- IBM is introducing Sovereign Core to tackle control of AI workloads
- The software is built on Red Hat tech and works across on-prem and cloud environments that support Kubernetes
- It can help telcos rapidly launch and manage sovereign AI offerings for industries or individual customers
AI is making digital sovereignty more complex than ever. It’s no longer just about where data is stored and accessed but also where it is created. For enterprises, wrapping their arms around what AI-enabled applications are generating data and where they’re doing it is tricky. But IBM has a new tool it thinks can help: IBM Sovereign Core.
Built on Red Hat’s open-source technology, IBM Sovereign Core is software designed to help enterprises, telcos and managed service providers build, deploy and manage AI workloads across cloud environments within their chosen jurisdiction. It allows enterprises to maintain direct control over their software operations and system configurations, ensure local inference execution and AI agent operations, and keep things like encryption keys and authentication close to home.
Here's how it works: once the software is deployed, customers can set their sovereign boundary and controls. The software supports multi-tenancy, meaning multiple departments or customers can operate within the boundary a user has set. They can also bring their own AI models, which will then run inside the specified boundary area – a key feature for demonstrating compliance with sovereignty regulations.
Notably, IBM Software VP Will Streit told Fierce that Sovereign Core was made to be used as bundled software. That is, IBM won’t be the one operating it. Instead, enterprises and telcos can use it to rapidly set up sovereign environments and offerings for specific industries or individual customers.
Streit added that IBM is working with a number of telcos as design partners and is working on building in the ability for the software to feed into billing, OSS and BSS systems.
Flexibility in focus
While he didn’t name names in terms of compatible clouds, Streit told Fierce that Sovereign Core can basically run in any environment on-prem or in the cloud that supports Kubernetes.
“Generally, anything that Red Hat runs on, this runs on,” he said. That flexibility, he added, is intentional, allowing customers to move workloads not just across regions but also across infrastructure. This, combined with the ability to set unique controls will allow customers to adapt as regulations catch up with AI advances.
“I can say very clearly what makes something FedRAMP [compliant]. Sovereignty isn’t that clear. It’s evolving. So, building something flexible is important,” Streit explained.
Indeed, Moor Insights and Strategy VP and Principal Analyst Jason Andersen pointed to Sovereign Core’s ability to offer a single interface that works across on-premises and regional cloud hardware as a standout feature.
He added it also addresses emerging real-time needs associated with AI inferencing workloads, which is key, given that more GenAI-specific regulatory requirements are likely on the way.
“IBM can direct where to run AI inference at deployment time to stop an issue before it happens,” he explained. “Previous sovereign cloud solutions focused on storing and processing data, now we need to deal with generating new content altogether.”
Asked who and what Sovereign Core competes with in the market, Andersen pointed to cloud-specific sovereignty solutions from AWS, Google and Microsoft. But he added he has “yet to find something that is built for hybrid environments like IBM is proposing here.”
One other potentially big deal? The ability to easily demonstrate compliance with sovereignty regulations, and thus cut audit-related costs.
“Compliance and audits cost a lot in terms of time and resources, and it’s only getting worse,” Andersen said. “Anything to simplify and streamline that process will be of big benefit to IT and MSPs.”
IBM Sovereign Core will be available in tech preview next month, with full general availability planned for mid-year 2026.