here. We're seeing it in every market that we participate in, whether it be medical, industrial, telecommunications. And we're going to take our industry leadership of OpenRAN and as that moves to AI RAN, continue that dominance.

Dave Vellante

>> So Paul, we've talked a lot about the telco and the operator space. My understanding is you've got momentum in the enterprise. Can you tell us about that?

>> Yes. We've had quite a few wins recently over the past year. We've got a significant agreement signed with ServiceNow where we're bringing IT and enterprise functionality into our cloud technology. We recently did a VMware migration for over 20,000 servers from VMware to our platform in just a few weeks. And so we're really seeing some exciting opportunity to bring our cloud technology from where it was built in the telco domain as a hardened platform that's now mission-critical and proven into the enterprise sector and helping support those businesses.

Dave Vellante

>> I know there was a lot of enthusiasm for a while around network function virtualization, which never really happened. It didn't really affect the application, but it sounds like you guys are broadening that scope and making it much easier for enterprises to take advantage of your technology.

>> Yes. We see that Kubernetes technology and virtualization technology is equally applicable to other vertical markets, industrial, manufacturing, medical, automotive. They're all interested in the technology.

Dave Vellante

>> Paul Miller, thanks so much for coming on theCUBE. Really appreciate it.

>> Thanks.

Dave Vellante

>> And thanks for letting us use your space.

>> Thank you.

Dave Vellante

>> And thank you for watching Wind River. You may not see them every day, but you touch them without even knowing it. Thanks for watching. This is Dave Vellante for theCUBE.