Red Hat Summit 2026 | Raghu Chakravarthi, Core42

Clips
More from Red Hat Summit 2026

Raghu Chakravarthi

Chief Technology and Product Officer

Core42

play_circle_outline Core42/G42: UAE AI Infrastructure Powering Global Customers — Khazna Data Centers with Core42 Compute, GPUs, Storage, Networking

play_circle_outline Diverse AI Infrastructure: Storage and Silicon Partners Power Bare-Metal Labs, Managed GPU Slicing, and Compass Token Factory

play_circle_outline AIOS Intelligence Grid: Single Hybrid-Cloud Platform for Distributed Inference and Unified Workload Management

play_circle_outline Greenshield and Digital Embassies: UAE's Global AI Investments Between US Production and Asian Consumption

Info
Transcript

Raghu Chakravarthi, Core42

Raghu Chakravarthi

Chief Technology and Product Officer Core42

In this interview from Red Hat Summit 2026, Raghu Chakravarthi, chief technology and product officer of Core42, joins theCUBE's Rob Strechay and Rebecca Knight to discuss how sovereign AI infrastructure is reshaping the global distribution of compute power. Chakravarthi introduces Core42 as the AI infrastructure arm of G42, a UAE-based organization operating data centers across the Middle East, Europe and the US — including a five-gigawatt facility in Abu Dhabi. He outlines the company's two founding principles: diversity across silicon and geography, and sov... Read more

explore Keep Exploring

What is G42 (and its related companies Core42 and Khazna), why is it called "42", and what services and infrastructure does it provide? add

What does Core42 (and its sister company Khazna) do, what AI infrastructure and data-center services do they provide, and what are their key differentiators? add

What is an AI intelligence grid, and how does a sovereign AI OS enable unified deployment across public clouds (Azure, AWS), private clouds, and on‑premises to provide operational simplicity, control (sovereignty), and distributed inferencing/workload management? add

Are you seeing the global picture play out right now—specifically that the US is the main producer of AI while other countries (like the UAE and parts of Asia) are the main consumers, and that initiatives such as UAE “digital embassies” and government-to-government trust arrangements are emerging to host foreign data and compute? add

bolt Powered by CUBE AI

Raghu Chakravarthi, Core42

search

Rebecca Knight

>> Hello everyone and welcome back to day two of theCube's live coverage of the Red Hat Summit here in Atlanta, Georgia. I'm your host, Rebecca Knight, alongside Rob Strechay talking sovereign AI right now.

Rob Strechay

>> Yeah. I mean, we're talking how you actually get AI to be AI and funny enough, you need a place for it to live. I know it's called cloud, but there's actually servers and stuff underneath and storage and other things.

Rebecca Knight

>> Indeed. There is some hardware there. Yes.

Rob Strechay

>> Yes, there is some hardware.

Rebecca Knight

>> Excellent. So I would like to welcome our next guest. He is Raghu Chakravarthi, chief technology and product officer at Core42. Welcome.

Raghu Chakravarthi

>> Thank you. Thank you, Rebecca. Good to be here.

Rebecca Knight

>> Fresh from the main stage. You were part of the keynotes this morning. You did a great job.

Raghu Chakravarthi

>> Thank you.

Rebecca Knight

>> Core42 is one of those companies that we are hearing a lot more about. How would you describe what Core42 does and its role in sovereign AI infrastructure?

Raghu Chakravarthi

>> So that's a good question. People always ask why 42?

Rebecca Knight

>> Yeah, right. Exactly.

