VMware Cloud Foundation Is Now an AI-Native Platform
Broadcom is anticipating that most organizations will want to use AI in some fashion, and that the only economical way they can do so is through running a cloud in-house.
This year, at the company’s annual VMware Explore 2025, the company has unveiled that it has equipped its private cloud platform, VMware Cloud Foundation (VCF), to make it easy for developers to create and run AI workloads.

Screenshot from Broadcom.
Built on the vSphere compute engine, VCF 9.0 is now generally available. Enterprises will get one unified platform for their AI and non-AI workloads without an additional purchase.
New AI services include:
- A Model Store
- Model Runtime
- GPU Monitoring
- Agent Builder
- A Vector Database
- Data Indexing/Retrieval
The platform enhances privacy and security and streamlines model deployments with a multiaccelerator model runtime that will enable organizations to deploy AI models across a range of hardware, including AMD and NVIDIA GPUs, without refactoring.
Organizations can run multiple tenants on a single platform, allowing different business units to share models while keeping separate namespaces for full data privacy and isolation. Models can be located on premises or in the cloud.
To help with model design and fine-tuning, Broadcom will offer an AI-driven support assistant, which is currently in tech preview. The assistant can help diagnose and resolve issues, using Broadcom’s knowledge base.
The company has also standardized the process of managing Model Control Protocol (MCP) instances by providing a unified way to link AI assistants with internal content repositories and external tools such as Oracle, Microsoft SQL Server, ServiceNow, GitHub, Slack, PostgreSQL and other services, without the need to write custom connectors.

Other VCF Enhancements
In addition to the AI-related augmentations, VCF comes with other improvements as well.
In the realm of storage, VCF cloud can now offer a full range of storage formats. It has long supported file and block storage, and with this release, it will natively support object storage as well through an S3-compatible object storage interface. This offers customers the ability to centralize all their storage through VCF. Also, object storage is favored for AI workloads, thanks to its ability to store large chunks of unstructured data.
For those that want to move into GitOps deployment practices (particularly with its Kubernetes workloads), VCF now supports Argo CD, the cloud native community’s favored open source CI/CD tool built on GitOps principles. Infrastructure and apps can be stored as code in Git and deployed through a service mesh run on Istio.
Broadcom is even standardizing on the operating system of choice. It has expanded a partnership with Canonical to support the Ubuntu Linux distribution and other Ubuntu-related software on the platform.
Nine of the top 10 Fortune 500 companies have adopted VCF, with customers worldwide having licensed more than 100 million cores of VCF.
