Enterprises can achieve enormous productivity and business transformation from AI. With the VMware Private AI Foundation on NVIDIA, Broadcom and NVIDIA aim to unleash AI potential with a lower TCO and boost productivity. Recently, Broadcom and NVIDIA released several features in VMware Private AI Foundation in VCF 9.0 to further our mission of providing private and secure AI models for businesses. Today, we are excited to announce even more features to help enterprises achieve this mission.
VMware Private AI Services Now Included in VCF Subscription #
At the ongoing VMware Explore conference, a major update has just been announced: the VMware Private AI Foundation, which was previously sold separately, is now directly included in the VCF platform. Private AI services ensure privacy and security, simplify infrastructure management, and streamline model deployment. These services include features like GPU monitoring, model storage, model runtime, agent builder, vector database, and data indexing and retrieval. By embedding all the benefits of Private AI into VMware Cloud Foundation, businesses get a unified platform for both AI and non-AI workloads without an extra purchase.
New Supported Features Now #
We are releasing an exciting new feature on the platform.
NVIDIA Blackwell Architecture Support: https://www.nvidia.com/en-us/data-center/technologies/blackwell-architecture/
NVIDIA Blackwell GPUs provide enterprises with exceptional performance, efficiency, and scale to unleash the potential of generative, agentic, and physical AI. VCF now supports the NVIDIA Blackwell architecture, enabling businesses to get industry-leading AI training and inference capabilities at an unprecedented scale. Let’s take a look at the supported GPUs.
For workloads that require the compute density and scale provided by a data center, the NVIDIA RTX PRO 6000 Blackwell Server Edition GPU offers powerful performance for next-generation AI, scientific, and visual computing applications in industries such as healthcare, manufacturing, retail, media, and entertainment. The RTX PRO 6000 Blackwell Server Edition is designed for enterprise data center deployment and can be configured with up to 8 GPUs per server.
Future Versions #
We will continue to introduce exciting new features in the future. Let’s take a look at some of them.
Enabling Privacy and Security #
Broadcom’s partnership with NVIDIA on this platform aims to help businesses build and deploy private and secure AI models with integrated security features within VCF and NVIDIA AI Enterprise. Let’s look at one of the features we will release in the future that is designed to enhance privacy and security.
Multi-tenant Model-as-a-Service
In VCF 9.0, we announced the general availability of the Model Runtime feature, which allows enterprises and cloud service providers to deploy models as a shared and scalable service for their users. Model Runtime services will be enhanced with model endpoint sharing, allowing secure sharing of AI models between tenants or different business lines while ensuring complete data privacy for each model. This means that enterprises and cloud service providers can host a single model instance and scale it horizontally across the organization, with each team or department having a separate, private, and secure namespace to store their private data. With this feature, businesses can reduce power consumption and improve AI efficiency by eliminating model redundancy while still ensuring that the way data is shared with the model is secure and reliable for each tenant or business line. This puts private AI in enterprises on the same level as AI model services in the public cloud.
Simplifying Infrastructure Management #
The VMware Private AI Foundation on NVIDIA architecture provides purpose-built features that help simplify infrastructure management and optimize costs for AI environments. Let’s review one of the features we will soon release to simplify infrastructure management.
DirectPath Support for GPUs with VMware Private AI Foundation and NVIDIA
VMware Private AI Foundation with NVIDIA now supports DirectPath for NVIDIA accelerated computing. This feature provides an easier infrastructure path for businesses to launch and scale AI projects, minimizing the required licensing.
It supports high-performance, exclusive GPU access for a single virtual machine, allowing the VM to take full advantage of the GPU’s capabilities. With this new feature, businesses can easily conduct AI experiments, prototype new applications, and deploy AI projects on VMware Private AI Foundation in partnership with NVIDIA.
VCF Smart Assist
This LLM-based assistant, built on the capabilities of VCF Private AI services, will be integrated into VCF as an AI assistant for VCF users. When a problem arises, the smart assistant will allow users to quickly access Broadcom’s knowledge base for a solution, significantly reducing downtime in physically isolated and interconnected private cloud deployments.
