Storage News Ticker - February 23

Datacenter-in-space business Starcloud says it will launch AWS Outposts hardware into orbit in October. Outposts is AWS’ on-prem version of its public cloud.

…

Databricks is partnering DBeaver Corp. and launching of DBeaver's native Databricks integration, providing advanced SQL development and data exploration capabilities. With DBeaver and Databricks, joint customers can work more efficiently across their entire Databricks workflow while maintaining access to more than 100 additional databases through a single, unified interface. DBeaver's integration provides complete visibility into all data processes, from raw ingestion tables to production-ready datasets, enabling users to inspect data, metadata, and database structure at every stage. With native Databricks authentication support and seamless cloud platform integration, teams can leverage DBeaver's advanced capabilities without disrupting their existing operations.

Databricks connectivity is available in DBeaver Community and all DBeaver PRO editions, with advanced features such as enterprise authentication methods and visualization tools offered exclusively in DBeaver PRO. To learn more about the partnership with Databricks, please visit dbeaver.com/databricks.

…

DDN has appointed Bhaskar G Manickam as its new Director of Product Management to “help customers advance their AI initiative from pilot to production.”He is a 12-year veteran of HPE, where he had been Director of Product Management for nearly 3 years.

…

Formula 1 racing team McLaren Racing has extended its relationship with Dell Technologies as its Official Innovation Partner. Dell says its infrastructure, PCs and services continue to support the McLaren Mastercard Formula 1 Team from the factory to the racetrack. McLaren can process 1.5 terabytes of data per race weekend, Using the Dell AI Factory, including PowerEdge servers, PowerStore and PowerScale arrays, and AI systems. Read more in a blog. 

…

Tiered deduping backup target system supplier ExaGrid has reached the 5,000+ customer level. It’s growing at the 150 to 200 new customers per quarter rate. ExaGrid says it has the easiest POC (Proof of Concept) test program in the industry, allowing organizations to test ExaGrid in their production environments. Bill Andrews, President and CEO of ExaGrid, said: “When ExaGrid is in an opportunity, the customer chooses ExaGrid over 70% of the time versus primary storage or inline deduplication options from Dell, HPE, NetApp, Pure Storage and others.”

...

HPE's Alex Veprinsky, Chief Architect, Storage, has written a blog about how HPE's Alletra Storage MP X10000 using RDMA/S3 is great at participating in KV Cache workloads. A table supplies comparative data:

Read the blog here.

…

We’re told IBM Storage Ceph 9.0's near-zero impact bucket resharding represents a fundamental shift in production object storage operations, eliminating one of the most significant pain points in large-scale deployments. As Ceph continues evolving with features like in-order sharding, the vision becomes clear: single-bucket architectures that scale infinitely without operational complexity. For data lakehouse architects building on Apache Iceberg, for AI/ML engineers managing billions of training artifacts, and for enterprise architects demanding the highest availability without operational friction, IBM Storage Ceph 9.0 delivers the operational maturity that production workloads require. Read this document to find out more.

…

A study reported in Harvard Business Review on how AI is changing work habits found that AI tools didn’t reduce work, they consistently intensified it. Greg Hanson, Group VP and Head of EMEA North, Informatica from Salesforce, says: “AI intensifies workloads when companies fall foul to the AI Verification Tax. If AI can’t be trusted to work unsupervised, the productivity promise collapses and instead adds time to the task. This leaves employees spending more time checking and correcting AI outputs as they would if they were doing the task themselves. 

“This verification burden is compounded by a skills gap. 75 percent of data leaders tell us their workforce lacks data literacy, and 74 percent say more AI literacy training is needed to use AI responsibly. But it isn’t inevitable. Where data is well governed and employees have the skills to challenge AI outputs, verification drops, decisions scale more safely, and productivity gains become real rather than theoretical.” 

…

A Keepit blog by Chief Vision Officer (CVO?) Frederik Schouboe asks “what fully agentic AI could enable in SaaS data protection, and how to get the speed without surrendering the non-negotiables of independence, immutability, and testable rollback.” Agentic AI is the next interface layer for work. Agentic AI doesn’t just make progress faster, it also makes mistakes faster, too. If an agent can act, it can also act wrong — confidently and at scale. Rollback is the locked door, and without it, Lecter walks out. Read the blog for much more detail.

...

Cloud file services supplier Nasuni has an expanded partnership with intelligent content lifecycle management supplier DryvIQ through the launch of a new connector. The new connector enables organisations to migrate content into Nasuni, while providing a complete picture of their Nasuni content plus other unstructured data systems including collaboration platforms, foundational storage, and line-of-business applications. We’re told DryvIQ translates content insights into clear, defensible actions, enabling customers to manage Nasuni file data alongside other enterprise content through a unified view of their data environment. The Nasuni connector is available immediately within the DryvIQ platform.

