IBM’s Sovereign Core Software Foundation Set To Give Partners More Control Over AI

Sovereign Core “is really intended to be taken by partners, by integrators and by enterprises as well to deploy their sovereign environment,” says IBM executive Will Streit.

IBM plans to launch its artificial intelligence-ready Sovereign Core software foundation into tech preview next month, moving the offer into general availability in the middle of 2026 and giving partners a way to build repeatable deployments in multi-tenant models they operate, unlock more AI use cases for customers in regulated industries and have an audit trail for compliance purposes.

The software promises enterprises, governments and service providers a way to deploy and operate AI-ready environments quickly and at scale while advancing infrastructure and tooling they’ve already invested in, Will Streit, a vice president in IBM Software, told CRN in an interview.

Sovereign Core “is really intended to be taken by partners, by integrators and by enterprises as well to deploy their sovereign environment,” said Streit (pictured above), who leads the product management and business dimensions for core IBM Data and Analytics software portfolios including Cognos Analytics, SPSS, Data Replication, Master Data Management (MDM) and OpenPages.

IBM Sovereign Core

The Armonk, N.Y.-based provider of mainframes, AI and hybrid cloud products positions Sovereign Cloud as a way for MSPs and SIs to achieve infrastructure optionality and flexibility in reusing existing infrastructure for their customers while saving partners time and effort from building digital sovereignty from parts, Streit said.

“Our initial focus really is on this partner community,” he said. “By providing this as a single install product, it’s an accelerator. It accelerates laying out the foundation, for example. It accelerates establishing a sovereign boundary. It accelerates a lot of the key capabilities. Again, we’re not trying to be a cloud the box, but there are key things like metering or how do I feed into billing or how do I how do I then set up a catalog so that, if I’m a partner, I can then add things into that catalog that my specific customers might want to use. A lot of this value that is tailored to exactly that audience.”

How Sovereign Cloud Works

Users also gain greater control and the ability to create audit trails by owning the software and its different access components, not trusting that to a third-party software provider, and gaining more granular levels of access and AI model authorization, according to IBM.

Sovereign Core was created to meet new digital sovereignty needs in the AI era. Sovereign Core enforces sovereignty architecturally, not contractually, according to IBM. Policy overlays and provider-managed control planes don’t solely dictate control. Instead, automated, always-on, user-owned control planes are at work.

IBM built Sovereign Core on the OpenShift platform offered by its Red Hat division with air-gapped software that works like software-as-a-service (SaaS) but gives the user full, local authority.

The OpenShift foundation allows organizations to advance the infrastructure, platforms and operational tooling they’ve already invested in on-premises, regional clouds or partner-operated environments, according to IBM.

The sovereign boundary holds identity, encryption keys, logs, telemetry and audit evidence, with ongoing compliance capabilities embedded into the software for regulator-ready proof on demand. In the boundary, users can deploy clusters based on central processing units (CPUs) and graphics processing units (GPUs) and bring in open and proprietary AI models.

Sovereign Core’s controlled gateways govern access to models, tools and knowledge source. And AI inference and agent-based applications run locally without exporting data or telemetry to external providers.

IBM positions all that’s avaialble within Sovereign Core boundaries as saving users time from manual, audit-driven processes. Sovereign Core enforces sovereignty at runtime and configuration time with continuous monitoring and recording of identity, access, model usage and operational activity for audit trails needed in regulated sectors.

The customer-operated control planes promise rapid deployment, consistent operation at scale and the ability for localized central teams to operate thousands of cores and hundreds of nodes with different sovereign requirements and controls from one plane, according to IBM.

Users can self-service provisioning for CPU, GPU, virtual machine (VM) and AI inference environments for repeatable deployment across enterprise-run and partner-operated multi-tenant models while maintaining control. And the planes have automated configuration of identity, security and compliance built in from the get-go.

Sovereign Core also provides a curated catalog of atomic accelerators for application deployment and operating traditional and AI applications, according to IBM.