Red Hat Summit 2026 | Gunnar Hellekson, Red Hat

Clips
More from Red Hat Summit 2026

Gunnar Hellekson

Vice President and General Manager, Red Hat Enterprise Linux

Red Hat

play_circle_outline Linux/RHEL Powers AI Breakthroughs Through Ecosystem Integration and NVIDIA Accelerator Partnerships

play_circle_outline Red Hat Hardened Images GA: fast upstream runtimes packaged for RHEL compatibility

play_circle_outline Fedora Hummingbird: Fedora-based fast-moving OS, supported under RHEL subscriptions

play_circle_outline OS as AI Control Plane: OpenShell and NVIDIA Add SELinux, cgroups, and Resource Controls for Agent Workloads

Info
Transcript

Gunnar Hellekson, Red Hat

Gunnar Hellekson

Vice President and General Manager, Red Hat Enterprise Linux Red Hat

In this interview from Red Hat Summit 2026, Gunnar Hellekson, vice president and general manager for Red Hat Enterprise Linux at Red Hat, joins theCUBE's Rob Strechay and Rebecca Knight to discuss how Linux is evolving into the definitive foundation for enterprise AI innovation. Hellekson frames RHEL not simply as a server OS but as the natural collection point for the open-source AI ecosystem — where the most meaningful model and tooling contributions are happening. He explains how Red Hat's partnership with NVIDIA ensures support for the latest GPU accelera... Read more

explore Keep Exploring

What role does Linux—and Red Hat Enterprise Linux (RHEL) specifically—play in enabling and supporting modern AI innovation and workloads? add

What customer problems led to the development of RHEL Hardened Images? add

What is Fedora Hummingbird Linux, how does it relate to the Red Hat Hardened Images and Red Hat Enterprise Linux support, and who is it intended for? add

How should we think about AI relative to traditional IT workloads, and what role does the operating system play in supporting AI (security, resource management, and hardware access)? add

bolt Powered by CUBE AI

Gunnar Hellekson, Red Hat

search

Rebecca Knight

>> Hello, everyone, and welcome back. We are live in Atlanta at the Red Hat Summit here at the Georgia World Congress. I'm your host, Rebecca Knight, alongside Rob Strechay. It's the number one open-source show.

Rob Strechay

>> Yeah. I mean, again, outside of some of the foundations and stuff like that that Red Hat is a big contributor to, but when you start to look at it, it's so much fun seeing everybody, the enthusiasm about open and really how that has literally driven for decades the foundation of what we're talking about today.

Rebecca Knight

>> Indeed. Well, with that, I would like to welcome back to the show, Gunnar Hellekson, Vice President and General Manager Red Hat Enterprise Linux at Red Hat. Thanks so much for coming on.

Gunnar Hellekson

>> Thanks for having me. It's nice to be here.

Rebecca Knight

>> So there's a line in your messaging that every major AI breakthrough happens on Linux. For those who are still thinking of RHEL as server OS, how do you frame the way you think organizations should be thinking about it in 2026?

Gunnar Hellekson

>> So you said it great. Open-source and specifically Linux is the foundation for all AI innovation that's happening right now. And so, what that means for RHEL specifically is, first it means supporting the hardware that people need to run by AI. So our partnerships with NVIDIA, making sure that we are supporting the latest and greatest accelerators, this kind of thing. It also means that the entire ecosystem of AI software that's being developed is all out in open-source today. The most meaningful contributions are happening through open-source. And so, that makes the operating system a natural collection point for all of that innovation. And when we look at not just hosting AI applications, but actually including AI features in the operating system itself with tools like Lightspeed and the MCP servers that we've shipped, the idea here is that RHEL is both producing AI work and also the beneficiary of AI work. So yeah, Linux and RHEL specifically is kind of deeply integrated with the AI ecosystem.

Rob Strechay

>> So also kind of help us understand, because I think one of the big things with RHEL and Linux in general is that reliability and stability. I mean, I was a Unix admin coming out of college. And then I got into Linux and I'm like, "Wow, this is just awesome about how fast it moves and the community around it." But really, it was, why'd you go down that path? Because reliability. So kind of talk to that a little bit and how that's playing a role.

