Alation announced six new products for its enterprise AI operating system AIOS, including a first-of-its-kind product to enable real-time monitoring of enterprise AI agent compliance that helps solve one of the biggest roadblocks to AI agent performance in large companies today:
- AI Governance expands to add agent lineage tracing, enabling visibility into every AI agent's regulatory risk, all the way through the live data it consumes. There are now six native connectors bringing every major AI agent and model into a cross-platform registry: Amazon Bedrock, Amazon SageMaker, Databricks MLflow, Microsoft Copilot Studio, Microsoft Foundry, and Snowflake Cortex. It is the first cross-platform registry to continuously connect an agent's compliance posture to the live quality and policy status of its underlying data.
- Console is a single intelligent interface that routes natural language requests to the right AIOS capability without requiring users to know where to go. With AIOS, your intelligent applications operate like a single integrated system – the same is now true of the operating system that manages them.
- Governed Collections makes unstructured data from outside the catalog, including compliance policies, standard operating procedures, and business glossaries, available to agents as governed, up-to-date context. Documents from Confluence, SharePoint, and S3 become governed catalog objects and agents consume them by reference. When a source document changes, the agent reads the current version automatically.
- Intelligent Feeds deliver governed, interactive intelligence, combining metrics, visualizations, narrative, and chat to the teams responsible for recurring operating decisions. Data intelligence stops being a destination users visit and becomes something that finds them and enables them to dig as far as their curiosity takes them.
- Ontologies are a new enterprise knowledge object that guide agents through business processes with precision. Agents are only as reliable as their understanding of the business. Where an agent would otherwise infer the rules and constraints of a process, ontologies make them explicit: a governed, machine-readable model of how the business works, mapped to the catalog and reviewed by subject-matter experts. Built on open standards, ontologies can be exported and consumed by any agent runtime.
- Semantic Model Mastering ingests semantic models from Databricks and Snowflake, enriches and governs them in Alation, and syncs the enriched definitions back to both platforms automatically. A term like 'active customer' resolves to one definition, and every downstream agent runs on it. Every downstream consumer, including the agents that run on them, operates on the highest quality, most current definition without anyone re-entering Alation.
More information can be found in a blog. …
Reuters reports Chinese DRM fabber CXMT is preparing to build NAND chips as well as DRAM.
…
Content collaborator Egnyte introduced the Egnyte Context Layer that maps relationships across content, people, projects, business systems, and organizational knowledge. This gives AI agents a purpose-built context map upfront, eliminating redundant processing and reducing costs while helping them deliver more accurate, relevant outcomes faster. The Egnyte Context Layer powers task-based agents across workflows, including industry-specific workflows across financial services, life sciences, media and entertainment, and architecture, engineering, and construction (AEC). Learn more here.
…
Hitachi Vantara announced a commitment to reach net-zero emissions across its global value chain by FY2040, with near and long-term targets independently validated by the Science Based Targets initiative (SBTi). It has committed to:
- Cut absolute Scope 1 and 2 emissions by 98% by FY2030, against an FY2024 baseline.
- Reduce Scope 3 emissions per usable petabyte of storage sold by 97% by FY2040, against the same baseline.
Hitachi Vantara reduced its Scope 1 and 2 emissions by 43% in FY2025. Customers are also reporting efficiency gains, including DestekBank, which cut data centre energy consumption by 25%, and Malayala Manorama, which reduced power and cooling costs by 70%.
…
Tel Aviv, September 15, 2026 – Keewano announced general availability of KeewanoDB, the first database architected from the ground up for machine reasoning at scale. The company also revealed total seed funding of $12M, led by Hetz Ventures with participation from a16z speedrun, Remagine Ventures, DIG Ventures and angel investors. KeewanoDB stores information about every action a user, device, or AI agent undertakes, and does so in the way machines need it: live, complete, in order, and with context. Analysis runs inside the database, processing trillions of events in parallel, with agents receiving compact results, grounded in the context required for true reasoning. compact results, grounded in the context required for true reasoning.
…
Quantum has appointed Mike Heuer as its Chief Information Officer, to lead Quantum’s global information technology organization, including enterprise applications, data, AI adoption, infrastructure, and cybersecurity. Heuer most recently served as Chief Information Officer of Legence, a Blackstone portfolio company, where he built the company's IT function and supported its growth through more than 20 acquisitions. Prior to Legence, he was Head of Global Information Technology at Blackhawk Network, where he led the deployment of enterprise ERP and CRM platforms across North America, Europe, and Asia.
