While some of the top storage vendors pledged support for the new Nvidia AI Data Platform reference design for AI infrastructure and AI inference, others introduce a wide range of hardware and software aimed at expanding the overall AI ecosystem.
While the Nvidia GTC 2025 event this week is centered on what Nvidia is doing to build an AI ecosystem, including the introduction of its next-generation Blackwell Ultra GPU for AI data centers, it is also an opportunity for others in the industry to showcase what they bring to that ecosystem.
One key part of building AI infrastructures is the storage industry. The GPUs needed to process AI training and inference are data-hungry, requiring not only ever more capacity to store that data but also performance to match their requirements.
Therefore, it is no surprise that storage has become front and center at GTC 2025.
[Related: DDN CEO On The Company’s AI Mission And The ‘Essential Role’ Partners Play]
Nvidia itself made storage news at the conference with the Tuesday introduction of the Nvidia AI Data Platform, a customizable reference design that many storage developers are using to build what Nvidia calls a “new class” of AI infrastructure for AI inference workloads.
Leading storage vendors including DDN, Dell Technologies, Hewlett Packard Enterprise, Hitachi Vantara, IBM, NetApp, Nutanix, Pure Storage, Vast Data, and Weka, are working with Nvidia to build infrastructures to speed AI workloads with specialized AI query agents to help businesses generate insights from data in near real time working with Nvidia Blackwell GPUs, Nvidia BlueField DPUs, and other technologies.
Other storage vendors are also at the conference showing a wide range of hardware systems and components as well as software aimed at helping businesses take advantage of AI.
CRN here presents information on the latest storage offerings from 13 vendors looking to help build high-capacity and/or high-performance systems aimed at making sure the data pipeline to AI moves as quickly as needed.
Also look for CRN to examine how vendors outside the storage industry are helping build other parts of the AI infrastructure.
Graid Technology SupremeRAID AE (AI Edition)
Santa Clara, Calif.-based Graid used Nvidia GTC 2025 to introduce its SupremeRAID AE (AI Edition), which was designed to help enterprises running GPU servers and AI workloads deliver optimized data management with GPUDirect Storage (GDS) and an Intelligent Data Offload Engine. By enabling direct NVMe-to-GPU transfers, it helps eliminate bottlenecks and reduces latency for faster model training and inference. Its offload engine optimizes GPU utilization by handling data tasks, freeing GPU resources for AI processing. SupremeRAID AE supports NVMe-oF for scalable AI storage, and provides enterprise-grade RAID protection to help ensure uninterrupted access to critical datasets. Its flexible GPU deployment model allows enterprises to scale workloads while maximizing AI infrastructure performance.
Kioxia 122.88TB NVMe SSD
San Jose, Calif.-based Kioxia’s new enterprise-class Kioxia LC9 SSD is aimed at GenAI applications. The SSD is optimized for capacity, and includes a PCIe 5.0 interface and dual-port capability for fault tolerance or connectivity to multiple compute systems. It can help accelerate AI model training, inference, and Retrieval-Augmented Generation (RAG) at scale. Kioxia claims the LC9 drives are the first enterprise QLC SSDs to use 2Tb QLC BiCS FLASH 3D flash memory 8th generation technology.
Pure Storage FlashBlade//EXA
Pure Storage FlashBlade//EXA is a new data storage platform aimed at meeting the requirements of AI and high performance computing (HPC). The new platform, from Santa Clara, Calif.-based Pure storage, helps provide multidimensional performance with massively parallel processing and scalable metadata IOPS to support high-speed AI requirements, with performance of 10-plus terabytes per second in a single namespace. The platform also helps eliminate metadata bottlenecks with high metadata performance, availability, and resiliency for massive AI datasets with no manual tuning or additional configuration needed. Its configurable and disaggregated architecture uses industry standard protocols, including Nvidia ConnectX NICs, Spectrum switches, LinkX cables, and accelerated communications libraries. FlashBlade//EXA is slated to be available in summer 2025.
Cohesity Gaia On-Premises
Cohesity is expanding Cohesity Gaia, the enterprise knowledge discovery assistant, to deliver what the Santa Clara, Calif.-based company calls the industry’s first AI search capabilities for data stored on-premises. The new Cohesity Gaia will be available for use with Cisco UCS, HPE, and Nutanix platforms. The expansion will enable enterprises to uncover AI-powered insights by unlocking the full potential of their on-premises backup data. For enterprises adopting hybrid cloud strategies and looking to keep their valuable, critical data on-premises to meet security, compliance, and performance requirements, the expanded Cohesity Gaia can help them access more high-quality data and remain in control of their infrastructure. Availability is slated for mid-2025.
IBM Content Aware Storage In Fusion
IBM’s new content-aware storage (CAS) capability helps enterprises extract the meaning hidden in their rapidly growing volumes of unstructured data for inferencing, without compromising trust and safety, to responsibly scale and enhance AI applications like retrieval-augmented generation (RAG) and AI reasoning. IBM Storage Scale responds to queries using the extracted and augmented data to help speed communications between GPUs and storage using Nvidia BlueField-3 DPUs and Nvidia Spectrum-X networking. The multimodal document data extraction workflow also leverages Nvidia NeMo Retriever microservices, built with Nvidia NIM. Availability is slated for the second quarter of 2025.
