Deskside AI systems built on NVIDIA’s GB10 have one job, and they do it well: put 128GB of unified memory and a petaflop of compute on a desk close enough to the person using it that iteration is instant. What they don’t have is a connection to the storage an organization already runs. Every DGX
Brian recently sat down with Max Manekia, CEO of ServerPartDeals, at the company’s facility in Sanford, Florida, on the same trip that produced our ServerPartDeals facility tour. We’ve been a ServerPartDeals customer for years. Brian put them on our drive leaderboard not long after they opened their doors in 2010, and our 2025 look at
Simplyblock is building business continuity and storage-level disaster recovery into its software-defined storage platform for Red Hat OpenShift Virtualization and OpenShift Container Platform, with no separately licensed DR product on top. The platform now ships with synchronous stretch-cluster replication for zero-RPO metro deployments, asynchronous replication with RPOs as low as one minute, automated failback, live
Google Cloud has made its Storage Intelligence advisor generally available and expanded batch operations for Google Cloud Storage. The advisor monitors storage environments for operational anomalies, while the enhanced batch engine applies policy changes to up to 1,000 buckets per project in a single managed job. Both features are now available to all Storage Intelligence
IBM has added two beta capabilities to IBM Digital Asset Haven, its digital asset platform for banks, governments, and other regulated organizations: an on-premises deployment that runs entirely inside a client’s own data center on IBM Z or IBM LinuxONE, and connectivity to permissioned blockchain networks, starting with Swift’s blockchain-based shared ledger. The Swift piece
Synopsys has certified its digital and analog EDA flows on TSMC’s A14 process and completed 64G UCIe IP tape-outs on TSMC’s 2nm and 3nm nodes, the two hard milestones in a broader set of joint work the companies announced around TSMC’s Open Innovation Platform ecosystem. The rest of the payload covers agentic AI design flows
NVIDIA has launched the NVIDIA Open Agent Safety Platform, an open software platform and reference system design that puts guardrails around autonomous AI agents at two layers: a secure runtime on the host CPU and an out-of-band watchdog on the DPU that the agent can’t see. NVIDIA says the pattern in recent agent security incidents
AMD has posted the full SKU list for its 6th Gen EPYC 9006 series, the Zen 6 parts it launched as Venice at Advancing AI 2026 in July, and all of the processors offer 1Ku pricing guidance. There are 31 parts across the two sockets: nine for SP7, the 16-channel flagship platform, running from a
NetApp has agreed to acquire PEAK:AIO, the Manchester, UK software company whose metadata and parallel NFS work NetApp says will accelerate its AI infrastructure roadmap. NetApp plans to combine PEAK:AIO’s metadata services and parallel namespace technology with ONTAP, aiming at shared storage that scales alongside growing GPU clusters, in an architecture it says is intended
Supermicro is now shipping NVIDIA Vera Rubin NVL72 racks integrated with its Data Center Building Block Solutions (DCBBS) and its DLC-2 direct liquid cooling stack. The racks follow the VR NVL72 design NVIDIA launched at CES 2026: 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs in a single liquid-cooled rack that operates as one
Amazon