Supermicro Open Storage Summit 2026: 12 sessions on AI data architecture

The Supermicro Open Storage Summit 2026 will run from Aug. 11 through Sept. 3 as a 12-session theCUBE video series focused on storage and data architecture for enterprise AI. Supermicro and theCUBE are positioning the seventh annual program around a practical problem for infrastructure teams: fitting inference, agentic-AI workloads and existing enterprise data into systems that can be operated at scale.

Sessions will air at 10 a.m. PDT and be available on demand afterward. Registration is free, according to the event page. The program lists 38 industry leaders from 21 companies, including AMD, DDN, Hammerspace, Intel, MinIO, Nutanix, Scality, Solidigm, VAST Data and WD. Crusoe and Iron Mountain are also scheduled to offer practitioner perspectives.

What will the Supermicro Open Storage Summit cover?

The published agenda groups its discussions around decisions that storage, platform and AI teams are already making. One session will examine multi-tier storage for inference, including the performance and cost tradeoffs between an all-flash parallel file system and an object-storage layer based mainly on hard drives. WEKA, Scality, Samsung, WD and Supermicro are listed for that discussion.

Enterprise implementation is another major thread. A panel involving Nutanix, MinIO, PEAK:AIO and Supermicro is set to address the path from proof of concept to production, covering testing, deployment, token economics, scalable infrastructure, data readiness, governance and user onboarding. A separate vertical-AI session will use financial-market exchanges and genomics workflows to examine how requirements can differ by response time, data and workflow design.

The agenda also reaches beyond a new build. Hammerspace, Sandisk and Supermicro are scheduled to discuss connecting legacy capacity and new storage systems to AI workflows. For teams assessing cloud infrastructure options, the Crusoe session concerns neoclouds, providers that specialize in large-scale AI workloads and maintain an AI-specific focus. It will cover their services, scale and storage requirements alongside Supermicro, VAST Data and Kioxia.

Several sessions center on inference data paths. IBM, Kioxia and Supermicro are slated to discuss compute, storage and network optimization, plus KV cache and AI data-platform design. Another session with VAST Data, Solidigm and Supermicro will cover a proposed “context memory” layer for agentic AI. The concept is a storage tier intended to retain and retrieve KV-cache data, positioned between local GPU-system SSDs and network storage.

Data lakes, lakehouses and AI data platforms round out the published topics. Those sessions will consider AI-oriented data pipelines, inference workloads, enterprise-data preprocessing and systems that combine data management with GPU infrastructure. The event organizer says attendees will receive actionable recommendations, but the sessions are vendor and partner panels, so implementation claims should be assessed against each organization’s own workload, data controls and cost constraints.

This story draws on original reporting from SiliconANGLE.