Simplifying Model Deployment #
Broadcom and NVIDIA also provide software and features that greatly simplify model and AI agent deployment for data scientists. Let’s look at some of the features in this category that will be released in the future.
Model Context Protocol (MCP)
In a future version, MCP support will be added to the VMware Private AI Foundation in partnership with NVIDIA. MCP will provide a standardized method for integrating AI agents with internal content repositories and external tools like Oracle, Microsoft SQL Server, ServiceNow, GitHub, Slack, and PostgreSQL without the need for custom connectors. This will also include end-to-end authentication and RBAC, ensuring secure, scalable workflows and empowering developers to create context-aware applications and task automation using real-time, licensed data.
High-Speed Networking with Enhanced DirectPath I/O
VCF will support NVIDIA ConnectX-7 and NVIDIA BlueField-3 SuperNICs with enhanced DirectPath I/O. This will allow customers to leverage advanced features such as GPUDirect® RDMA and GPUDirect Storage for high-speed, multi-host AI model training and data transfer, which is crucial for demanding generative AI workloads.
The cornerstone of this integration is the customer’s ability to retain familiar VCF operational workflows and enterprise-grade virtualization features such as vMotion, High Availability (HA), Distributed Resource Scheduler (DRS), and live patching.
With support for NVSwitch on the HGX platform equipped with Blackwell GPUs, enterprises will be able to perform large-scale AI deployments on the VMware Private AI Foundation with NVIDIA.
The NVIDIA HGX platform seamlessly integrates NVIDIA accelerated computing, NVIDIA NVLink, NVSwitch, NVIDIA networking, and a fully optimized AI software stack with NVIDIA AI Enterprise, providing the highest AI application performance and the fastest insights for every data center. NVSwitch can connect up to 8 GPUs per node, forming a high-speed GPU communication network of 900 GB/s. This powerful combination with the VMware Private AI Foundation and NVIDIA provides unparalleled performance for LLM inference and training in private and secure AI deployments.
VCF Support for NVIDIA HGX B200
Future versions of VCF will support the NVIDIA HGX B200 system to meet the needs of advanced accelerated computing and generative AI workloads. As an outstanding accelerated-scale x86 platform, the HGX B200’s real-time inference performance is up to 15 times faster than the previous generation Hopper, with 12 times lower cost and 12 times lower energy consumption, and it is designed for the most demanding AI, data analytics, and high-performance computing (HPC) workloads.
New Partners in the VMware Private AI Ecosystem #
In addition to core features, the VMware Private AI Foundation, in partnership with NVIDIA, is expanding its ecosystem through strategic partnerships, further enriching the enterprise value proposition. We warmly welcome the following partners to the VMware Private AI Ecosystem:
- Zenera: https://zenera.ai/ This partnership will leverage the power of the VMware Private AI Foundation and NVIDIA to enable intelligent applications with autonomy, allowing them to understand, plan, code, and complete tasks independently. Zenera’s expertise in autonomous AI solutions complements our platform, enabling businesses to deploy more complex and autonomous AI applications.
- Xtravirt: https://xtravirt.com/ Xtravirt, a leading cloud consulting and managed service provider, is partnering with Broadcom to help enterprises realize value from the VMware Private AI Foundation and NVIDIA collaboration. From designing and deploying secure, scalable infrastructure to integrating AI applications and services, Xtravirt’s extensive experience will simplify complexity and accelerate the long-term success of our shared customers. Their comprehensive services ensure that businesses can navigate the complexities of AI applications with confidence.
- ITQ: https://itq.eu/ Adopting AI while meeting sovereignty, compliance, and performance requirements is a complex challenge. ITQ solves this by combining Broadcom’s deep expertise, award-winning ITQ engineers, and a sovereign, enterprise-grade AI infrastructure built with the VMware Private AI Foundation and NVIDIA. This collaboration enables enterprises to innovate faster and protect sensitive data when deploying AI, ensuring that regulatory and performance requirements are met without compromise.