Gunnar Hellekson

>> Yeah. Well, I mean, certainly that's what distinguished Linux from other competitors and also what distinguishes REL from other Linuxes is our commitment to stability. It's kind of our hallmark. We have 10-year life cycles, now 14-year life cycles. We do reliable releases every three years and minor releases every six months. And we really kind of made our bones imposing order on the chaos and innovation that's going on upstream. We make it consumable by enterprises and be able to run workloads like SAP HANA, Oracle databases, and so on. That's where we earned our reputation, right? But now that's starting to mean something different too, because we're seeing enterprises now operate, yes, in the kind of traditional reliability, kind of 10-year release cycle, 15-year release cycle, but we're also seeing them move much faster and being able to actually do both things at the same time. In some cases, you want to move as fast as humanly possible because you're trying to keep up with CBEs, you're trying to keep up with innovation and AI. And in other cases you are slowing down and like, "We don't want to mess with this thing any more than we have to, right?"

Rebecca Knight

>> So it's that fast, slow mode.

Gunnar Hellekson

>> That's right. Yeah. Yeah. And both simultaneously. It's often within the same project, right?

Rebecca Knight

>> Right, right. Okay. So talk a little bit about RHEL Hardened Images which just hit general availability. What were the kinds of problems that customers were bringing to you that made you say, "Okay, this is what we need to build now."?

Gunnar Hellekson

>> Yeah, just that. So Red Hat Hardened Images is a set of fast moving run times. So we take what we can from the upstream communities, package it up as fast as possible and produce a RHEL-compatible version of that thing. We've been doing work like this for decades, but we do it in kind of the slow RHEL mode, where we have a release every six months and then we harden it and we backport security fixes and all the rest of it. With Red Hat Hardened Images now, you are basically plugged right into upstream and as soon as they fix something, you get the fix as well. Now, not everybody can consume this, because it means updating the images every day if you want to stay current. So if you're willing to go on the ride, we can offer you that ride. If you're willing to go on the traditional enterprise Linux very predictable release cycle, you can go on that ride too.

Rob Strechay

>> Do you see people looking at it because the regulatory environment is always changing? I mean, you got FIPS, you've got a whole number of things and the regulations definitely vary. Your mileage will definitely be adjusted depending on where you live and the regulations in your neighborhood. So kind of talk to that a little bit.

Gunnar Hellekson

>> Yeah, sure. So the challenge is, especially say you're a bank, you will have workloads that you need as stable and reliable as possible. And as I said before, I don't want to touch them for 15 years. And then you're going to have other workloads that are moving fast. Maybe you're doing an AI project, but you still need things like FIPS compliance to make sure your encryption is working the way that it's supposed to. And so, with Red Hat Hardened Images, we actually offer FIPS versions of these containers as well. So you can actually operate either in fast mode or in slow mode, but still get the benefits of all the regulatory compliance that you need.

Rebecca Knight

>> You're talking about shifting security earlier in the development process. So what does that really change for a developer in practice?

Gunnar Hellekson

>> Yeah. So there has been a lot of conversation about shifting left in the development process, so moving all the security. You don't want to do the security work after you build the thing, you want to move it left in the process so that you're doing it before you've actually built and deployed the item. So Red Hat Hardened Images make it easy to do that. But shifting left is, we used to talk about it in terms of humans, but now especially as more and more developers are using AI agents to do their development, you want to make sure that those AI agents are actually using good ingredients for the application. And so, what I would recommend everybody do, of course, is to go tell your Claude code or your Cursor or what have you to make sure that you're using a Red Hat Hardened Image when you have the opportunity so that you start with a fixed, compliant, secured, supportable runtime on which you can build your application.

Rob Strechay

>> How do you see organizations that may be down the path already, like how would they say, "Okay, I want to go to that direction."? How easy is it for them to say, "Hey, I want to go towards the hardened image direction so that I know..." And to your point, because the shift left put a lot on developers, but this takes it off their plate as well.

