ETL vendor Adeptia has rolled out Automate 5.3, its latest agentic platform release. This update extends native MCP support, letting users and AI agents build integrations through natural language and leverage context-aware insights. It reduces reliance on deep technical skills, speeds partner onboarding and simplifies large-scale integration development.
A redesigned AI Mapping workflow enables developers, business users and AI agents to construct and validate mappings in plain language, cutting onboarding from days or weeks down to minutes. The AI learns continuously from mappings to become familiar with the organization’s data, terminology and business rules. Release 5.3 also brings BYO-LLM capability and the new Observe module for platform health monitoring.
Chelsio unveiled its seventh‑generation AI interconnect portfolio, covering SmartNICs, storage controllers and DPUs built to accelerate AI training, inference, storage and cloud networking. Built on its established Unified Wire architecture, the platform supports native 400GbE, unified RDMA (iWARP and RoCEv2), advanced storage offload and AI infrastructure tuning, while maintaining backward software compatibility with prior Chelsio generations.
The lineup splits into three categories under its seventh‑gen offload design: S7 SmartNICs, memory‑free single‑chip adapters optimized for cloud, virtualization, edge and AI; T7 storage controllers that offload NVMe/TCP, NVMe-oF, iSCSI and RDMA; and T7 DPUs, fully programmable units with native 400GbE for AI networking, storage and security acceleration.
CloudCasa by Catalogic released a feature update, headlined by expanded DR support for an additional enterprise storage array, native MCP integration for AI agent interoperability, and a preview of Changed Block Tracking for VM backups. The updated feature set is live, with CBT available in preview.
Storage‑integrated DR now adds Dell PowerStore alongside HPE Alletra Storage MP B10000, HPE Alletra 9000 and SUSE Storage. The storage system replicates persistent volume data to the recovery site, while CloudCasa orchestrates Kubernetes and KubeVirt VM failover and failback.
Its new MCP server allows AI agents to retrieve and report data protection and DR status. CBT backup tracks modified disk blocks since the last snapshot, drastically shortening backup windows.
Capital markets risk intelligence provider ActiveViam partnered with Databricks to deliver validated real‑time risk analytics directly on the Databricks Data + AI Platform. The collaboration helps financial institutions deploy AI agents built on governed enterprise data and trusted risk calculations.
UNSW Sydney will deploy DDN storage for its Research Data Experience program. Arcitecta’s MediaFlux data management software will pair with DDN’s EXAScaler Lustre parallel file system. After an 18‑month evaluation, the university will replace 11 legacy storage systems with this unified stack for high‑performance storage and enterprise data governance.
Arcitecta built an EXAScaler connector to feed filesystem metadata changes into MediaFlux. The architecture automates data movement between fast on‑campus cache and a central off‑campus repository, giving researchers low‑latency access while using cheaper HDDs for non‑performance‑critical data. The on‑premises platform also supports cloud workflows. The project kicked off September 1, 2026; hardware delivery is scheduled for early 2027 and full service launch later that year.
John McKillop has joined Hammerspace as VP of Americas Sales. He brings deep enterprise and AI technology leadership, including senior executive roles at Quantum Corporation. Tom Whaley, the prior Americas sales lead, departed in May for NetApp.
Tony Asaro, Hammerspace Chief Business Officer, cited strong year‑over‑year growth and said McKillop will help expand sales momentum, scale the sales team and grow partnerships to convert market traction into sustained revenue. McKillop noted the massive AI market opportunity and said his priorities are scaling sales, broadening strategic alliances and helping customers unlock distributed AI data value.
Hitachi Vantara, together with Dutch thermal specialist Duraflow, is running a pilot of one of the world’s largest phase‑change material cooling installations at its Netherlands data center to free power for AI workloads. Duraflow’s solution absorbs, stores and releases heat, cutting dependence on energy‑heavy mechanical refrigeration.
The setup delivers 240 kW cooling with only 5–8 kW of power, versus an estimated 60–80 kW for conventional cooling. It can reduce thermal management power consumption by up to 90% depending on workload and requires no cooling water, enabling more compute within existing power limits.
Supermicro published a solution brief detailing a combined IBM Storage Ceph and GrandTwin server platform. The configuration uses NVMe storage, 100GbE networking and Intel Xeon 6740E processors to deliver a scalable software‑defined storage stack.
At IBM TechXchange 2026, IBM introduced new ransomware and data theft detection capabilities for its FlashSystem portfolio. FlashSystem already triggers encryption alerts within one minute; now FlashSystem 5600, 7600 and 9600 can alert administrators at the start of data exfiltration, before attackers encrypt data for ransom.
The system runs validation checks to confirm suspected data theft in under 30 minutes, targeting a false‑positive rate below 1%. Attackers typically stay inside networks for 11–14 days before deploying ransomware, so this detection gives security teams a critical early warning window.
Earlier this year, IBM updated FlashSystem with AI agent automation for storage operations. The latest FlashSystem.ai enhancements add actionable guidance, cutting routine provisioning effort by up to 90%. Administrators can query the AI agent for contextual recommendations, while retaining final approval authority governed by policy.
Danish startup KineDB secured an €850K pre‑seed round led by Superhero Capital to build a general‑purpose database for engineering teams. It runs across cloud infrastructure and local offline devices, targeting industries that frequently lack stable internet connectivity, such as mining field crews logging equipment status on underground tablets. Data updates happen locally and sync once connectivity returns.
The firm says AI has lowered software development barriers, and Android and iOS client builds are actively in development and due soon.
Kioxia outlined that generative AI‑driven data growth demands higher‑capacity, higher‑density 3D flash. Stacking additional word lines is one path forward, yet beyond roughly 1,000 stacked layers, cell current degradation, wafer warpage and larger block sizes become prohibitive.
