Optical character recognition (OCR) replacement tech provider Adeptia, an intelligent data automation company, announced AIDP, an AI-driven Intelligent Document Processing offering, for use in the cloud, on-prem, and in hybrid environments. Leveraging advanced AI technologies, including ML and NLP, AIDP enables users to extract insights from unstructured documents automatically. It intelligently processes and interprets a variety of documents, including those with complex layouts, handwritten entries, and large, multi-page documents.
For decades, OCR has been the go-to technology for digitizing physical documents, but Adeptia claims this is not enough. AIDP aims to ensure highly accurate data extraction and seamless integration into business workflows, helping to extract insights and enabling users to put their data to work in minutes.
Adeptia tells us: “For decades, Optical Character Recognition (OCR) has been the go-to technology for digitizing physical documents, but OCR alone is not enough. Enterprises need a solution that goes beyond simply converting images to text to understanding the meaning and context of the data within those documents. This is where Adeptia AIDP steps in, enabling the extraction of unstructured data into a structured format. This is immensely helpful for further processing and integration into business applications and proprietary internal systems of record. AIDP also includes support for humans-in-the-loop, allowing users to verify extracted data in a structured format. It also provides support for AI-driven business rules to validate data automatically, a must-have for ensuring data quality, and includes support for handling large files and other complex variations.”
Deepak Singh, Adeptia’s Chief Innovation Officer, tells us: “We completely replace OCR with a much more complete solution. The input to Adeptia is same as the input to OCR which may be scanned images or PDF forms or Word or Excel documents.”
…
William Blair analysts say Cisco noted at its San Diego-based Live conference a renewed acceleration in on-prem infrastructure demand, driven in large part by AI inference. While Cisco expects continued growth in cloud, it notes that the need to run large and small language models in company datacenters or at the edge is driving a resurgence in on-prem spending. This aligns with broader commentary we have received from other vendors and our VAR checks, which have highlighted a resiliency in on-prem demand – a positive sign for Cisco alongside other on-prem vendors like Dell, Nutanix, NetApp, HPE, Pure Storage, and Broadcom/VMware.
…
William Blair analyst Jason Ader took a look at Commvault and told subscribers: “Against the backdrop of exponential data growth and the increasing sophistication of ransomware threats, Commvault remains confident in its ability to sustain midteens revenue growth going forward. This outlook is underpinned by structural market tailwinds and the company’s ability to execute across multiple fronts. In the $9 billion on-premises backup market—still 75 percent of the $12 billion backup and recovery TAM—Commvault is taking share from legacy vendors (mainly IBM and Dell, which continue to cede ground due to their lack of hybrid cloud support. Management noted that 30% of this market is still controlled by such legacy players, creating an ongoing share gain runway.
“Simultaneously, Commvault is benefiting from strong momentum in the $3 billion cloud backup market, which is expected to grow to $4 billion by 2027. With 80 percent of data currently unprotected by third-party backups and 80% of new data expected to be cloud-native, management sees significant white space for expansion. Furthermore, Commvault is aggressively expanding into the $12 billion data and cloud security market via relatively newer offerings such as Threat Scan, Security IQ, Cleanroom Recovery, and Active Directory. These newer products not only expand the company’s total addressable market, but also unlock significant cross-sell potential; management highlighted that 70 percent of its 7,000-plus SaaS customers currently use only one of its roughly 15 SaaS offerings.”
…
Objective Analysis says the market for CXL (Compute Express Link) products has become a hot topic, bringing a broad range of features aimed at enabling computing architectures that were previously challenging to implement. This emerging technology will enjoy rapid growth to nearly $3.4 billion by 2028 as computing systems are configured to take advantage of its many features. This forecast is part of Objective Analysis’ new 60-page report titled “CXL Looks for the Perfect Home.”
…
More on Hitachi Vantara’s EverFlex AI Data Hub as-a-Service, a fully managed, all-in-one infrastructure consumption service addressing AI data preparation challenges by providing a modern data lakehouse with integrated workbench capabilities for AI, business intelligence (BI), and data needs. It helps customers better manage AI workloads by paying for only the infrastructure they need with the security, control and cost savings of hybrid cloud infrastructure. Designed to support robust data quality and governance, customers can leverage the service to enable a single view of enterprise data to feed their AI and BI initiatives with complete and timely data, unlocking faster insights and better business outcomes.