Raghu Chakravarthi

>> Hitchhiker's Guide to Galaxy, that book had a computer called Deep Thought, which took about a million years to the question which was, "What is the answer to life?" Came back with the number 42. Nobody knows why. But our chairman and our founders were really impressed with that book. So they said, "Everything's going to be 42 with us." So our parent company is G42. It was formed in 2018. It was formed out of a lot of the government entities which had IT operations, and all of those were brought together in a private organization to run a lot of the infrastructure and the software solutions that UAE has. So we are a UAE-based company in Abu Dhabi, but we are now very global. I am here in the US running as the GM for Americas for Core42, running a lot of the things. But G42 services the entire life cycle of AI for any kind of customers - government, enterprise, and so on. So at the lowest level, we have our sister company called Khazna, which builds data centers. We are right now building a five-gigawatt data center in Abu Dhabi. And we've also gotten a lot of data centers here right here in the US. We have five that we operate. And we are building data centers and operating them in Europe - in France and Italy and so on. So Khazna builds the data centers, and Core42 is the infrastructure company. So you'll be hearing a lot about infra because that's where everything is these days. So we take over a data center, and we acquire both the compute nodes and the GPU nodes and storage, and we assemble it, operate it, and we network it. As you know, networking is key. So we have expertise in each one of these areas and we operate them. And we literally offer three basic services. On the AI side, we offer bare-metal as a service to our large AI lab customers. We offer a GPU as a managed service for our mid to smaller customers who can come in and do GPU slicing. And then we have our token factory service through our portfolio product called Compass. So Compass is not only a model router and . It also services a lot of the needs that customers have for tokens. So in the Core, we have two main first principles. One is diversity - diversity in silicon, diversity in location, and diversity across who we serve. So diversity in silicon, we work with NVIDIA, we work with AMD, we work with Qualcomm, and we are a large investor in Cerebras, and we are the largest operators of Cerebras clusters. Diversity and location, we operate in the public cloud. You want GPT models, we service them through a private offering in the UAE, through our own PTU cluster. And we also offer models from Anthropic, from AWS. So it's diverse. And from a storage standpoint, we work with all of our major partners - VAST, WEKA and DDN - and we offer a diverse set of storage solutions. And the second thing is about ... Model diversity and the whole diversity is important, but we talked today about sovereignty. Our key differentiator and a key offering is all about sovereignty, and that's a major focus for us. And then above us sits a lot of the sister companies which actually build large language models and also those which build SaaS applications and agentic applications that we host. We are the infra that we host and offer it to the world. And some of the things we are doing that is innovative is this. In UAE, the UAE government is interested in propagating AI because they think that's the next oil. So there's heavy investments going on in terms of that. And the UAE is trying to work with various nations that need AI capabilities, but they can consume it right away, and we are doing these things called digital embassies and we call it Greenshield, where we can actually host some of the sensitive workloads that these countries want to run and provide them with the infra, provide them with the operational capabilities and take it from there.

Rob Strechay

>> I mean, I just love it because again, we were talking before coming on here and I've known Core42 since its founding and as a partner in a previous life. And I think what's really nice about it is it really was focused on sovereign, it was focused on open, it was focused on a lot of it. You said a word on stage this morning that is near and dear to my heart because I've actually built some of the software underneath there is OpenStack and then there's OpenShift. Help folks understand the relationship with Red Hat as a partner and those two things and how they play a role in what services you guys provide.

Raghu Chakravarthi

>> Yeah. So Red Hat enables sovereignty for us. So we've been working with Red Hat for the last two years in our sovereign private cloud offering. So when it comes to sovereignty, there are two major dimensions. One is the data classification and the other ... So in data classification, it's about top secret, secret and confidential and public. This is how the classification is done in terms of sovereignty. So that is an important component that Red Hat, OpenStack and OpenShift lets us implement. And the other half is about three things. Sovereignty, when I talk about sovereignty, everybody says, "Oh, data sovereignty." Yeah, that is one of the components, not everything. So the most important component there is data. You need to have data residing in-country. So that's good. We have our own software stack and the hardware stack and we do that. But the second thing is about operational sovereignty. They want people in-country operating the whole environment. So it cannot be a public cloud-type environment. It needs to be on premise and so on. And the third thing is about technology sovereignty. I talked about the kill switch this morning. We cannot have an environment which can be having a kill switch that can be controlled from the outside and the whole environment dies. Red Hat is the main thing. We evaluated other solutions. There are other solutions, which is USA-based and European-based and Chinese-based, and we don't like any of those. Red Hat was the ideal thing because it doesn't have a kill switch. It gives me the technology sovereignty I need, and it enables the other two dimensions that we need in terms of data and operational. So that's why we started working with them on the sovereign private cloud, now with OpenStack on OpenShift. Now, they have this whole OpenShift AI, and now, it's being rebranded as Red Hat AI. I'm bringing the same component and the same level of flexibility and sovereignty to the AI side, which is very nascent by the way. There isn't many players there who offer a cloud management platform and so on. So Red Hat and us, we are going to pioneer and work together to build that same level of flexibility and sovereignty on the AI side.

Rob Strechay

>> So talk to that because again, with sovereignty, a lot of the folks that you guys have clients and customers of are looking at, "Hey, I have something in the UAE, but I also need to have it operate in the EU somewhere as well." Talk to that flexibility and resiliency and how that all plays together.

Raghu Chakravarthi

>> Correct. So we have this whole notion of AI intelligence grid. So these are connected data centers that enables compute that lives in UAE, whether it's GPU or CPU compute, to be connected with the ones in Europe, to be connected with the ones in US. It brings about many different dimensions, but first of all, it's the same AIOS that we will use. And that OS is sovereign by nature. We built sovereign capabilities in it. But what does that mean, right? So sovereignty, what we have built in is a set of controls and a landing zone, and that landing zone can exist in Azure or in AWS in the public cloud, and the same landing zone can be enabled in your private cloud and on-prem by the way, all these three environments. So what that means is it makes it a single OS, a single platform that I can deploy across multiple locations, multiple types of deployments, and what that brings about is operational simplicity. We are reducing the complexity around all of these. It brings about an operational simplicity. So with a single click of a button, I can see the entire network all connected across multiple data centers. I can see what is more utilized, what is underutilized, and I can actually now do distributed inferencing, distributed workload management through all of these in a sovereign manner. So we are bringing those capabilities around. So that connected AI intelligence grid is the core foundation that we have built everything on, and the enabler for this is the OS, and we are building a lot of those capabilities with Red Hat and building it on OpenStack and OpenShift.

