Sovereign by design: Lessons from Red Hat Summit

Digital sovereignty used to sit somewhere between a compliance checkbox and a future roadmap item. That’s changing fast. At Red Hat Summit, Mohammed Retmi of Core42 in the United Arab Emirates and A.S. "Raj" Rajgopal of NxtGen in India joined me for a roundtable on how 2 very different markets are building sovereign clouds, and why the conversation now shows up in nearly every customer meeting.

Their starting points could not be more different. In the UAE, national regulation defines what partial and full sovereign cloud mean, from data residency and encryption to government pre-screening of cloud operators. In India, there is no sovereignty mandate today; adoption is driven by risk mitigation, customer awareness, and heavy investment in AI. Yet both leaders described the same shift: sovereignty is no longer optional. 

You can watch the full roundtable discussion from Red Hat Summit here:

From data residency to full-stack control

Customer conversations have moved well beyond where data is stored. Raj described how discussions in India now extend to operational sovereignty (who runs the platform) and technological sovereignty (freedom from a single vendor stack). Government, healthcare, and financial services are the most active verticals, but AI has made the topic universal.

In India, Rajgopal said sovereignty today is a strategic choice rather than a regulatory requirement. Data privacy rules protect citizen identity, and government-led AI initiatives may shape future policy. For now, organizations are making deliberate decisions about exposure, especially as AI workloads multiply.

In the UAE, Retmi and Core42 distinguish partial sovereign cloud (data in-country, encrypted at rest, in transit, and in use) from full digital sovereignty, which also requires operational control and technology independence. Working with open infrastructure, Retmi noted, gives Core42 more independence than relying on hyperscaler platforms deployed in-country.

Sovereign AI: Models, languages, and borders

AI and sovereignty now travel together. In India, NxtGen supports model developers backed by government GPU funding, deploys Indic language models that can converse in 22 languages, and routes reasoning through open source large language models (LLMs)—all without sending tokens outside the country. The team built a multivendor GPU platform on Red Hat OpenShift spanning NVIDIA, AMD, and Intel, so customers are not locked into a single chip architecture.

Core42 takes a layered approach in the UAE. Its sister company, Inception, trains and hosts fully sovereign Arabic models. A dedicated in-country OpenAI cluster serves customers through a partnership with Microsoft. Open source models run on GPU-as-a-Service infrastructure, with future plans to deploy Red Hat OpenShift AI underway.

The architectural goal is the same in both regions: keep sensitive data and inference inside national borders while still delivering modern AI experiences.

The migration reality

Moving workloads into sovereign environments is a journey, not a project you finish in a quarter. Retmi described common friction points: subjective data classification, resistance to change from security and IT leaders, and the need for phased roadmaps that start with less critical applications. He has also seen customers move data to hyperscaler platforms, only to pull secret or top-secret workloads back into private cloud when classification or risk posture changes.

India is seeing a different kind of urgency. Rajgopal has seen many customers move their proprietary private clouds to NxtGen hosted private cloud or public cloud. This is driven partly by cost but increasingly by the need to run modern workloads on OpenShift and OpenShift AI. 

Once customers land on NxtGen infrastructure, their data stays in a private environment where GPU resources and models can power new use cases. His team has delivered more than 150 AI use cases to date, including large-scale public-sector data projects.

"In the short term, we have a massive, almost 54-55% of the VMware private clouds moving to our cloud… I think next year we'll be busy just migrating customers from VMware to our cloud and then adding AI use cases."
A.S. "Raj" Rajgopal, NxtGen

Why the platform choice matters

Both partners stressed that sovereignty is not only about geography. It is about control over the full stack.

Core42 and Red Hat have spent several years building a sovereign private cloud that delivers classical infrastructure-as-a-service (IaaS) and platform-as-a-service (PaaS) in an environment designed to feel familiar to hyperscaler users, with stronger security and sovereignty assurances. Retmi was direct about the outcome:

"We could not achieve full digital sovereignty except with Red Hat in this private cloud run. We cannot achieve it with the hyperscalers."
Mohammed Retmi, Core42

NxtGen runs its public cloud on Red Hat OpenStack and private cloud on OpenShift. Customers include a Japanese automotive manufacturer and major regional consumer goods provider for fully sovereign private cloud deployments, as well as critical government workloads in finance, health, and elections.

Private-sector demand is rising in the UAE as well. Retmi said non-regulated customers increasingly want in-country backup and secondary copies of data in private cloud environments, —further evidence that sovereignty is becoming a mainstream business requirement, not a niche compliance exercise.

Barriers, futures, and advice for IT leaders

The biggest barrier Retmi sees is speed. AI adoption and geopolitical pressure are pushing cloud providers to deliver sovereign capacity faster than ever. Raj offered a more philosophical view: perfect sovereignty may be impossible for data already collected at a global scale, especially in a country with hundreds of millions of internet users. The practical goal is choice and control over what happens next.

"I really don't know whether we can get to digital sovereignty in that sense, because there's so much data that has already gone away from us. But at least if we can have control over the actions that are being taken, that's a massive challenge." 
A.S. "Raj" Rajgopal, NxtGen

Looking ahead, Core42 is exploring how to export UAE sovereign cloud expertise to other nations through a "digital embassy" concept, pairing government agreements with infrastructure, applications, and operational know-how. NxtGen is betting on people: more managed services above the infrastructure layer and faster enterprise AI adoption so customers are not left behind.

For IT leaders weighing their own path, Mohammed's closing advice may be the one to remember:

"We used to say 'security by design.' We should have this 'sovereignty by design' in all what we are developing. So owning and having the upper hand on our data is having the upper hand on our future."
Mohammed Retmi, Core42

Raj urged leaders not to wait for AI to mature before acting. Security threats evolve daily on traditional workloads; delaying AI adoption risks falling behind globally.

Take the next step

Whether your organization is responding to regulation, managing AI risk, or replatforming your virtualized infrastructure, sovereignty is a design decision, not a retrofit. Assess your posture across data, operations, and technology, and explore how open infrastructure, from metal to AI agent, can help you maintain control without sacrificing innovation.

Watch the full Sovereignty Customer Roundtable video to hear firsthand from Core42 and NxtGen about their strategic approaches to building sovereign clouds.

Learn more about Red Hat's approach to digital sovereignty and sovereign cloud.