By bringing together Hitachi EverFlex STaaS, VSP One and iQ technologies with Cisco Powered Hybrid Cloud IaaS solutions containing GPU Compute and Networking as a Service, customers benefit from:
- Unified Data Management: VSP 360 offers a unified data management software platform that supports block, file, object and software-defined storage infrastructures. It orchestrates and automates data services, manages data policy compliance, and provides AIOps observability for predictive data infrastructure insights.
- Comprehensive AI Data Infrastructure: Hitachi iQ provides the necessary infrastructure for AI and analytics workloads, including GPU servers, high-speed storage, and networking components to form a comprehensive AI infrastructure with integrated systems for AI and analytics workloads. It emphasizes data governance, security, and quality, and offers flexible deployment options.
- AI Data Integration and Preparation with a Modern as-a-Service Data Lakehouse management: With Zetaris modern lakehouse for AI software, the EverFlex AI Data Hub as-a-Service enables real-time connectivity to diverse data sources and performs federated analytics for data in place, emphasizing data governance and quality at the source, reducing cost inefficiencies and technical barriers to analytics and BI. The service accelerates data pipelines, enabling rapid AI and machine learning deployment, and provides a unified platform for AI, BI, and data operations.
More information here.
…
HPE’s Q2 2025 revenues were up 6 percent to $7.63 billion with a GAAP loss of $11 billion. Its GreenLake (subscription) ARR was up 47 percent to $2.2 billion. Server revenues were up 6 percent to $4.1 billion, intelligent edge revenues up 7 percent at $1.2 billion, and hybrid cloud (inc storage) up 13 percent to $1.5 billion. HPE noted solid storage performance including Alletra revenue up high double-digits year-over-year with Alletra MP storage orders growing more than 75 percent year-over-year for the fourth consecutive quarter.
…
Cloud backup storage supplier Keepit says it has been selected by the French Tennis Federation (FFT) to independently back up its Microsoft 365, Microsoft Entra ID, and Power BI data. The FFT chose a sovereign backup offering, independent of major global cloud providers. This deal began in 2022, alongside SCC France, a partner of the FFT for over 15 years, with the initial aim of safeguarding Microsoft 365 environments. Since then, the partnership has gradually expanded to include Power BI and Microsoft Entra ID. FFT now plans to systematically integrate any new Microsoft solution it adopts into the Keepit ecosystem, ensuring continuity and consistency in the protection of its digital assets.
…
Micron is now shipping HBM4 to key customers. It’s engineered to meet the extreme memory bandwidth and capacity demands of next-generation AI accelerators and high-performance computing (HPC) platforms. It says it’s among the first in the industry to start shipping HBM4 to customers.
…
Micron announced the integration of its 36 GB, 12-high HBM3E into the upcoming AMD Instinct MI350 Series GPUs. This highlights the role of power efficiency and performance in training large AI models. AMD’s Instinct MI350 Series integrates 288 GB of HBM3E memory capacity, delivering up to 8 TBps bandwidth for exceptional performance. This enables Instinct MI350 series GPUs to efficiently support AI models with up to 520 billion parameters on a single GPU. AMD’s tightly integrated architecture, combined with Micron’s power-efficient HBM3E, empowers datacenters to scale seamlessly while resourcefully maximizing compute performance per watt.
…
Nasuni has appointed three new executives as CPO, CIO, and CISO:
- Alison Bayiates joins as Chief People Officer, bringing extensive experience in scaling people strategies for high-growth technology companies. She previously held leadership roles at RSA and Veracode.
- Dalan Winbush is Nasuni’s new Chief Information Officer, bringing more than 25 years of experience leading enterprise IT, cybersecurity, data, and AI initiatives, working for companies including Quickbase and Comcast.
- Elyse Gunn joins as Chief Information Security Officer, bringing deep expertise in security and compliance leadership for SaaS businesses. She has previously worked for Pax8 and Quickbase.
These additions come not long after the appointment of new Nasuni CEO Sam King in April 2025, and reflect Nasuni’s strong growth in the hybrid cloud storage market.
…
Fabless Numem has a patented AI Memory Engine with foundry-ready MRAM architecture supporting die densities up to 1 GB and delivering SRAM-class performance with up to 2.5x higher memory density in embedded applications and 100x lower standby power consumption. It claims it has transformed MRAM into a production-ready, scalable memory building block for next-generation, AI-focused workloads. It has high endurance, enabling MRAM to support SRAM and DRAM-like architectures, and is scalable, software-defined memory. The company claims it delivers 30-50 percent power savings over existing high-bandwidth memory solutions.
