Agentic AI shifts enterprises from models to platforms

As agentic AI moves from experimentation to production, enterprises are grappling with cost, data exposure, and infrastructure control rather than simply choosing AI models. Rising token costs and compliance concerns are pushing organizations to reconsider exclusive reliance on public cloud AI services, according to Red Hat VP Joe Fernandes. Platform teams are becoming central to AI strategy as autonomous agents replace simple assistants and require sandboxing, access controls, and auditability. Red Hat is contributing to Nvidia's OpenShell sandbox runtime and advocating for hybrid, open-source infrastructure spanning public cloud, private environments, sovereign clouds, and edge deployments.