SC25: Qumulo Announces Global Data Supply Chain for AI Factories

Qumulo, a player in unstructured data management, unveiled three major innovations that redefine how enterprises integrate their data into AI factories. Together, these advancements – Qumulo Helios Agent, Qumulo Cloud AI Accelerator, and Qumulo AI Networking – provide customers the oversight, scalability and speed needed to innovate on their terms, any data, any location, with total control.The Future of the Enterprise Data Platform
Qumulo Helios, Cloud AI Accelerator, and AI Networking offer intelligent, distributed, and autonomous data infrastructure. They extend Qumulo’s ability to deliver the only platform capable of unifying EB-scale data across every cloud, edge, and data center environment, while continuously learning, optimizing, and accelerating the workflows of the world’s most data-intensive organizations.

“Enterprises in today’s competitive environments need more than just storage-they need systems that think, adapt, and accelerate,” said Douglas Gourlay, CEO, Qumulo. “Helios gives our customers predictive awareness of their entire data ecosystem, Cloud AI Accelerator puts their data into motion wherever insight is needed, and AI Networking redefines what’s possible in performance. This is the foundation for the next gen of reasoning infrastructure and competitive advantage.”

Helios: AI-Ops Agent for Enterprise Data Systems
Helios, Qumulo’s AI agent, integrates system-wide telemetry to provide a self-managing, self-diagnosing, and self-optimizing data environment. Built into Qumulo’s Data OS, Helios continuously analyzes billions of operational events per day across hybrid and multi-cloud infrastructures. It identifies emerging anomalies, predicts capacity or performance issues before they occur, and automatically generates prescriptive recommendations or remediation workflows.

Helios represents the first gen of machine-coordinated platform intelligence for unstructured data. It fuses telemetry from compute, storage, cloud, and network layers into a unified model that enables proactive reasoning. The result is a system that not only automates troubleshooting and debugging but anticipates architectural and workload challenges-helping enterprises stay ahead of potential disruptions. With MCP support, Helios extends its reach into Qumulo’s partner ecosystem, allowing external agents and orchestration frameworks to participate in the same reasoning fabric, creating a truly autonomous data platform.

Qumulo Cloud AI Accelerator: Accelerating Insight Across Every Cloud
The Qumulo Cloud AI Accelerator bridges the company’s Cloud Data Fabric with public cloud compute environments to create a frictionless path for AI and analytics workloads. Leveraging Qumulo’s NeuralCache technology, it enables intelligent, predictive prefetching and accelerated data streaming between on-premises and public cloud environments. Whether data originates in an enterprise data center, a sovereign edge enclave, or a cloud region, Cloud AI Accelerator ensures that data moves only when and where it is needed-with minimal latency and no manual orchestration.

Through this accelerator, Qumulo allows customers to connect EB-scale datasets to hyperscale compute resources from AWS, Azure, Google Cloud, Oracle Cloud Infrastructure, or emerging AI service providers. Cloud AI Accelerator dynamically optimizes data pathways, ensuring that hot datasets for AI training, inference, or rendering are delivered to the right accelerator clusters at the right time. It transforms the economics of AI, eliminating the duplication and idle replication that often plague cloud-based AI architectures.

Qumulo AI Networking: Redefining Performance for Accelerated Computing
The third innovation-Qumulo AI Networking-introduces a new gen of high-performance data movement protocols optimized for GPU-based AI training and inference environments. Qumulo now natively supports RDMA, RDMA over Converged Ethernet v2 (RoCEv2), and NFS over RDMA, with S3 over RDMA in active development. These transports provide near-memory bandwidth between storage and accelerated compute clusters, dramatically reducing latency and CPU overhead for large-scale AI operations.

By extending the Cloud Data Fabric with these high-speed, low-latency data paths, Qumulo enables seamless integration with NVIDIA DGX, AMD Instinct, and other GPU-rich compute infrastructures. The result is unprecedented throughput for model training, simulation, and AI reasoning workloads-bringing the data fabric as close as possible to the computational fabric. Qumulo AI Networking unifies the traditionally siloed world of storage and comp into a single, coherent performance domain, accelerating innovation across scientific research, media production, genomics, and advanced industrial design.

Qumulo’s new capabilities are available in preview for select customers starting today, with availability across the next quarter.