Its new MSA‑CBA architecture stacks two or more memory cell array wafers onto a CMOS wafer via copper direct wafer bonding. Splitting layers across multiple array wafers limits per‑wafer stack height, easing degradation and warpage, while shared word‑line switch circuits shrink block size and cut power. Kioxia has demonstrated the world’s first QLC operation on MSA‑CBA prototypes, with clean threshold voltage separation and adequate margins for both individual arrays and the combined structure.
MSP‑focused cyber resilience vendor N-able expanded its strategic partnership with SentinelOne. The joint security scope extends beyond EDR to cover AI, cloud, identity and managed response, integrating with N-able’s UEM, Adlumin SOC and Cove Data Protection. N-able can now also offer SentinelOne FedRAMP‑certified offerings.
N-able also launched AI‑accelerated vulnerability and patch management, designed to compress patch cycles from days to minutes. The closed‑loop workflow scans third‑party applications, prioritizes critical flaws, remediates directly and verifies successful deployment within its UEM platform.
Nasuni has selected AHEAD as its first delivery partner for Nasuni Professional Services. AHEAD, a global provider of AI‑enabled cloud, data, infrastructure and security platforms with a new Reading headquarters, will deliver architecture design, data migration and cutover services, expanding customer access to hands‑on deployment expertise.
Pinecone released pg_pinecone, an Apache 2.0 open‑source Postgres extension built on its vector search engine. It indexes existing vector columns inside Postgres, using compact vector summaries for fast search with a small memory footprint and quick index builds. Filters are applied during search rather than afterward, and all data stays within Postgres, eliminating the need for a separate vector database.
Vector search provider Qdrant 1.20 simplifies hybrid keyword and semantic search. It runs BM25 ranking directly on full‑text indexes and computes statistics automatically, letting developers combine lexical and dense retrieval in a single query without maintaining separate sparse pipelines. Semantic search can overlook exact matches such as part numbers or contract clauses; combining BM25 ranking lifts these precise hits to the top as catalogs and document sets evolve.
At GMIF2026, Samsung’s Kevin Yoon presented the BM1773 SSD, 10th‑generation BV‑NAND and zNAND‑O. The BM1773 is a flagship capacity drive built on V9 QLC, spanning 16TB to 256TB and marketed as the world’s smallest 256TB SSD.
Gen 10 V‑NAND exceeds 400 layers, delivering 58% greater bit density, 33% faster I/O and 25% lower power than V9. It uses vertical bonding of memory and peripheral circuit wafers. zNAND‑O is a next‑gen high‑bandwidth flash for on‑device AI, under development in 4‑layer and 8‑layer variants. Co‑packaged with an MPU over UCIe, it delivers 10× DRAM‑equivalent bit density and 7× better read bandwidth and energy efficiency than conventional NAND. Simulations using Moonshot AI’s Kimi K2‑1T model showed zNAND‑O cut memory cost by 75% at equal performance and supported nearly twice the model size.
Seagate stated its Mozaic HAMR HDDs are the first and only SMR drives added to Broadcom’s HBA approved vendor list, enabling large‑scale data center deployment. This claim is nuanced, as Broadcom’s compatibility documentation also lists WD SMR drives. The publication has asked Seagate to clarify the discrepancy. Seagate added that Microsoft Azure has completed qualification of this Broadcom HBA + Seagate HAMR/SMR stack, providing hyperscale validation.
Starburst announced limited general availability of Torin, its GPU‑native Trino analytics engine built on NVIDIA cuDF for Starburst Enterprise. Torin accelerates SQL with native GPU execution, complementing the existing CPU engine so users can switch compute resources without rewriting Trino SQL, applications or agent workflows. Benchmarks on Blackwell GPUs showed up to 7× faster TPC‑H queries and a peak 11.4× speedup; on L4 instances, ClickBench throughput reached 3.4× the CPU baseline at comparable hourly cost.
Starburst is integrating NVIDIA’s open Nemotron Retriever and Kumo models into its AIDA platform as part of the Agent Knowledge Foundation toolkit. These components help AI agents retrieve context, generate grounded answers, run predictive analysis and optimize decisions for fraud detection, credit risk and KYC workflows.
Enterprise AI infrastructure provider Smartbird announced a partnership with WEKA. The pairing combines WEKA NeuralMesh with Smartbird’s managed AI clusters to deliver a customer‑controlled infrastructure stack for sensitive production AI. Smartbird builds and operates dedicated AI clusters as a managed service, deployed on‑premises or in colocation, tailored to individual workload requirements.
Sovereign AI cloud vendor Zadara earned NVIDIA AI Cloud Ready validation for its multi‑tenant sovereign AI cloud platform. The audit covered the full stack including compute, networking, storage, data platform and GPU orchestration, testing HGX H100 and GB200 NVL72 alongside Quantum InfiniBand and Spectrum‑X Ethernet, confirming readiness for production AI workloads.
Beijing Qianxing Jietong Technology Co., Ltd.
Sandy Yang/Global Strategy Director
WhatsApp / WeChat: +86 13426366826
Email: yangyd@qianxingdata.com
Website: www.qianxingdata.com/www.storagesserver.com
Business Focus:
ICT Product Distribution/System Integration & Services/Infrastructure Solutions
With 20+ years of IT distribution experience, we partner with leading global brands to deliver reliable products and professional services.
“Using Technology to Build an Intelligent World”Your Trusted ICT Product Service Provider!
コンタクトパーソン: Ms. Sandy Yang
電話番号: 13426366826
Japanese
