VDURA Launches All-Flash NVMe V5000 to Power AI Factories and GPU Clouds with Unmatched Performance, Scale, and Reliability

VDURA  announced the launch of its V5000 All-Flash Appliance, engineered to address the escalating demands as AI pipelines and generative AI models move into production. Built on the revolutionary VDURA V11 data platform, the system delivers GPU-saturating throughput while ensuring the durability and availability of data for 24x7x365 operating conditions, setting a new benchmark for AI infrastructure scalability and reliability.

VDURA V11 Software is now available on the new ‘F Node,’ a modular 1U platform with Intelligent Client-Side Error Coding. It delivers over 1.5PB per rack unit and offers seamless scalability and reliability.

Effortless Scalability for AI Workloads

Unlike traditional AI storage solutions that require overprovisioning, VDURA enables AI service providers to expand dynamically, scaling from a few nodes to thousands with zero downtime. As GPU clusters grow, V5000 storage nodes can be seamlessly added—instantly increasing capacity and performance without disruption. By combining the V5000 All-Flash with the V5000 Hybrid, VDURA delivers a unified, high-performance data infrastructure supporting every stage of the AI pipeline—from high-speed model training to long-term data retention.

Designed for AI at Scale: Performance, Density & Efficiency

The VDURA Data Platform’s parallel file system (PFS) architecture is engineered for AI, eliminating bottlenecks caused by high-frequency checkpointing. VDURA’s Intelligent Client architecture ensures AI storage remains high-performance and reliable at scale, implementing lightweight client-side erasure coding without adding unnecessary CPU overhead—unlike alternative solutions that impose heavy compute demands.

By combining client-side erasure coding, RDMA acceleration, and flash-optimization handling, the V5000 sustains AI workloads at any scale, delivering:

• Seamless AI Storage Expansion – Scale effortlessly as GPU clusters grow, with no downtime or overprovisioning.
• AI Checkpoint Optimization – Eliminates write bottlenecks that slow AI training.
• Data Center Efficiency – 1.5PB+ per U, reducing power, cooling, and footprint costs.
• RDMA Acceleration – Architected for NVIDIA GPU Direct and RDMA, with optimizations rolling out later this year.

“The V5000 All-Flash Appliance, powered by our next-gen V11 Data Platform, delivers not only the peak performance enterprises expect—but while high performance is necessary, it is not enough,” said Ken Claffey, CEO of VDURA. “AI workloads demand sustained high performance and unwavering reliability. That’s why we’ve engineered the V5000 not just to hit top speeds but to sustain them—even in the face of hardware failures—ensuring every terabyte of data fuels innovation, not inefficiencies or downtime.”

Radium Selects VDURA to Power Massive-Scale AI Training

As AI models grew in complexity, Radium, an NVIDIA Cloud Partner, required a co-located storage solution that could scale dynamically with its expanding GPU infrastructure without increasing the data center footprint or energy consumption. After evaluating multiple AI storage solutions, Radium selected VDURA for its unique combination of performance, efficiency, and seamless scalability.