Red Hat Summit 2026 | Mike Ferris, Red Hat

Clips
More from Red Hat Summit 2026

Mike Ferris

Senior Vice President, Chief Operating Officer and Chief Strategy Officer

Red Hat

play_circle_outline AI-Driven Strategy and Token Economics: Transforming Organizations into Internal Token Providers

play_circle_outline Red Hat Enables Choice: Open Hybrid Cloud to Operationalize Secure, Multi‑Cloud AI with Accelerators

play_circle_outline Business-as-Code Meets ROAI: Achieving Measurable Productivity Through Policy-as-Code and CI/CD Governance

play_circle_outline Sovereign AI and compliance: governance, auditing tens of thousands of agents

Info
Transcript

Mike Ferris, Red Hat

Mike Ferris

Senior Vice President, Chief Operating Officer and Chief Strategy Officer Red Hat

In this interview from Red Hat Summit 2026, Mike Ferris, senior vice president, chief operating officer and chief strategy officer of Red Hat, joins theCUBE's Rob Strechay and Rebecca Knight to discuss how Red Hat is acting as its own customer zero in the enterprise AI era, deploying its open hybrid cloud platform to build a repeatable blueprint for agentic AI transformation. Ferris outlines Red Hat's AI pyramid — foundational models at the base, skills development in the middle and agent networks at the top — explaining how the company moved from rolling out... Read more

explore Keep Exploring

How should organizations adapt their operations and strategy to align with emerging technical capabilities and evolving ecosystems—such as adopting token economics and shifting from being token consumers to token providers? add

How has Red Hat's emphasis on choice — from open source and hybrid cloud to supporting various hardware, accelerators, and deployment agents — played out in practice for your customers and for Red Hat itself? add

How can organizations make documents, policies, and procedures "AI-ready" (e.g., via a "business as code" approach), and what impact will that have on operations, governance, and measuring return on AI? add

How are organizations addressing government and industry pressures for control, data sovereignty, and choice over where and how AI models and resources are run and governed? add

bolt Powered by CUBE AI

Mike Ferris, Red Hat

search

Rebecca Knight

>> Hello everyone and welcome back to theCUBE's live coverage of the Red Hat Summit here in Atlanta, Georgia. I'm your host, Rebecca Knight alongside Rob Strechay. We are both fresh from the main stage. We heard Matt Hicks, we heard the CTO, we heard the who's who.

Rob Strechay

>> Ashesh and Chris.

Rebecca Knight

>> Exactly.

Rob Strechay

>> That CTO and CPO with Ashesh. It was fantastic. I think the message is just so strong about how AI is really changing organizations from a business perspective, not just from a technology perspective. I think we're going to have a great conversation now about that.

Rebecca Knight

>> And Red Hat is doing it alongside the organizations that it's working with. And indeed, we are going to have a great conversation about that with Mike Ferris, Senior Vice President and Chief Operating Officer and Chief Strategy Officer at Red Hat. Thank you so much for coming on the show.

Mike Ferris

>> Happy to be here. This is awesome.

Rebecca Knight

>> You're a CUBE veteran.

Mike Ferris

>> I am. I am.

Rebecca Knight

>> And you have some of the biggest jobs at Red Hat rolled into one. You're also the person that is tasked with making sure that the announcements that we heard up there on the main stage translate into real customer adoption. Talk a little bit about some of the conversations that you're having with customers here before we dig into those announcements in terms of their pain points.

Mike Ferris

>> Absolutely. And I think, as you said, my role is unique, right? Chief strategy and chief operations. I think that is certainly systemic and evidence of what's happening in the industry. And I'm hearing that from customers I talk to in that you do have to align not just the operational aspects of what you do and make sure those are efficient, but you also have to look forward to the future and start saying, "How do we change what we're doing in a way that will make the organizations function in line with the technical capabilities, but also more importantly, how the ecosystems, how partners, how your customers evolve?" And so that's a big part of my job as well. So I'm seeing that happen universally. And as we heard this morning in the keynotes where Matt was up on stage talking about what we're doing internally, we're seeing a lot of this focus as we heard token economics being a key part of this. How do you actually get the most efficient use and turn from being just a consumer of tokens to actually a provider of them to your organization? And that's something that we've done inside Red Hat and we've made that transition and it's starting to really show how we can elevate the value and the capabilities that we have across the entire company.

Rob Strechay

>> Yeah. I mean, again, Matt talked about the NVIDIA stack that you're working with as well as some of the Granite models and model optionality. And a lot of the theme carried through all of the main stage keynotes today was really choice and being able to have optionality. I mean, I've always thought this, I think that Red Hat being an open source company first and upstream and really working through that and with the standards, bodies and stuff like that really provides that choice and has been a big ... Talk to how that's really been in effect for not only your customers but yourself as well.