Rebecca Knight

>> There's a real conversation happening right now about whether AI is separating along geopolitical lines. I'm curious from your vantage point, is that something that is already happening now? Are you seeing that play out right now in terms of the big picture?

Raghu Chakravarthi

>> 100%. 100%. Yeah. So US is the number one producer of AI, all of the hardware, the software and all the agentic stuff. But the number one consumer, if you go read the Microsoft distribution report, diffusion report that they published in December, January, the number one consumer of AI is not the US. US is actually the 28th or the 23rd consumer of AI. So the actual consumption's happening ... The number one consumer in terms of population is UAE. It's 68% of the population is consuming AI. And it's folks in Asia which are consuming. So we need the technology and the resources to be in locations where it's useful. Right now, a lot of the technology and the deployments are happening in the US. So we are actually serving inferencing needs halfway around the globe from here in the US. Latency, cost and everything, it's not efficient. So I think we need to propagate AI more across the world and make resources and things available. And this is exactly what UAE is doing with two types of initiatives. UAE has already realized this is the next oil. So they are diversifying, and they are doing two things. One is what we call digital embassies. So digital embassy is ... Think about it, what is an embassy? Embassy is a jurisdiction or a piece of land in a foreign country that is completely administered from a foreign land. I mean, it's a foreign property in US. So we think about that for digital. So we are creating these embassies in UAE because we have the resources, we have the capability and the expertise to build AI and all of these clouds, and we are renting these embassies to foreign governments to perform their critical tasks. But it's not just technology. The UAE government is putting its mind behind saying, "I'll do a government-to-government relationship and a trust deal where you can bring your data from your country, secret data, maybe confidential data, and you can do that compute in my country, but I won't interfere. This is your property. I'll give it to you." And they are funding it. There is a billion-dollar investment UAE has made into France and Italy and India and Kenya and so on to fund AI resource and AI initiatives. And there is a similar commitment that UAE has made to the US to do equivalent amount of investments right here. So I think the geopolitics is playing a thing, but I think UAE and G42, we are trying to make it easier and make it better for the world to consume AI, and we are going about it in our own way in multiple directions. Yeah.

Rob Strechay

>> I mean, just playing off of that and the fact that I love the idea of embassies, that's such a good way to think about it, about the sovereignty of that land you're landing, and I'm going to steal that. So that was great. That is fantastic. I love that you guys are doing that because when you start to look at sovereignty, it's not just about place.

Raghu Chakravarthi

>> Correct.

Rob Strechay

>> It's about the data. It's about the administration. How do you look at AI sovereignty into the future, and what role is Red Hat helping you guys with that?

Raghu Chakravarthi

>> I had a good talk with the Neural Magic folks yesterday. The founders and-

Rob Strechay

>> Yeah. Brian and the team there. Yeah.

Raghu Chakravarthi

>> Oh, amazing. Amazing. I think there is something there, all right? So what they are thinking about is from the ground up as Red Hat always thinks. So vLLM, that is our inferencing platform. Model serving platform is vLLM. Wonderful. So we use that right now to serve a lot of our open source models and so on. Next is how do I scale it? llm-d. How do I scale that? So llm-d is the next level of token economics. So now with the scale that I want to bring, I want to have as much cache hit as much as possible so that I can produce more tokens for the same amount of GPUs I got. So perfect. They're building that. And now, I was brainstorming with them yesterday about the next level. So sovereignty requires confidential compute. AI and confidential compute is very nascent right now. Yeah, NVIDIA enables it, but I'm thinking diversity. I have AMD there, I have Qualcomm there, I have Cerebras there. How do we bring confidential to all of these so that it can be in that mode. And then comes there thinking about encryption. All the way from the prompt from your chatbot to all the way to the back, can we encrypt so that we can bring a little bit more security to the sovereignty that we are talking about? So all of these are areas that Red Hat's already thinking about, I'm interested in. So it's a wonderful partnership. So we did great on the OpenStack on OpenShift, and we are taking it to that next level on the compute side, but AI is still nascent, and I think Red Hat's thinking about it the right way in many ways.

Rebecca Knight

>> Raghu, thank you so much for coming on the show. Really fascinating conversation.

Raghu Chakravarthi

>> Thank you very much. Appreciate it.

Rebecca Knight

>> I'm Rebecca Knight for Rob Strechay. Stay tuned for more of Day 2 of the Red Hat Summit here in Atlanta. You're watching theCube, the leader in enterprise tech news and analysis.