Mavenir and Red Hat launch carrier-grade AI platform for operators
Mavenir has announced an Integrated AI Platform built in collaboration with Red Hat, aimed at giving telecoms network operators the infrastructure to sell AI capabilities directly to subscribers and enterprise customers. The platform uses Red Hat's Kubernetes-based OpenShift environment and Red Hat AI tooling, layered on top of Mavenir's existing telco software stack, and is designed to let operators meter and bill for AI consumption in the same way they already bill for mobile data.
The platform supports three commercial modes: operators can deliver AI services under their own brand to retail subscribers; provide AI compute infrastructure for "AI grid" deployments hosting third-party workloads; or offer metered AI access to enterprise customers as a value-added service bundled alongside connectivity contracts. Token-accurate billing integrates with operators' existing business support systems, and on-premises small language models handle the majority of traffic, with policy-governed routing to external frontier models available for tasks requiring more advanced reasoning.
The architecture
At its core, the platform is a hybrid deployment model. On-premises, open-source models serve routine workloads, keeping subscriber data and model weights on operator-controlled infrastructure to address data sovereignty requirements. Mavenir's model router manages selective, policy-governed access to frontier models when needed, with hard spend caps intended to give operators predictable economics. The stack also includes zero-trust identity controls, closed-loop service assurance, MLOps tooling powered by Red Hat AI, and agentic orchestration capabilities.
Bejoy Pankajakshan, Chief Technology and Strategy Officer at Mavenir, said the platform is designed so that "operators gain sovereign control over models and data, token-accurate monetization that integrates with their existing BSS, and the service assurance to back their own managed AI services with contractual SLAs." Red Hat CTO Chris Wright added that the collaboration gives operators "a path from AI experimentation to AI monetization without rebuilding their operations model."
Market context
The launch reflects growing pressure on mobile network operators, who have watched hyperscalers and specialist AI providers capture revenues from AI services that run over operator-owned connectivity. Operators are well placed to offer AI as a managed service given their existing billing relationships, regulated infrastructure and carrier-grade reliability track records; the challenge has been assembling a credible, production-grade AI platform capable of meeting telecoms-grade SLA commitments.
A number of vendors are targeting this space, including network equipment makers and cloud-native telco software specialists, while hyperscalers are pursuing operators as distribution channels rather than competitors. Mavenir's differentiation rests on its existing deployments with more than 300 operators across 120 countries, which the company argues gives it a reference architecture others cannot yet match at scale.
From a regulatory standpoint, the platform's sovereign-first design is well timed. The EU AI Act's general-purpose AI obligations are phasing in, and regulators across multiple jurisdictions are tightening rules on where subscriber data can be processed. An on-premises default architecture, with documented data residency and zero-trust security controls, is increasingly a commercial prerequisite rather than a differentiator for operators serving regulated sectors such as financial services and public infrastructure.
Mavenir will demonstrate the platform at DTW Ignite 2026, running 23 to 25 June, where it is exhibiting at booth 334. No commercial launch customers, pricing tiers, or specific throughput benchmarks were disclosed in the release.