VMware has integrated some of the world's most widely used AI models into its VMware Cloud Foundation private cloud platform, enabling partners to deliver model-as-a-service capabilities to customers running on-premises infrastructure. The updates, announced this week at VMware Explore 2026 in Las Vegas, include newly validated AI models from Google, Nvidia, NEC, Alibaba Cloud and Z.ai. According to Bob Keblusek, CTO at VMware partner Sentinel Technologies, the integration removes barriers for customers looking to operationalize artificial intelligence at scale.
The newly validated models on VCF include Google's Gemma 4, an open-source multimodal model family designed for building and deploying autonomous AI agents, and Nvidia's Nemotron 3, which uses hybrid Mamba-Transformer MoE architecture with 1 million context and multi-environment reinforcement learning to power long-running agentic workflows. Additional models include Z.ai's GLM 5.2 General Language Model for coding and reasoning agents, NEC's cotomi model, and Alibaba's Qwen 3.8-27B open-weight model. With these additions, customers can now run more than 150 open-source models on VCF, giving enterprises a broad selection of AI models validated for their on-premises infrastructure. The platform allows organizations to run inference workloads, agentic applications, containerized services and traditional virtual machines together, eliminating the need to manage separate technology stacks.
VMware also launched new AI-ready data foundations for its Tanzu Platform, providing what the company describes as an end-to-end framework for transitioning safely from initial AI pilots to fully production-ready AI agents inside secure private clouds. New capabilities include hardened agent sandboxes that enforce a "deny-by-default" security containment model isolating credentials to prevent prompt injection attacks and unauthorized network access. The update also features AI-ready data foundations that process structured and unstructured enterprise data on-site, delivering high-precision context to AI agents to improve accuracy, reduce hallucinations and lower token costs. Other innovations include a developer harness with preapproved skills, step-by-step workflows and human-in-the-loop controls, plus a curated marketplace offering a centralized catalog where developers and agents can discover vetted AI models, tools and data products.
Keblusek said VMware is providing clients a clear path to data sovereignty and cost-effective AI at scale, with leading models securely available and delivered as a service through VCF. He noted that for Sentinel Technologies, cost optimization dashboards and financial operations around tokenization become critical as AI adoption accelerates, particularly when using frontier models, allowing many use cases to be localized. On the security front, Keblusek emphasized that governing, controlling and orchestrating agentic AI represents a major topic for customers adopting agentic systems, and VMware's approach through the Tanzu platform makes sense for organizations looking to run AI agents in their own data centers or in VMware secure enclaves, as security remains the top concern when customers discuss AI adoption. He said VMware's focus on monitoring and building secure, trustworthy infrastructure through policy enforcement will drive AI adoption rates even higher. The company's strategy of easier bundling and simplified operationalization directly addresses adoption barriers partners see in the field. By combining data sovereignty with a diverse model catalog and built-in security controls, enterprises gain the confidence to move AI workloads beyond experimentation into full production, a shift that requires both technical capability and trust in the underlying infrastructure.