Mike Ferris

>> It's been awesome. So I've been with Red Hat a very long time and this is how we founded essentially their enterprise message was really all about choice. It was focused on choice of open source. You had the choice to continue using the technology that you purchased indefinitely. That still applies, but it's been awesome watching that mature over the decades now to where we see similar problems repeat. We saw it with the public clouds where, absolutely, as I think Chris or Ashesh mentioned this morning, you could go to the public clouds, forget AI, but you could go to the public clouds and you had the option of using them, but you were limited in what you could do because you're really constrained. And so hybrid cloud really became the default behavior and we helped innovate, helped enable that with our platforms by focusing on ensuring any hardware was available for any application. Now, we're seeing the same thing happen as well where models really kind of drove the AI revolution now, but we're also seeing how do you get inference, how do you get the capabilities out to your organizations in the most efficient way possible, but also allow them to take advantage of the latest technologies. And so certainly when I talk to customers and when I look internally at Red Hat, it's all around how do we give the tools to our associates, the customers, give it to their customers, how do we make that accessible so we can get that level of innovation while allowing them to choose what accelerators they run on, what clouds potentially they run on, what agents they run on those. And so when we talk about things like metal to agent, it's about allowing any agent on any piece of hardware available using our common platform across those in a secure, governed and measured way. And that's really what we're doing internally as well.

Rebecca Knight

>> And as you've said, Red Hat really is customer zero here. One of the other things that you talk internally about is this AI pyramid with foundational models at the base, your own tuning in the middle and then the agents at the top. When was the moment for you when this, "This is what clicked. Okay, this is how we're going to go forward here?"

Mike Ferris

>> Yeah, it's been pretty amazing, because one of the first things we did is we rolled out Gemini from Google to every person in the company. We were a Google shop, have been for many years and it was a natural evolution to say everybody's using Gmail, using Google Docs to communicate, let's add a layer of AI on top of it. Great partner to start the evolution with. What that really started to engender across the organization was what I consider that bottom layer of the pyramid where everyone was starting to engage and create their own scripts, start to look at how we can use different tools. And it really created this upswell of integration across the company, but more at a personal or departmental layer. What we then said is, "Look, these are things that are starting to create needs. How do we standardize those? How do we make a structured way of creating agents, governing them appropriately, using our own tools, specifically Red Hat AI Enterprise, to be able to provide an infrastructure that allows everyone to connect?" That required a couple of things. Down below the pyramid, how do we get all our data together? How do we get common workflows and common structures around what types of skills do people need to develop in this? And so that was really the productivity gains that we started to focus on in that middle layer of the pyramid. What Matt Hicks spoke about this morning in the keynote was really around how do we get all this information together in a common language and a common way to access it. And so that top of the pyramid, which we really consider, how do you actually find information and use it better is where we started to look across the organization and ask everyone, create agents that provide access to your information. We did that across structured and unstructured data in the organization. So now, as an executive, I can sit down with my identity, ask our agents what is happening in a specific region with a specific product in specific Jira tickets across the entire organization and get that information almost instantaneously, but then combine it with other levels of information in the organization. That in turn then spawns other dialogues and other individuals in the company can do that. But as we talk about this pyramid, it's really around connecting individual productivity with departmental, with cross company to really say all of these things work together against a common set of data on a common infrastructure that is what we're doing with Red Hat AIE.

Rob Strechay

>> Yeah, I totally agree. And I think to me it's so much not about the models anymore. I mean, the models are important, but most companies aren't building their own foundational models. I mean, that's just not where they're going. They're taking ones and they may even look to make them smaller and be able to make them more functional by doing fine-tuning or what have you. I know there's a lot of technologies here with vLLM and llm-d and some of the other stuff for distributing those models. Where do you see the organizations you guys are interacting with from a model size perspective? Because, obviously, that has impact on costs and things of that nature and where they can run. How do you see the kind of juxtaposition between big models and smaller specific language models?

Mike Ferris

>> Yeah. So first, I think organizations, we're going through it still, but we got a good head start on it. All organizations are coming to grips with what are the value streams, the workflows, the skills that I really need to understand to run my business effectively. That in turn says which models are most effective in those types of environments. And so if we start talking about things like document understanding, small language models work very, very well for that. How do you actually aggregate information, summarization? They also can work across that. Planning may take some larger models depending upon how they talk to all the agents in your organization, but specific to the tasks, how you build this infrastructure really will depend upon what types of skills, what types of needs you have in your work. That said, I think there's some great fundamental ones. NVIDIA released a blueprint called AIQ several months ago. We had already implemented it, but that said, how do you actually use Nemotron? How can you use, in our case, Granite models for certain aspects of this? And through our learning through this and the exploration and frankly doing multiple evaluations against these infrastructures that we built, we started to get native understanding of how we can actually deploy the right models for the right workloads, including things like ensuring our core data doesn't get moved out to a frontier model's infrastructure, even with appropriate contracts. We're protecting that along the way. Again, using these models as a service infrastructure components that we have inside our own platforms.

Rebecca Knight

>> Can you talk a little bit about that internal experimentation that you're doing and the fact that, as you say, you're learning as you go too and sometimes things might not work as you expect them to, but talk a little bit about what you've experienced.

