Broadcom Launches VMware AI Factory to Secure Enterprise AI Agents
Broadcom's VMware AI Factory aims to simplify enterprise AI agent deployment and governance with automated infrastructure, centralized security controls, and secure sandboxes.
Broadcom has officially launched VMware AI Factory, a new platform designed to streamline the deployment, governance, and security of artificial intelligence workloads within enterprise private cloud environments. Unveiled at VMware Explore 2026, the platform addresses the significant challenges enterprises face in moving from raw hardware to production-ready AI, promising to reduce deployment times from weeks to mere hours.
At its core, VMware AI Factory leverages VMware Cloud Foundation's automation capabilities to orchestrate the entire AI lifecycle, from bare-metal infrastructure provisioning to model inference. This comprehensive automation aims to cut down on the complexity and cost typically associated with setting up and managing AI operations, allowing organizations to deploy AI models more rapidly and efficiently. The platform offers flexibility, enabling customers to choose their preferred hardware and vetted AI models while maintaining predictable private cloud costs.
Security and governance are central tenets of the VMware AI Factory. The platform introduces several key features to address the unique security concerns of enterprise AI. These include multi-tenant model sharing, an AI Gateway for centralized policy enforcement, and secure AI sandboxes designed to isolate potentially risky agent-generated code. These capabilities are intended to provide organizations with greater control over their AI deployments and mitigate the risks associated with autonomous AI behavior.
One of the standout security features is the AI Gateway. This component acts as a unified management interface, centralizing governance across both on-premises and cloud environments. It provides intelligent prompt routing, rate limiting for token usage, and application-level authorization controls, ensuring that AI interactions adhere to enterprise policies and security standards. This centralized approach is crucial for managing the increasing complexity of AI deployments and preventing unauthorized access or misuse.
Furthermore, VMware AI Factory introduces Secure AI Sandboxes, which utilize virtualised container spaces to isolate the execution of code generated by AI agents. This isolation is critical for preventing malicious or unintended actions by autonomous agents. Coupled with a robust control layer that defines how agents are invoked, what tools they can access, and how their outputs are validated, these sandboxes offer a significant layer of protection against the potential for rogue AI behavior.
Broadcom is also enhancing the hardware and model ecosystem for AI Factory. The platform is being paired with certified VCF AI ReadyNodes from major hardware vendors like Cisco, Dell, Lenovo, and Supermicro, and is collaborating with AMD to integrate its Instinct MI350 Series GPUs and ROCm software. Additionally, customers gain access to a governed model gallery featuring over 150 open-source and commercial models, delivered as a service, allowing for data-sovereign and cost-controlled AI deployments on private infrastructure.
The integration of MetalSoft further streamlines the physical server provisioning process, reducing it from weeks to minutes and allowing IT teams to manage heterogeneous hardware directly through the VCF console. This comprehensive approach, combining automated infrastructure, robust security features, and a flexible ecosystem, positions VMware AI Factory as a significant offering for enterprises looking to adopt AI responsibly and securely.