At VMware Explore 2025, Broadcom announced the further integration of NVIDIA's latest AI technology into VMware Cloud Foundation (VCF). This will enable enterprises and cloud service providers to build, deploy, and scale next-generation AI models in private cloud environments, while continuing to build a trusted service operations experience through VCF.
Generative AI drives data center architecture transformation
Generative AI, agent AI, and physical AI applications are rapidly changing data center infrastructure, placing unprecedented demands on computing and network performance. The collaboration between Broadcom and NVIDIA not only continues their long-standing partnership but also expands the introduction of NVIDIA's latest Blackwell graphics architecture GPUs, includingRTX PRO 6000 Server Edition, and the soon-to-be-supported Blackwell B200 GPU, while also introducing NVIDIA's network technology to provide greater choice, flexibility, and operational efficiency.
Broadcom said these technologies will expand the application capabilities of the two parties' previous collaboration project on VMware Private AI Foundation with NVIDIA, allowing enterprises to more effectively grasp the driving force of AI innovation.
Combining AI with enterprise-grade stability
Paul Turner, vice president of Broadcom's VCF division, emphasized: "Customers want the freedom to innovate on familiar enterprise-grade platforms. Our collaboration with NVIDIA enables AI workloads to coexist with existing applications without sacrificing performance, efficiency, or stability."
“The rise of generative AI is driving a new demand for accelerated computing infrastructure,” said Justin Boitano, vice president of Enterprise AI at NVIDIA. “By leveraging Blackwell GPUs and NVIDIA networking technology, enterprises can build powerful AI applications directly within their existing private cloud environments and leverage the comprehensive capabilities of the VCF platform.”
Core technology integration and advantages
Specific highlights of this collaboration include:
• Support Blackwell graphics architecture GPU:CoverRTX PRO 6000 Server EditionIn the future, it will also support Blackwell B200 GPU to cope with large-scale AI and HPC computing and workloads.
• High-speed networking and DirectPath I/O:The NVIDIA ConnectX-7 NIC and BlueField-3 400G DPU are introduced, supporting GPUDirect RDMA and GPUDirect Storage technologies to accelerate multi-node AI training and data transmission.
• Retain core VCF functionality:The new hardware will coexist with existing VCF features such as vMotion, HA, DRS, and Live Patching, ensuring seamless parallel operation of AI/ML and traditional enterprise applications.
Enhance the flexibility of private cloud AI deployment
These integrations mean that businesses can leverage the latest NVIDIA hardware and network performance when implementing next-generation AI applications without completely rebuilding their infrastructure, while maintaining VCF's existing operational processes and enterprise-grade reliability. The in-depth collaboration between Broadcom and NVIDIA will further promote the large-scale implementation of private cloud AI, enabling AI to move beyond laboratory technology and become a true part of daily enterprise operations.
A partnership between friends and foes
While Broadcom and NVIDIA currently compete in the AI application market, particularly in the field of network connectivity for AI servers, they are also partners in certain circumstances. VMware, even before its formal acquisition by Broadcom, established a deep partnership with NVIDIA, leveraging NVIDIA's GPU acceleration and networking technologies in virtualization architectures and cloud infrastructure services. This partnership continued even after the VMware acquisition.
From the perspective of market supply and demand, NVIDIA has continued to define AI server architecture in recent years. Through the acquisition of Mellanox, it has used its Spectrum Ethernet platform to build a server network connection system that is more suitable for NVIDIA GPU acceleration, thereby taking a share of the server network system market demand that was previously dominated by Broadcom. In particular, it proposed this yearNVLink Fusion Project, further impacting Broadcom's existing market share of server technology applications. Therefore, Broadcom continued to emphasize the use of Ethernet open architecture to meet more diverse market deployment needs, thereby highlighting the competitive flexibility of its products that are not limited to specific fields.
However, given the current popularity of the generative AI market, a significant portion of the market is shifting towards NVIDIA solutions. NVLink Fusion can even be used to build AI server architectures that better meet deployment requirements, resulting in a continued impact on Broadcom in this market.
However, as many companies still widely use VMware's cloud-based services and also have a high demand for NVIDIA's GPU-accelerated computing applications, Broadcom continues to maintain a deep collaboration with NVIDIA, forming a partnership in which they are both competitors in the market and partners in commercial applications.
