Cloudian

Cloudian Cloudian is the leader in hybrid object storage solutions. Enterprises and service providers deploy

08/20/2026

✨ KV Cache Offloading to Object Storage Transforms AI Inference ✨

Every return visit to an LLM conversation forces the model to recompute context from scratch, burning GPU cycles and slowing response times.

Cloudian's new benchmark, run with NVIDIA Dynamo, shows there's a faster path.

By offloading KV Cache to Cloudian HyperStore, an S3-native object storage platform with RDMA acceleration, we measured:

📊 Up to 20x lower Time to First Token vs. full recompute at 120K tokens
⚡️ A 31.77% mean latency advantage for RDMA vs. TCP across all context lengths tested
🚀 Flat, scalable retrieval latency — even as context windows grow to 128K+ tokens

The takeaway: the storage tier is becoming a first-class part of inference architecture. For multi-turn conversations, RAG pipelines, and long-context workloads, how fast you can recall model state is just as important as how fast you can compute it.

Read more for the full breakdown on methodology, results by context length, and what it means for GPU utilization at scale: https://bit.ly/452X4d0

08/13/2026

Egress fees. API charges. Cross-region replication costs. If you're storing petabytes on AWS S3, you already know the bill never stops growing.

We modeled the real 5-year total cost of ownership for 2.6 PB of S3-native storage - Cloudian HyperStore in a colocation data center vs. AWS S3 - using current 2026 pricing.

The result: 68% lower TCO with Cloudian, and a savings gap that only widens for AI/ML and high-access workloads.

See the full cost breakdown, capability comparison, and pricing assumptions in our latest report: https://bit.ly/4bxyTHb

08/06/2026

Neoclouds compete on GPU economics, but there's a hidden margin killer: storage that can't keep pace. When your storage layer stalls, GPUs sit idle - and that idle time is capacity you've already paid for in capital, power, and cooling but can't bill for.

Our whitepaper, Neocloud Storage Blueprint - Architecting Object Storage for AI Training & Inference at Scale, breaks down:
📊 Why storage-induced GPU idle can cost $288K–$384K/year on just a single 100-GPU cluster
⚡ The 5 storage challenges unique to neocloud environments, from the "egress tax" to multi-tenant isolation
🏗️ A practical 5-layer blueprint for architecting (or evaluating) the storage layer beneath your GPU fleet
✅ A buyer's checklist to score any storage platform, including your own

Plus how Cloudian HyperStore, built on NVIDIA-Certified Storage validation, delivers up to 35 GB/s per node and over 1 TB/s in a single rack — with S3-native support so nothing needs to be rewritten.

If storage is quietly eating into your utilization and your margin, this is the read.

👉 Get the full whitepaper: https://bit.ly/4wem8Jd

07/31/2026

Your GPU margins are getting killed by overpriced storage. Neoclouds can’t deliver better GPU economics if all your training data and checkpoints sit on expensive, all-flash arrays.

Cloudian HyperStore fixes the math:
🔹Slashed TCO: Pair high-density HDDs with fast flash to match storage costs directly to workload needs
🔹Automated Tiering: Hot data stays on flash; cold data moves to HDD seamlessly
🔹Zero GPU Downtime: Feed compute pipelines with up to 35 GB/s throughput per node
🔹NVIDIA-Certified & S3-Native: Plug-and-play simplicity for AI infrastructure

Stop overpaying for performance you don't need—and protect your cost-per-GPU-hour.

👇 See how neoclouds tune storage economics with Cloudian:

https://bit.ly/4vUWGbv



07/01/2026

✨ AI success starts with the right data foundation. When 🇨🇭 Swiss IT provider Begasoft set out to build a sovereign AI data lake 🌊 for its Brandbot platform, it needed more than scalable storage.

Begasoft needed an S3-native foundation that could keep sensitive customer data entirely within Switzerland while supporting enterprise RAG, multi-tenant AI services, and future growth.

By choosing Cloudian HyperStore, Begasoft built an AI-ready platform that delivers:
✅ Swiss data sovereignty for regulated industries
✅ Native S3 compatibility for seamless AI integration
✅ Multi-tenant isolation for secure customer environments
✅ Linear scalability from PoCs to enterprise production

The result? A single AI data lake powering both AI/RAG workloads and large-scale data integration, enabling Begasoft to move from pilot projects to production deployments with confidence.