….

Australia’s New Cloud Dynamics’ Neo One device “is an entirely new category of secure private infrastructure, one that combines hardware security module (HSM) technology, distributed architecture, secure communications and user-centric convenience into a single, self-contained platform.” It “is a secure, self-contained personal hardware node that stores data in a distributed, encrypted form across a private network of trusted Neo One devices. It is a personal, private cloud comprised only from these devices and does not use any central servers or datacenters.” 

“Unlike conventional encrypted storage, Neo One is not locked away from day-to-day use. It runs applications. It manages data dynamically. It behaves much more like a secure personal or organisational database, rather than a static vault. It also enables defence-grade communications security by providing a proprietary messaging system that doesn’t route through any cellular or satellite network. This means the messages will only ever be seen on the private network, they can’t be intercepted by ISPs or foreign jurisdictions.”

Users interact with Neo One much like they would a smartphone, browsing files, managing data, and using applications, but underpinned by defence-grade security architecture. It’s available in1, 2, 4, 8 and 12 TB SSD configs. Find out more here.

… 

Phison CEO K.S. Pua says DRAM and NAND shortages will extend past 2030 and thinks some SSD-based consumer businesses could cash as a result. More here.

…

Vector database supplier Pinecone has a Bring Your Own Cloud (BYOC) offering, available now in public preview for all Enterprise users. BYOC is Pinecone’s split-plane architecture applied to a zero-access enterprise security posture. It brings Pinecone into your AWS or GCP account with a zero-access operating model. With BYOC, the Pinecone Database’s data plane runs inside your VPC. Your vectors are stored and queried there. Pinecone doesn’t need SSH, VPN, or inbound network access to your cloud account, VPC, Kubernetes cluster, nodes, or workloads. You don’t open inbound firewall ports for Pinecone to operate the system. You still use the Pinecone control plane (API and console) for index lifecycle, auth, and usage visibility. 

Provisioning creates the data plane environment in your cloud account, including a dedicated VPC and an EKS/GKE cluster, plus supporting cloud services for storage, system state, and TLS/DNS. See the BYOC GitHub repository and documentation for more information.

...

Vector database supplier Qdrant says engineering teams running vector search at scale face two persistent problems: retrieval that misses user intent, and latency that degrades under write load. Available today, Qdrant 1.17 addresses both directly. The headline feature is Relevance Feedback Query, a vector-native approach to iterative retrieval refinement. Rather than rewriting queries or rescoring large document batches, it collects lightweight feedback on a small set of top results, builds "context pairs" of more- and less-relevant examples, and adjusts the similarity scoring function on the next retrieval pass. No model retraining, no expensive reranker. On the infrastructure side, a new update queue buffers up to one million pending changes with back pressure under heavy ingestion, and delayed fan-outs hedge against slow replicas by querying a second in parallel. Full details in a blog post.

...

Bloomberg reports Samsung shares surged 5.4 percent to a record high after a report said the company plans to price HBM4 chips at ~$700 per unit, 30 percent higher than the previous generation.

…

Samsung Electronics is going to increase its new DRAM production capacity by nearly 20 percent this year, centered around its Pyeongtaek Plant 4 (P4). This will boost the production of 6th-generation 10nm-class (1c) DRAM, the primary material for 6th-generation High Bandwidth Memory (HBM4). Read more here.

…

Samsung and SK Hynix plan to increase production of their latest generation NAND chips, 280-layer VP for Samsung, and 321-layer 9th gen NAND for SK Hynix. They’ll convert existing prior generation NAND fab capacity to the new higher layer count generations. Read more here.

…

Sandisk announced the pricing of a secondary public offering of 5,821,135 shares of its common stock currently owned by Western Digital Corporation (WDC), the Company’s former parent. In a debt-for-equities exchange, WDC is expected to deliver its Sandisk shares to J.P. Morgan Securities LLC and BofA Securities who will sell them. WDC is then expected to own 1,691,884 shares of the Company’s common stock, which it intends to dispose of in one or more subsequent exchanges for outstanding shares of WDC common stock and/or through pro rata distributions to WDC stockholders. WD hopes to raise $3.1 billion from deal.

…

Object storage supplier Scality has boosted its F5 partnership “o to help organisations securely scale AI, analytics, and data-intensive workloads. By integrating the F5 Application Delivery and Security Platform (ADSP) with Scality’s highly scalable, S3-compatible object storage, the companies are enabling enterprises to deliver faster, more resilient, and more secure access to data across on-premises, cloud-native, and hybrid environments.” The two have integrated “F5 BIG-IP with Scality RING object storage to create a resilient, secure, and high-performance S3 environment.” 

