The rapid shift toward agentic AI has introduced critical operational, cost, and governance challenges. Unlike linearly growing inference usage by users, autonomous agents generate non-linear inference usage and can quickly overwhelm IT infrastructure. Performance is a critical aspect of today’s agentic era.
Announcement
Today, we are announcing VMware vSphere 9.1 (and all future vSphere 9 releases) is now an NVIDIA -Certified Hypervisor, with near bare metal performance for AI workloads. This certification enables customers to confidently choose VMware Cloud Foundation (VCF) as the performance-optimized private cloud platform to power AI and accelerated computing applications in enterprise data centers and AI factories. By accurately exposing hardware topology and implementing key performance optimizations, VCF achieves near-bare-metal performance for AI workloads.
Details of the Certification
Optimized Performance with Virtualization Benefits of VCF
NVIDIA-Certified vSphere 9.1 delivers near bare-metal performance for representative AI and accelerated computing applications. Deploying AI and accelerated applications on vSphere in VCF makes datacenter management easy and efficient because of the virtualization benefits of VCF.
Workload-Based Validation
Certification testing is based on representative performance-critical behaviors across compute, memory, data-path efficiency, and LLM inference as defined by NVIDIA. Below listed are the various test categories vSphere 9.1 was validated against and the corresponding Use cases these tests are relevant to.
| Test Category | Use Case |
| GPU Collective Communication | Distributed AI training and fine-tuning, multi-GPU inference,communication-intensive HPC |
| GPU, CPU, andMemoryBandwidth | Data-intensive AI training and fine-tuning; high-throughput or batch inference,preprocessing, and analytics |
| ComputePerformance | AI training and fine-tuning, HPC simulation and scientific computing, GPU-accelerated analytics. |
| AI Inference | Gen AI, high-throughput and batch model serving large-scale LLM andmultimodal inference, agentic AI applications |
| RDMA DataPath | Multi-node AI training, distributed large-scale inference, MPI and network-bound HPC |
| Topology andDevice-MappingEvidence | Cross-cutting evidence for multi-GPU/multi-node training, scale-out inference, and HPC; not a workload category |
Note that vSphere in VCF met the performance thresholds set by NVIDIA for all the above test categories while providing the virtualization benefits of VMware Cloud Foundation. For further details, please refer to the NVIDIA Technical Whitepaper published here.
Enterprise-Grade
VCF is the industry’s first unified private cloud platform that combines the scale and agility of public cloud with the security, resilience, and performance of private cloud, delivering increased productivity and lower total cost of ownership. The NVIDIA-Certified Hypervisor program provides a clear, reliable way for enterprise customers to identify VCF as one of the platforms that satisfies the rigorous demands of production AI deployments.
Ecosystem Readiness
VCF is certified across the major NVIDIA GPU architectures—Blackwell and Hopper architectures—to support broad deployment flexibility across enterprises’ AI factories. With the NVIDIA Blackwell series GPUs, NVIDIA unlocks the potential of generative, agentic, and physical AI by delivering exceptional performance, efficiency, and scale for enterprises. The partnership between Broadcom and NVIDIA addresses the need for increased choice, flexibility, and operational efficiency for AI infrastructure, ensuring VCF customers can harness the full potential of NVIDIA’s comprehensive data center architecture.
Want to know more?
- Complete this form to contact us!
- Visit VMware.com/AIML for more information.
- Connect with us on Twitter at @VMwareVCF and on LinkedIn at VMware VCF.
Discover more from VMware Cloud Foundation (VCF) Blog
Subscribe to get the latest posts sent to your email.
