Rafay Systems has secured certification under the NVIDIA-Certified Hypervisors program, validating its Virtual Machines-as-a-Service solution for high-performance AI workloads. The designation confirms that Rafay’s platform maintains near bare-metal efficiency while running on NVIDIA HGX and NVL72 rack-scale systems, supporting complex multi-tenant AI environments.
As enterprises and cloud operators scale their AI infrastructure, the ability to slice expensive GPU resources into secure, manageable segments has become a priority. The certification evaluates critical performance metrics, including data-path efficiency and large language model inference, ensuring that virtualized environments can handle the demands of modern accelerated computing.
Haseeb Budhani, CEO and co-founder of Rafay Systems, described the achievement as a bridge between raw hardware acquisition and scalable production utility. By integrating virtualization into its orchestration framework, Rafay allows operators to move beyond isolated GPU clusters toward shared, monetizable AI platforms. This approach provides developers with self-service access to resources while maintaining centralized governance over identity, quotas, and tenant isolation.
The certification covers a range of hardware, including NVIDIA HGX H200 and H100 systems, alongside the GB200 and GB300 NVL72 architectures. For infrastructure providers, this adds a layer of operational confidence, allowing them to offer virtual machines alongside bare metal and Kubernetes clusters from a unified control plane. Beyond basic provisioning, the platform supports the full lifecycle of AI services, including usage metering and chargeback models designed to maximize return on hardware investments.
Comments (0)
No comments yet. Be the first!