…

SpaceX boss El Musk says AI expansion will cause Earth-bound data centers to run out of energy, with orbiting space data centers using solar energy 24 hours a day the answer. Talking to Dwarkish POatel, Musk said: “In 36 months, but probably closer to 30 months, the most economically compelling place to put AI will be space. The limiting factor once you can get to space is chips, but the limiting factor before you can get to space is power.”

…

Speedata, developer of the world’s first purpose built Analytics Processing Unit (APU), announced a strategic partnership with Nebul, a European Private AI Neocloud provider. Nebul will be the first to deploy Speedata’s APU technology within its sovereign cloud infrastructure, enabling European enterprises to run advanced analytics and AI data processing workloads at scale, while maintaining full jurisdictional control over their data. 

By integrating Speedata's analytics accelerator with Nebul's European-owned infrastructure, enterprises can achieve significant performance gains and lower total cost of ownership on Apache Spark workloads while their data remains under European legal authority. Learn more at www.speedata.io and www.nebul.com.

...

Trino open source, distributed SQL supplier Starburst has just passed $100 million ARR, up nearly 40 percent YoY, with 82 percent growth in financial services. It says it now works with four of the top five banks in the Americas and seven of the top ten in EMEA, claiming this is largely because those firms want to push ahead with AI without moving sensitive data into a centralized cloud stack.

Justin Borgman, Founder and CEO of Starburst, said: “For the past decade, the industry pushed a single idea: move all your data into one platform, centralize first, analyze later. That model was built for BI. AI changes the rules. Companies can’t unlock AI productivity gains with a data strategy designed for dashboards. They need a federated context layer that connects distributed data, governance, and business meaning into a single AI-ready foundation. That’s what Starburst delivers. And with AIDA, we’re bringing that foundation directly to the business user, transforming governed, federated data into an intelligent, conversational interface for the enterprise." Read more in a Borgman blog.

…

Stravito, which provides an Insights Intelligence Platform, has upgraded AI Assistant with a Deep Research Agent. AI Assistant autonomously plans multi-step research, analyzes full reports in parallel, and iteratively quality checks its own work before presenting results. It works exclusively from trusted inputs, including primary and third-party research. Thor Olof Philogène, Founder and CEO of Stravito, said: “AI that can't be verified risks becoming a source of wrong evidence. Deep Research Agent is built to close that gap: delivering answers with the depth, sourcing, and credibility that major commercial decisions require.”

The Deep Research Agent upgrade to Stravito AI Assistant will roll out to all customers during Q1 2026. For more information, click here. 

...

SurrealDB 3.0 is the first multi-model, AI-native database that combines relational, document, graph, time-series, vector, search, geospatial and key-value data types with real-time capabilities - all in one database. It solves AI agents’ memory problem - giving AI systems a more reliable memory layer directly inside the database. AI Agents need a unified view of state (semantic context, structured facts, and durable memory) but struggle to remember facts consistently, understand relationships, and to keep context and memory - particularly when data size and complexity grows. SurrealDB 3.0 is the only database to enable first-class agent memory and context graphs inside the database, so models run next to the data, context stays synchronised, and agent logic remains simple.

The release of 3.0 comes as SurrealDB raises an additional $23M Series A funding, bringing its Series A total to $38M, and its total funding to-date to $44M. This makes SurrealDB one of the most well-funded database startups of all time. Read more in a blog and here.

…

TrueNAS - iXsystems as was - announced that Brett Davis has been appointed CEO, replacing Mike Lauth, co-founder, President and now ex-CEO. Davis, a 20-year iXSystems vet, was EVP Sales & Marketing, and he will “llead TrueNAS … by delivering what legacy platforms cannot: predictable storage economics, lifecycle control without lock-in, and adaptable infrastructure that supports both current production workloads and emerging AI use cases on a single unified platform. 

Lauth said: ”Brett has been instrumental in building the TrueNAS business into what it is today. He understands the enterprise market, our partners, and the operational discipline required to scale a platform responsibly. This transition reflects both confidence in our strategy and the market opportunity ahead. Brett is the right leader to help our customers and partners navigate this inflection point in enterprise storage." 

…

 Following its separation from Sandisk, WD has rebranded its Sandisk Professional products to G-DRIVE, “as the unified identity for its content creator and creative professional external storage product portfolio. The brand consolidation brings together high-capacity, high-performance storage solutions under a single name brand – G-DRIVE – known for its quality, performance, and reliability. Products currently branded as SanDisk Professional will transition to the G-DRIVE® brand by the end of this month.

Existing SanDisk Professional HDD-based products will continue to receive full support and warranty coverage by WD. Products with new G-DRIVE branding have already begun shipping. 

…

WD is investing $74.1 million in its R&D center in Thailand to develop HAMR technology in the components and laser-based data recording processes areas. Read more here.