Cloudian and NVIDIA partner to deliver the first on-premises AI server with cloud-scale performance.
Approximately 80% of enterprise data remains inaccessible to AI systems, hindered by data sovereignty restrictions and infrastructure complexity that force companies to choose between innovation and control over their most valuable assets.
The main issue involves a fundamental mismatch between how companies store data and how AI systems access it. Most enterprise data is stored in unstructured formats, such as emails, documents, medical images, sensor data, and more. Yet traditional AI infrastructure requires either moving this sensitive data to cloud services or investing in costly, complex on-premises AI stacks that demand specialized expertise, which many IT teams lack.
In regulated sectors like healthcare, finance, and government, this results in a painful tradeoff: sacrifice security and compliance to access AI capabilities or keep data on-premises and risk falling behind competitors who are using AI to drive faster insights.
Cloudian’s Strategic Move with NVIDIA
Cloudian has partnered with NVIDIA to address this challenge with the industry’s first fully on-premises AI server, offering cloud-scale performance. The new Cloudian HyperScale AI Data Platform, announced on September 30th, brings AI computing directly to enterprise data stores, eliminating the need for risky and costly data movement.
Built on NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, the solution debuts with S3-over-RDMA (Remote Direct Memory Access), providing an eightfold faster performance boost. The platform achieves 35GB/s per node on reads, scaling linearly to terabytes per second per rack, meeting the speed demands of generative AI workloads.
Key Technical Innovation
- Embedded vector databases: Unlike traditional systems that bolt on separate AI layers, Cloudian’s platform directly integrates vectorization into the storage layer. Data gets automatically classified and converted into vectors on write, enabling instant AI-readiness through standard S3 APIs.
- Direct data paths: Leveraging NVIDIA AI Enterprise software and Spectrum-X Ethernet networking, the system connects storage, system memory, and GPU memory via ultra-low-latency pipelines. This architecture bypasses bottlenecks that often hamper enterprise AI projects.
- On-premises sovereignty: The HyperScale AI platform keeps all operations within the enterprise perimeter, ensuring compliance, security, and governance while harnessing the latest AI technologies.
Implications
For enterprises, this breakthrough could shift the balance in the ongoing debate between AI innovation and data governance. By enabling organizations to run large‑scale AI directly on sovereign data, regulated industries like:
- Healthcare can analyze vast amounts of patient records and imaging data without risking HIPAA violations.
- Financial firms can enhance fraud detection and risk modeling while remaining compliant with stringent data residency laws.
- Government agencies can unlock intelligence insights from historical and real-time data securely, without the data ever leaving the national infrastructure.
Industry analysts see this as a significant step toward “cloud‑like AI performance without the cloud.” Early adopters are expected to use the platform for vector search, real‑time recommendations, predictive analytics, and multimodal generative AI, all while ensuring full data residency.
This development may also intensify competition in the AI infrastructure market, where hyperscalers such as AWS, Microsoft, and Google have dominated by integrating AI with cloud data storage. Cloudian’s move, in partnership with NVIDIA, signals a new path forward for enterprises that demand both performance and sovereignty.
Here’s a short demo of the HyperScale AIDP.
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | TikTok | RSS Feed