Gunnar Hellekson

>> That's right.

Rob Strechay

>> Well, hopefully. Yeah, that's the design, right? Yeah. And well, the good news is that it is actually very easy to start with, because those hardened images are free and downloadable today. So you don't need a note from your doctor in order to get access to them, you can just start using them. And we imagine that most development shops are going to be using some blend of the two. So you're going to have one application or one dependency that's going to require some long-term support. Other use cases, other portions of the application are going to be able to move quicker. But in either case, you want to be able to produce the regulatory documentation you need. You need to be able to produce the FIPS credentials that you need. And so, we're making it as easy as possible to operate in both modes at the same time.

Rebecca Knight

>> In terms of the lifelong add-on, there's this tension that exists in IT of, "Modernize everything," versus, "But that works. Let's not touch it." How does this speak to that tension and allow CIOs to hold those truths at once?

Gunnar Hellekson

>> So I mentioned before RHEL's extraordinary lifecycle, right? 14 years is several eternities in IT world, right?

Rebecca Knight

>> Indeed.

Gunnar Hellekson

>> But even that is actually not enough. And especially now that we see customers are suffering from the disruptions in the hardware supply chain, it's getting increasingly difficult to get new servers, which means it's increasingly difficult to take your existing application and move it to a new platform, which means you're stuck, you have to stay where you are. And so, this long-life offering, now for a functionally unlimited amount of time, you can just keep extending another year and another year and another year and buy yourself some time on a known good system until you can get your hands on the new hardware that you want and then you can perform your migration. So it's an opportunity to help customers better align their hardware lifecycle with their application life cycle and hopefully make people's life a little bit easier.

Rob Strechay

>> So let's kind of change gears a little bit and talk Fedora, because I think it's really... I thought this was one of the almost least hyped announcements when I was getting some of the previews, but it has a significant play, especially with AI. So kind of go through what the announcement is and...

Gunnar Hellekson

>> So we were doing the work on the Red Hat Hardened Images and we were pretty far down that project and then we realized, "Oh, well, wait a minute, we should be able to do exactly what we did for Red Hat Hardened Images, we should be able to do it for an entire operating system." And so, we partnered with our friends in the Fedora community and developed a Fedora Hummingbird Linux. And you'll see several of us walking around with the Hummingbird pins at the show today.

Rebecca Knight

>> All right.

Gunnar Hellekson

>> So Hummingbird Linux is exactly what Hardened Images is, it's a fast-moving, stable release that is developed in the Fedora communities, which means it's free and freely redistributable, et cetera. But if you have a Red Hat Enterprise Linux subscription, we will support it, which means that now that idea of the fast and the slow extends from the application layer all the way down to the complete operating system. Who I expect to use this is humans, of course, as I said before, these developers who are trying to maybe move quickly on a project, but also this is extremely well-suited to a very fast-moving space like AI where we have, you need new hardware innovation as soon as humanly possible and you want to be able to get that hardware innovation in there so that you can take advantage of it in your AI application. Hummingbird is a great delivery vehicle for that.

Rebecca Knight

>> One of the things that really struck me about Matt Hicks' talk up on the main stage was talking about managers and how they'll spend their time and the metrics that they'll be judged by for their leadership in terms of how they delegate to the humans and to the agents. As a leader, how do you think about that and how has it changed your workflow?

Gunnar Hellekson

>> Yeah. Oh, well, we've only got about five minutes, right?

Rebecca Knight

>> Unfortunately, yes.

Gunnar Hellekson

>> So that shift left that we talked about in the development cycle, I have found that to also be true for all work when it comes to AI. It is shifting your response to everybody kind of takes one step to the left. Engineers are now able to do work that was previously product management work. Product managers are not able to do work that used to be technical marketing work, techno work, and so on and so on and so on. So I'm watching all of the roles in my organization change. And what I've found is that the most important thing you can do to enable that change is to ensure that everyone is working from a set of ground truths. If you have not written down your strategy, you are going to have a very difficult time having that strategy conducted through your agents. If you haven't written down your operational plan, you need to expose your org chart. All that rich context that all of us hold in our heads has to be available to that agent so that it can reason correctly about the problems that you give it.