…
Redpanda, alongside Aiven, Confluent, StreamNative and Ververica, announced the formation of e Streamhouse Working Group to establish an open, vendor-neutral category of data architecture for AI agents and real-time applications. As businesses give software more authority, agents can’t afford to operate on stale data – they need a continuously current, governed view of the business. A Streamhouse architecture is designed to provide this, keeping the current state of a business available to applications and AI agents at production service levels, without requiring enterprises to first consolidate their data into a single system. The new working group will maintain an open, vendor-neutral definition of Streamhouse and advance the technologies, standards and interoperability practices behind it. The aim is to give the industry a shared and maintained architecture for this shift, while ensuring the category remains open and no single company controls it. More information here.
…
Server/storage system virtualizer StorMagic and Mako Networks, a provider of secure, cloud-managed networking systems for distributed enterprises, are partnering to help organisations build more resilient and manageable IT environments across remote and distributed locations. The partnership combines StorMagic SvHCI, a purpose-built hyperconverged infrastructure platform that provides highly available compute, storage and virtualisation at the edge, with Mako Networks’ secure SD-WAN and cloud-managed networking platform, which provides PCI-certified centralised network management, secure connectivity and failover capabilities.By combining edge networking and edge compute that are simple to deploy, remotely managed and resilient by design, StorMagic and Mako Networks deliver data-driven outcomes that are meaningful to next-generation retailers’ bottom lines.
…
StorONE announced that Engage2Excel, a global workforce solutions provider, has expanded its StorONE deployment to support production storage, backup, Virtual Tape Library (VTL), business continuity, disaster recovery, and long-term retention. Engage2Excel manages more than one petabyte of enterprise data on StorONE across different StorONE systems. Up to 9x more effective storage from existing flash is made possible by StorONE's FlashSpan (formerly known as Real Time Tiering) technology. FlashSpan continuously analyzes data access patterns and intelligently places data on the most appropriate storage media, ensuring flash capacity is continuously reserved for the data that benefits most.
…
HDD supplier WD (Western Digital) says that, with so much attention paid to overcoming compute restraints in AI, the next infrastructure challenge looms large: where and how to store all this compounding and persistent data? To help answer that question, IDC surveyed over 700 IT and business decision makers in the AI infrastructure/hyperscale/data storage space. The research found that:
- 95% of organizations are storing more data because of AI
- 95% say that data is becoming more valuable
- Nearly 75% are retaining data longer
- Over 75% are bringing more archived data back online for AI workloads, effectively blurring the line between “cold” historical data and active data.
As organizations generate, retain and reactivate more data, storage capacity, accessibility and cost are becoming imperative infrastructure decisions. More information in a WD blog.
…
Yugabyte just reported its strongest first half in company history, with new paying customers up 125%+ YoY and more than 900 companies down YugabyteDB. The company is also seeing early traction around Meko, its agent-native context engine launched in May, with the number of databases created by agents each day growing 10x. Yugabyte has also expanded its leadership team with key hires across product engineering, and go-to-market to support its global expansion, including Sonny Purewal as VP Sales EMEA, and Linson Mathai as VP GTM India. The momentum comes as Yugabyte extends its product portfolio and global footprint with a newly expanded partner program to support enterprises moving AI and agentic applications into production.
…
SW RAID supplier Xinnor announced the results of a PoC project carried out at the Weizmann Institute of Science demonstrating a new approach to Tier-0 storage for HPC and AI: using xiRAID Opus to transform NVMe already installed inside GPU and compute nodes into a shared, protected storage resource. Modern GPU servers contain significant amounts of high-performance NVMe drives, but that capacity is typically tied to individual nodes and used as local scratch storage. Xinnor’s xiRAID Opus disaggregates local NVMe drives across the cluster and composes those resources into protected, software-defined storage pools that can be accessed from any participating node. This allows organizations to turn storage capacity already present in their compute infrastructure into shared storage, while adding RAID protection, multipathing, failover, and flexible resource placement in software.
The implementation used IBM Storage Scale, the established file-system environment at the Weizmann Institute of Science, above the xiRAID Opus block-storage layer to provide the shared file system and global namespace. With xiRAID Opus providing the distributed block layer beneath IBM Storage Scale, the system reached 131.7 GiB/s sequential read throughput, effectively matching the same performance measured at the raw NVMe-oF layer. Sequential write reached 46.9 GiB/s, retaining approximately 98.5% of the raw block-level bandwidth. For random 4KiB IO impressive 5.8M IOPS were achieved for read and 3.1M IOPS for write.
Using Nvidia GPUDirect Storage, aggregate reads from the shared file system into GPU memory reached 132.1 GiB/s across five nodes — effectively the same bandwidth as the maximum sequential read result of the shared storage system. Read a technical deep dive here. …