Mike Ferris

>> So a couple of things, right? First, we had to get everyone in the company to get an intuitive understanding of AI, and that's where I'll point to our use of Gemini to get things going, to really get people familiar with their own data and the capabilities that would come from that. Secondly, it was also motivation around exploration and learning within sandboxed environments. And so a lot of this was get comfortable using AI code generation tools. I, personally, I hadn't coded in many years. I was a coder when I started, but in the last two years, I've written hundreds of thousands of lines of code using the AI coding agents that were specifically directed at tasks that we had that I wanted to make sure could contribute to these models, these frameworks, these skills, but keeping up with that has been a core aspect of getting everyone going. And then finally, as we talk about how do you actually take the data that we have, we had a huge effort to actually collect our data, make it AI ready, put it into what we call Dataverse, which is a lakehouse that we have across the company. But secondly, motivating all of our functional departments in the company to collect their own documents that weren't structured data. So what are the policies and procedures that you have? How do you actually do your day-to-day work? How do you document that and put those in our case in a document repository that then the agents that we wrote access that? And those agents were written by the people that were collecting the documents too. So we have structured and unstructured data now accessible using agents written by people using AI coding tools that then actually plug into this overall infrastructure that they gather. It was this, how do you get everyone in the company motivated, but then also get them engaged in understanding how this works.

Rob Strechay

>> Yeah. I agree, because I think when you look at how organizations are evolving from, "Hey, we had a chatbot." Now, we've moved into agents in Agentic where it's a series of different tasks that are tied to a process and how they come back. I think that's a really smart way to think about it, about the people who are documenting it, building the agents that go along with it, because they have that institutional knowledge to be able to make sure it's running properly.

Mike Ferris

>> Yeah. It's amazing the innovations that come out of this, right? I keep talking about the pyramid, getting everyone involved. At this point now though, we're taking the next step and we're starting to see things like, well, documents are really defining things that I do and policies, procedure, discounting constructs. And we've been using those in various tools across the company. Now, how can I represent that in an AI accessible way itself? Just like we made data, AI ready, how do we make these AI ready, not just through these agents? And so this is a movement I'm coding business as code. Being able to look at everything as markdown language that you then manage with engineering principles, right? So CICD isn't just for code anymore, it becomes part of your business operations. And that will be, I think, a big movement inside Red Hat. I expect it to show up in the market more as well, that you then start being able to have much more governed structured access that is part of your overall AI story. And so when you start talking about as an executive, if I want to understand a major change, if we decide to move into a new region or build a new product, I want to understand the impact to our partners, our discounting policies, our SKUs, how does that work, not just in the data that we get a result of sales, but also how does it work across procurement? I can do that in an automated fashion with AI and then I can work with the experts in the team that help create those policies, help manage what they do to say, "Hey, these are the changes we possibly need to make given this major shift that we'll make."

Rob Strechay

>> Yeah. I mean, I think you hit on two really important parts in those two different answers. I think the businesses code thing makes total sense to me. I love that idea and pushing down there, because I think that really drives what I call the return on AI or ROAI for that because everybody needs another acronym, right?

Mike Ferris

>> Yeah, absolutely.

Rob Strechay

>> So you have your ROAI and how do you actually get ... I think 2026 is targeting what is going to be my return on AI. And I think a big part of that also you talked about was the data and how the data is handled. One of the things I think was another big theme that we was discussed on the main stage was really sovereignty and sovereign AI. And I look at AI and sovereignty going hand in hand because it's such a data intensive process and activity, but help us understand how you guys see sovereign AI as well.

Mike Ferris

>> So very similar to how we've been dealing with hybrid, except now you have governments around the world, we have industries. A lot of pressure is being put on how do you actually better control access and use of your resources that you do have. And so when we talk about this, we launched several things today. We talked about what we're doing in the EU with sovereign support. We've been doing in the US for many years, got capabilities built into the platforms. So we're enabling it in the environments that we engage, but we also see it as a significant driver of choice, right? The need for choice. How do I actually choose the models, where they run, how I actually govern the agents themselves? How do I use all of this and report effectively on it to ensure that I am compliant? And so providing that framework is critical for us for any of our customers, especially the ones that are working in this highly regulated, incredibly complex world. Keeping up with the regulations are challenged enough, but then showing and being able to audit what happens in these organizations as you have possibly tens of thousands of agents running around in them is really going to require that hybrid like access and understanding, but also on common platforms that work across all of these together.

Rebecca Knight

>> Okay. Last question. A CIO comes home from the Red Hat Summit Atlanta 2026, what is the one thing you want that CIO to take away from this?

Mike Ferris

>> So the capabilities that Open has brought to the world are now bringing to AI. And the more that we can engage together working on these solutions, we can address this complexity that has increased, but you need to keep all those same principles you built starting with Linux in the early days, moving through containerizations with OpenShift. Virtualization, certainly, as pressures pushing. All this sits on a common platform that builds on top of itself as things move forward. Just like you're seeing AI elevate, you're seeing the platforms become more and more important across these complex environments.

Rebecca Knight

>> Mike Ferris, thank you so much for coming on the show.

Mike Ferris

>> Thank you.

Rebecca Knight

>> Always a pleasure. 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, be leader in enterprise tech news and analysis.