At VMware Explore 2025, Broadcom announced an expansion of its partnership with NVIDIA to drive large-scale AI adoption in private cloud environments. The strategy involves integrating the latest NVIDIA Blackwell GPUs, as well as high-performance networking technologies, into VMware Cloud Foundation (VCF), Broadcom’s unified private cloud platform positioned as a cornerstone for business modernization.
This initiative responds to a profound shift in the industry: applications of generative AI, agentic AI, and physical AI are transforming data centers, demanding more powerful and flexible infrastructures. Through this alliance, organizations will be able to train and deploy state-of-the-art models without leaving their private environments, combining NVIDIA’s innovation with VCF’s operational reliability.
An integration to accelerate enterprise AI
Paul Turner, Vice President of VMware Cloud Foundation Products at Broadcom, explained: “Our customers want the freedom to innovate with AI, but on trusted platforms. With this partnership, they can deploy and scale AI workloads alongside their existing applications, without sacrificing performance, efficiency, or operational simplicity.”
On NVIDIA’s side, Justin Boitano, Vice President of Enterprise AI, highlighted that the rise of generative AI is driving the need for new accelerated infrastructures: “With Blackwell architecture and our high-performance network, customers will be able to run powerful AI applications directly in their private cloud, leveraging NVIDIA’s entire ecosystem together with VCF.”
Key benefits of the Broadcom–NVIDIA integration
The expansion of this collaboration is focused on several strategic areas:
1. Support for NVIDIA Blackwell GPUs
- NVIDIA RTX PRO 6000 Server Edition GPUs, designed for organizations needing to combine desktop virtualization (VDI) workloads and AI on the same hardware.
- NVIDIA Blackwell B200 GPUs, supported in future versions of VCF, with a focus on large-scale model training and HPC (High Performance Computing) applications.
2. High-performance networks for generative AI
VCF will support NVIDIA ConnectX-7 NICs and NVIDIA BlueField-3 400G DPU with DirectPath I/O, enabling technologies like NVIDIA GPUDirect RDMA and GPUDirect Storage. These capabilities are critical for distributed training and large-volume data transfers, reducing latency and increasing efficiency.
3. Compatibility with VCF features
One of the main advantages is that organizations can adopt this infrastructure while keeping critical VMware functions intact:
- vMotion for seamless workload migration.
- High Availability (HA) and DRS (Distributed Resource Scheduler) to ensure service continuity.
- Live Patching to apply updates without downtime.
Practically, this means AI/ML workloads can coexist smoothly alongside traditional enterprise applications within the same environment.
Context: private cloud as the engine of AI
This integration comes at a time when AI models surpass one trillion parameters and require more resilient and higher-capacity environments. Public cloud providers have led the initial phase of AI, but increasingly, companies seek control, data sovereignty, and cost predictability, driving interest in the modern private cloud.
Broadcom believes that VMware Cloud Foundation, enhanced with NVIDIA technology, will serve as the natural solution for organizations needing to combine regulatory security, scalability, and extreme performance.
Conclusion
The partnership between Broadcom and NVIDIA at VMware Explore 2025 reinforces a clear message: large-scale artificial intelligence will not be exclusive to big public cloud providers. With the arrival of Blackwell GPUs and next-generation networks in VCF, companies can build their own private “AI factories” with similar efficiency, but with greater control over their data and operations.
In a landscape where AI is redefining the digital economy, this collaboration bridges cutting-edge innovation and enterprise reliability.
Frequently Asked Questions (FAQ)
What is VMware Cloud Foundation (VCF)?
It is Broadcom’s unified private cloud platform, integrating compute, storage, networking, and management within a single cloud operating system.
What does the NVIDIA Blackwell architecture bring to VCF?
Blackwell GPUs are designed for large-scale AI and HPC workloads, offering significantly higher performance and optimizations for training generative models and advanced reasoning.
Can traditional applications run alongside AI in VCF?
Yes. Thanks to the integration, classic enterprise applications can coexist with AI/ML workloads within the same environment, maintaining essential features like HA, vMotion, or DRS.
What benefits does this integration provide for generative AI workloads?
High-speed networks and GPUDirect compatibility reduce latencies and accelerate training and deployment of models, which is critical for generative AI and agentic AI.
via: broadcom