Request a Demo
Join a 30 minute demo with a Cloudian expert.
Built for the AI Era.
Validated by NVIDIA.
Deploy NVIDIA-certified, exabyte-scalable object storage with native S3 API compatibility.
Cloudian’s growing collaboration with NVIDIA spans NVIDIA-Certified Storage, RDMA for S3-compatible storage, and the NVIDIA STX architecture — building the storage infrastructure the AI Factory demands.
Cloudian HyperStore is NVIDIA-Certified Storage, validated to meet NVIDIA’s rigorous performance and compatibility standards for AI infrastructure. That means faster model training, lower-latency inference, and a storage foundation to accelerate your AI workloads. No guesswork. No tuning. Just certified performance.
Integration Benefits
HyperScale AIDP: Your On-Premises AI Factory, Ready to Deploy
The Cloudian HyperScale® AI Data Platform is a fully integrated, on-premises AI system built on the NVIDIA AI Data Platform reference design. Pairing HyperStore’s S3-native object storage with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs and NVIDIA AI Enterprise software, it delivers a secure, data-sovereign platform for agentic RAG, semantic search, and enterprise knowledge retrieval — without sending data to the cloud. Process thousands of pages of documentation in minutes. Unlock the 80% of enterprise data that was previously inaccessible to AI.
RDMA for S3 Compatible Storage: GPU-Speed Data Access
Cloudian is a partner in RDMA for S3-compatible storage — enabling direct data transfers between object storage and GPU memory, bypassing the CPU entirely. The result: get TB/sec throughput in a single rack — 3× faster than non-RDMA flash — a 90% reduction in GPU server CPU utilization, and an 8× performance boost in vector database operations. Storage is no longer the bottleneck in your AI pipeline.
CMX: Extending AI Context Memory Beyond GPU Limits
As agentic AI reasoning sessions grow longer, KV cache sizes routinely exceed GPU memory, collapsing inference performance. The Context Memory Storage Platform (CMX), powered by BlueField-4 DPUs within the NVIDIA STX architecture, is a purpose-built storage tier for KV cache data — delivering up to 5× higher tokens-per-second and 5× greater power efficiency than traditional storage. Cloudian is proud to be a launch partner for NVIDIA STX, committed to delivering integrated solutions built on this architecture across all three STX system layers.
NVIDIA-Certified Storage: Validated for the AI Factory
Cloudian HyperStore 8.2.6 has achieved the Foundation level of NVIDIA-Certified™ Storage — validated to meet the essential demands of AI workloads involving up to 128 GPUs. The certification validates HyperStore’s performance across the precise I/O patterns AI workloads demand: sequential reads, random I/O, and low-latency retrieval. For enterprises building production-grade AI Factories, NVIDIA-Certified Storage means fewer unknowns, faster deployment, and proven confidence in the storage layer feeding your AI infrastructure.
Lenovo + Cloudian: A Proven Blueprint for NVIDIA AI Infrastructure
Cloudian and Lenovo have published validated reference architectures on NVIDIA accelerated computing platforms — proven blueprints for deploying scalable AI infrastructure with confidence. Delivering RDMA for S3 storage, integration with NVIDIA hardware and software, and full S3 compatibility, these designs have been validated with real-world AI workloads including model training, inference, and checkpointing. Reduce deployment risk, accelerate time-to-value, and scale from proof-of-concept to exabyte without rearchitecting.
The Storage Platform NCPs and Neoclouds Choose to Win
NVIDIA Cloud Partners and Neoclouds operate at the intersection of AI infrastructure and enterprise service delivery — and they can’t afford storage that slows them down. Cloudian HyperStore delivers the complete stack: true multi-tenancy across shared GPU clusters, native RDMA for S3 throughput exceeding 1 TB/sec in a single rack, NVIDIA Foundation Certification, geo-distributed multi-datacenter architecture, and government-grade FIPS 140-3 security. Software-defined and hardware-agnostic, Cloudian lets you start small and scale to exabytes.
See What HyperScale AIDP
Can Do For Your Industry
Every industry faces distinct challenges — from compliance and data sovereignty to speed and scale. These solution briefs show exactly how HyperScale AIDP solves them without months of integration work
FINANCIAL SERVICES
Empowering AI for Financial Services
Meet compliance demands and accelerate performance — on-premises AI delivers secure, high-speed intelligence for financial institutions.
Learn More
HEALTHCARE
Advancing Patient Care with Compliant AI
Protect patient privacy and accelerate clinical decisions with on-premises AI built to meet HIPAA and healthcare compliance demands.
Learn More
INSURANCE
Transforming Insurance Ops with AI
Slash claims cycle times and satisfy regulators with secure, on-premises AI that automates decisions without exposing sensitive data.
Learn More
LIFE SCIENCES
Secure AI for Life Sciences
Accelerate drug discovery and lock down IP with sovereign AI infrastructure engineered for the pace of life sciences research.
Learn More
MANUFACTURING
Manufacturing with On-Premises AI
Drive production excellence and protect proprietary process data with high-performance, on-premises AI at the factory edge.
Learn More
HIGHER EDUCATION
Higher Education AI Imperative
Advance breakthrough research and protect IP with on-premises AI that keeps sensitive data sovereign and drives student success.
Learn More
RETAIL
Retail Operations with Secure AI
Boost margins and reduce shrink with on-premises AI that analyzes customer behavior without putting sensitive data at risk.
Learn More
ENERGY
Optimizing Oil & Gas with Secure AI
Maximize asset yields and protect billion-dollar seismic data with secure, on-premises AI built for energy operations.
Learn MoreRelated Resources
SOLUTION BRIEF
Cloudian HyperScale® AI Data Platform
Cloudian HyperScale® AI Data Platform helps you fast-track AI adoption without compromising on data security, sovereignty, or accessibility.
Download Solution Brief
SOLUTION BRIEF
NVIDIA GPUDirect for Object Storage
Cloudian HyperStore with NVIDIA GPUDirect delivers exabyte-scalable, 35GB/s performance, streamlining AI workflows by eliminating data migrations and reducing infrastructure costs efficiently.
Download Solution Brief
Reference Architecture
Lenovo / Cloudian Reference Architecture
Cloudian and Lenovo’s architecture uses GPUDirect S3 RDMA to deliver high-performance, cost-effective object storage that maximizes GPU throughput for AI workloads.
Learn More
Reference Architecture
Supermicro / Cloudian Reference Architecture
Supermicro and Cloudian’s architecture leverages S3 RDMA to deliver high-performance, cost-effective object storage that maximizes GPU throughput for AI.
Learn More
WEBINAR
Transform Unstructured Data into AI Intelligence
Learn how NVIDIA and Cloudian simplify AI infrastructure, transforming unstructured data into actionable intelligence through high-performance, S3-compatible storage and compute.
Learn More