Read the full customer story to see how Begasoft is helping Swiss organizations embrace AI without compromising security, compliance, or data sovereignty.

https://bit.ly/3SvrDVM

06/09/2026

Enterprise AI doesn't have a model problem. It has a data problem. Data is spread across file and object silos, security requirements limit access, and public cloud AI costs can quickly spiral as workloads scale.

That's why we're excited to announce HyperScale® AIDP v1.1. This latest release expands Cloudian's turnkey on-premises AI platform with:

✅ Native support for NVIDIA AI Blueprints for Enterprise Document RAG and Video Search & Summarization
✅ Direct ingestion from both NFS file shares and object storage - no migration projects required
✅ End-to-end security that enforces user permissions across ingest, vectorization, retrieval, and inference
✅ Validation on NVIDIA-Certified platforms from both Supermicro and Lenovo

The result? Enterprises can deploy production AI on their own infrastructure, maintain data sovereignty, and reduce AI infrastructure costs by up to 60% compared to public cloud alternatives.

If you're looking to move AI from pilot to production while keeping control of your data, this announcement is worth a read.

Read more: https://bit.ly/4oe2ydE

06/03/2026

In an era where speed and efficiency are paramount, financial institutions must adapt or risk obsolescence. By embracing on-premises AI solutions, financial institutions not only safeguard sensitive data but also enhance their operational efficiency, laying the groundwork for a future where innovation drives growth.

The power of AI is on the horizon—are you ready to seize it?

Learn more at cloudian.com

06/01/2026

Agentic AI is creating new security challenges for enterprise infrastructure teams.

As AI agents gain access to data, context memory, and distributed inference environments, organizations need security that operates at AI speed—not after-the-fact monitoring.

That's why Cloudian is extending its collaboration with NVIDIA to support NVIDIA Vera BlueField-4 STX architecture and new NVIDIA DOCA-powered security innovations within HyperStore.

Key capabilities include:
🔹 In-silicon AI-native data protection
🔹 Secure context memory isolation for multi-tenant AI environments
🔹 Continuous AI agent governance and threat detection
🔹 Runtime threat detection up to 1,000x faster than traditional agentless approaches
🔹 Policy enforcement at up to 800Gb/s line rates

By combining exabyte-scale S3-native object storage with security enforced directly in the data path, enterprises can build a stronger foundation for secure, scalable agentic AI.

Read the announcement to learn how Cloudian and NVIDIA are helping organizations protect data, context, and AI agents across the AI factory.

https://bit.ly/3Sco0Ud

05/28/2026

As enterprises build AI factories, a familiar problem is re-emerging: storage silos. One cluster for training. Another for RAG. Another for regulated workloads. Another for inference.

The result? Rising operational complexity, stranded capacity, and higher costs.

In this blog, explore why multi-tenancy is the "way out" of this silo problem. It is becoming essential for scalable enterprise AI because it allows organizations to consolidate AI workloads onto a single high-performance storage platform while maintaining security, isolation, compliance, and performance guarantees.

https://bit.ly/4f3wab5

A great read for anyone building or scaling AI infrastructure.

05/21/2026

Everyone is talking about AI models. But enterprise leaders are discovering the real differentiator is data...specifically, who controls it, where it lives, and how fast it can be accessed securely.

New research from 203 enterprise IT decision-makers reveals:
• 79% of organizations are already moving some AI workloads on-premises or to private infrastructure
• 58% have delayed or scaled back AI initiatives over concerns about sensitive data leaving their control
• Cloud AI costs are exceeding projections for many enterprises

As NVIDIA’s Daniel Glogowski explains: “Sovereignty isn’t a niche requirement anymore — it’s a tier-one enterprise concern.”

That challenge is reshaping enterprise AI strategy. Organizations across financial services, healthcare, legal, and insurance are increasingly prioritizing data sovereignty, predictable economics, and low-latency performance — driving a major shift toward hybrid and on-prem AI infrastructure.

This article from Data Center Dynamics dives into why enterprises are rethinking AI infrastructure, how hybrid AI models are evolving, and why bringing AI to the data may become the defining architecture of enterprise AI.

Read the full article to explore the survey findings and expert insights from Cloudian and NVIDIA.
https://bit.ly/4eUmsb1

Address

San Mateo, CA

Opening Hours

Monday 10am - 6pm
Tuesday 10am - 6pm
Wednesday 10am - 6pm
Thursday 10am - 6pm
Friday 10am - 6pm

Website

Alerts

Be the first to know and let us send you an email when Cloudian posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Contact The Business

Send a message to Cloudian:

Shortcuts

Share