DDN Infinia Data Ocean
DDN, Chatsworth, Calif., used Nvidia GTC 2025 to introduce its Infinia Data Ocean, a platform aimed at unifying data across edge, core, and cloud environments. The platform aims to simplify AI data management by reducing silos and improving data mobility. It provides tools for real-time data processing, multi-tenant security, and seamless AI application integration. Key features include automated AI data pipelines, low-latency architecture, and cloud-native scalability. The platform is designed to support enterprises managing complex AI workloads, offering security, efficiency, and fault-tolerant data storage. Infinia Data Ocean targets the increasing demands of AI infrastructure across various industries.
Hitachi iQ M Series
Santa Clara, Calif.-based Hitachi Vantara introduced the Hitachi iQ M Series, the latest addition to the Hitachi iQ portfolio of AI-ready infrastructure solutions. This new offering lowers the entry cost for AI, with built-in adaptability and scalability as customer needs evolve. Integrating accelerated computing platforms with robust networking, Hitachi iQ M Series combines Hitachi Vantara Virtual Storage Platform One (VSP One) storage, integrated file system choices, and optional Nvidia AI Enterprise software into a scalable and adaptable AI infrastructure package. The Hitachi iQ M Series allows for a choice of Nvidia accelerating computing platforms to select the most suitable GPU for specific workloads, and allows for the scaling of compute and storage independently with the flexibility to adapt to diverse and fluctuating data sizes, data types, and workloads.
Pliops/vLLM Production Stack Collaboration
Santa Clara, Calif.-based Pliops and the open source vLLM Production Stack project have collaborated on technology to deliver high performance and efficiency for LLM inference. Pliops brings expertise in shared storage and efficient vLLM cache offloading, while LMCache Lab brings its scalability framework for multiple instance execution. The combined offering leverages Pliops’ KV (key-value) storage backend to increase performance and scalability in AI applications. Pliops’ XDP LightningAI, an accelerated KV distributed smart node, introduces a new petabyte tier of memory below HBM for GPU compute applications, and connects to GPU servers by leveraging the NVMe-oF storage ecosystem to provide a distributed KV service.
MinIO AIStor + NVIDIA
To further support the growing demands of modern AI workloads, Redwood City, Calif.-based MinIO is adding advancements to its AIStor AWS S3-compatible object store technology that deepen its support for the Nvidia AI ecosystem. The new integrations include support for Nvidia GPUDirect Storage (GDS) for object storage to deliver a significant increase in CPU efficiency on the Nvidia GPU server, native integration with Nvidia BlueField-3 networking platform to help drive down object storage total cost of ownership, and incorporation of Nvidia NIM microservices into MinIO’s AIStor promptObject inference to help simplify deployment and management of inference infrastructure and deliver faster inference via model optimizations for Nvidia hardware.
NetApp AFF A90 Validated for NVIDIA DGX SuperPOD
The new NetApp AFF A90 enterprise storage systems with NetApp ONTAP from San Jose, Calif.-based NetApp are now validated for Nvidia DGX SuperPOD. NetApp, by bringing what it terms the most secure storage on the planet to DGX SuperPOD deployments, gives customers the enterprise data management capabilities and scalable multi-tenancy they need to develop and operate high-performance AI factories and deploy agentic AI while eliminating data silos.
Vast InsightEngine with Nvidia DGX
New York-based Vast Data used Nvidia GTC 2025 to introduce the Vast InsightEngine with Nvidia DGX, a new converged system aimed at bringing enterprise customers a real-time AI data processing and retrieval platform designed to simplify AI deployments while delivering fast, scalable, and secure data services. Vast InsightEngine with Nvidia DGX converges instant automated data ingestion, exabyte-scale vector search, event-driven orchestration, and GPU-optimized inferencing into a single system with unified global enterprise-grade security to help businesses efficiently accelerate AI adoption and time-to-value by unlocking AI-driven insights. It is a pre-configured, fully integrated real-time AI stack to help ensure seamless data flow and scalable AI inferencing at exabyte scale.
Solidigm Liquid-Cooled Enterprise SSDs
Rancho Cordova, Calif.-based SSD manufacturer Solidigm unveiled one of the world’s first liquid-cooled enterprise SSDs (eSSDs) at Nvidia GTC 2025, designed to eliminate fans and enable fully liquid-cooled AI servers. Traditional SSD cooling has limited server design flexibility, but Solidigm’s cold-plate cooled eSSDs—developed with Nvidia—overcomes challenges like hot swap-ability and single-side cooling constraints. The technology features the D7-PS1010 E1.S 9.5mm form factor, which improves thermal efficiency and data center serviceability. The offering will be available in the second half of 2025. The company also said a 15mm air-cooled version is also launching for broader server compatibility.
Vdura V11 Data Platform and V500 All-Flash Appliance
The V5000 all-flash appliance from Milpitas, Calif.-based Vdura was engineered to address organizations’ increasing demands as AI pipelines and generative AI models move into production. The all-flash appliance is built on Vdura’s V11 data platform, and delivers GPU-saturating throughput while ensuring the durability and availability of data for 24x7x365 operating conditions. The update is aimed at helping organizations be confident that their AI infrastructures are scalable and reliable.