Rebecca Knight

>> AI is very literal.

Gunnar Hellekson

>> It's very literal. And so, what I found is in my organization, the most important thing we could do is make sure that we have all the right processes and make sure that all the right documents are in place so that our agents can actually be helpful instead of wrestling with us, right?

Rob Strechay

>> So when you look at that and you're looking at development, because you guys not only are building the OS, but at the same time you're building things on top of that, how should organizations feel about the OS and how the infrastructure layers, because again, they're not necessarily as oriented towards, like you said, a TPU or the stuff from NVIDIA or AMD or from Amazon or what have you, how should they feel about, "Hey, we're looking at this and we need flexibility within this."?

Gunnar Hellekson

>> Yeah, that's a great question. So people talk about AI as if it is some special case in the IT industry. And the truth is that AI is still a workload, it is still software, it still requires security controls, it still requires resource management. It still requires, for us, it's SELinux and cgroups and all these exotic controls that we put on a regular application, AI needs all that stuff too. And the application also needs access to hardware. And where we used to talk about CPUs, now we're talking about GPUs and TPUs and all the rest of it. But fundamentally, these are all the same problems. And so, the OS is a natural collection point for all those features. So we just announced OpenShell, for example, with NVIDIA, right? The controls that OpenShell puts in place for your OpenClaw or your agents, all those controls are fundamentally operating system controls. And so, when people make a choice about the operating system for their regular applications, they are also making a choice for where their AI stuff is going to get hosted. I think with that lens, I'm actually very excited for where the operating system space is going to go, because I think we've spent 20 years getting ready for the AI revolution and now it's here.

Rob Strechay

>> Yeah. And just to follow up on that, that also has a play in the sovereignty aspect as well, because people are worried about what data from where, and that's something where normally the OS controls that. So is that what you see as organizations are leaning in and saying, "Yeah, I want to control it from the bottom all the way up to the top."?

Gunnar Hellekson

>> Yeah. Well, certainly they want more visibility, more transparency, more measurability, observability on their entire stack all the way down to the operating system and even to the hardware. We got some very exciting work going on in confidential computing, for example, where we're able to encrypt containers and encrypt virtual machines, so that even if you don't necessarily trust the person who's hosting your workload, you can still be sure that they haven't tampered with or looked inside the workload. This is great for folks who are worried about sovereignty. And then, of course, there's the fact that we're developing an open-source in the first place, which is actually a comfort to a lot of customers. They know that if it has been developed in the open-source way and are getting it from a vendor that they can trust, that allays their fears they might have about sovereignty concerns.

Rob Strechay

>> And just to hit on that, because it was a question that was asked to me before the conference, was the whole confidential computing, you guys announced that a couple, I think it was two years back or something like that. But that's also helping secure the memory space and all that, which is really important to the AI.

Gunnar Hellekson

>> Yeah, that's right. And again, all fundamentally operating system work. I mean, there's lots of management tools that need to go around it, but fundamentally these are all operating system facilities that are making this possible. Yeah.

Rob Strechay

>> Awesome.

Rebecca Knight

>> If a developer looks back five years from now and considers 2026 a real turning point for Linux, what do you hope the reason will be for that?

Gunnar Hellekson

>> I will say we have never had so many tools and so many resources available to make us successful. I have a sneaking suspicion that AI will never be as available or as cheap as it is right now, and that has unlocked an unbelievable amount of innovation. I felt it myself, you probably have too, I see it in my teams, I see it in the products that we're shipping. They're going to look back and see that, "Oh, this is the moment when the cadence permanently changed for the industry." I certainly feel that every day.

Rebecca Knight

>> Gunnar, a pleasure having you on the show as always.

Gunnar Hellekson

>> Oh, thank you very much. It was great to be here.

Rebecca Knight

>> I'm Rebecca Knight for Rob Strechay, stay tuned for more of theCUBE's live coverage of the Red Hat Summit. You're watching theCUBE, the leader in enterprise tech news and analysis.