Broadcom introduces VMware AI Factory to automate private AI deployment

Broadcom has introduced the VMware AI Factory, a deployment model integrated into VMware Cloud Foundation (VCF) designed to accelerate the setup of private AI clouds. The system aims to reduce the time required to move from bare-metal servers to a deployed AI model from several weeks to as little as 30 to 45 minutes through automated hardware provisioning, Kubernetes integration, and lifecycle management.

The AI Factory supports a wide range of hardware, including accelerators from NVIDIA and AMD, and certified servers from vendors such as Cisco, Dell, Lenovo, and Supermicro. It features a ‘Model as a Service’ capability, allowing multiple business units to securely share model instances to optimize GPU utilization and reduce costs. The platform provides access to a catalog of over 150 verified commercial and open-source models, including Google’s Gemma and NVIDIA’s Nemotron.

To address security concerns regarding AI agents, Broadcom introduced AgentMinder, a gateway that verifies agent identities and evaluates requested actions against risk parameters. Additional security measures include the ‘Secure AI Sandbox’ for isolated agent execution and the TrueSource initiative, which provides clean-room builds for open-source software components like Java and Python.