…
Objective Analysis‘ Jim Handy writes about the iunvention of DRM: “Fifty seven years ago today, the US Patent Office granted a patent for a new memory technology to IBM. This approach used a single transistor and a single capacitor as an alternative to SRAM which is built with a six-transistor cell. This Memory Guy blog post was written to commemorate that milestone.”
…
CXL is getting closer. Panmnesia exhibited a full-stack HPC system—designed to accelerate parallel computing applications—based on a CXL server featuring CXL 3.x Switches at ISC High Performance 2025.
…
The PCI-SIG announced the official release of the PCI Express (PCIe) 7.0 specification, reaching 128.0 GT/s, and targeting data-driven applications like AI/ML, 800G Ethernet, cloud, and Quantum computing. Pathfinding for the PCIe 8.0 specification is already in progress to continue supporting the industry’s investments and product roadmaps in the PCIe technology ecosystem. PCI-SIG also announced a new optical interconnect specification revision to enable higher PCIe technology performance. The Optical Aware Retimer Engineering Change Notice (ECN) amends the PCIe 6.4 specification and the new PCIe 7.0 specification to include a PCIe retimer-based solution, providing the first industry standardized way to implement PCIe technology over optical fiber.
…
Phison announced its most advanced lineup at Computex by introducing new offerings purpose-built for write-intensive enterprise workloads, on-premises AI inferencing, and next-gen client devices.
- Pascari X200Z Enterprise SSD: Engineered for ultra-high endurance (up to 60 DWPD) and SCM-like responsiveness, the new X200Z redefines enterprise-grade performance for demanding AI, analytics, and database workloads. It’s Phison’s most advanced Gen 5 SSD for write-intensive environments to date.
- aiDAPTIVGPT: Announced for the first time at Computex, aiDAPTIVGPT is a plug-and-play inference services suite for on-premises LLMs. It offers conversational AI, code generation, voice services, and more, all in a unified toolkit. Phison says it fills a critical market gap for SMBs, universities, and government agencies seeking secure, localized AI performance without public cloud dependencies.
- E28 SSD Controller: Making its Computex debut as a 2025 Best Choice Award winner, the E28 is Phison’s flagship PCIe Gen5 SSD controller, now with integrated AI compute acceleration for faster model updates and unmatched Gen 5 performance.
…
Qumulo has added two board directors: Kelly Battles, a seasoned technology CFO and board member, and Bob Gleichauf, a technology leader with deep experience in enterprise IT infrastructure and cybersecurity. The company says it is delivering record revenue growth and has more than 1,100 customers in industry verticals including Media and Entertainment, Energy Exploration and Delivery, Gaming and Sports, Automotive and Autonomous Driving, Semiconductor Development, Healthcare and Life Sciences, and Financial Services. Since it launched Cloud Native Qumulo on Azure and AWS in October 2024, Qumulo has seen more than 400 percent year-over-year growth in cloud data storage.
…
Data protector and security supplier Rubrik announced 49 percent year-over-year revenue growth to $278.5 million for its first fiscal 2026 quarter with a GAAP loss of $102.1 million, subscription revenue up 54 percent to $265.7 million, and subscription ARR up 38 percent to $1.18 billion.
Operating and free cash flow were $39.7 million and $33.3 million respectively and it has $762.1 million in cash, cash equivalents & short-term investments.
There are now 2,381 customers with more than $100,000 in subscription ARR, up from 2,246 last quarter. With such a cash hoard, we wonder if an acquisition of some note could be on the table. Rubrik has just re-overtaken Commvault with its revenue total.
…
DCIG’s Jerome Wendt has an interesting take on edge HCI supplier Scale Computing after attending its Platform 2025 conference in Las Vegas. He mentions a Scale Computing installation above a commercial kitchen sink – the dirty dish datacenter – which uses SimplyNUC hardware.
…
Great Anton Shilov interview with Silicon Motion (SMI) founder and CEO in Wallace C. Kou in Tom’s Hardware about SMI’s SSD controller strategy, competitors (Phison), and customers.
…
Data warehouser/analayzer Snowflake announced a global strategic initiative with Acxiom (IPG) to build a next-gen, AI-powered marketing data foundation which is fully secure, transparent, and cloud-native. It says this partnership marks a significant shift away from “black box” agency models, empowering brands with direct access to data, real-time personalisation, and advanced AI capabilities – all within their own Snowflake environments. Acxiom’s identity tools (Real ID, clean rooms, predictive audiences) are embedded directly in Snowflake. It provides a first-of-its-kind AI marketing data stack with global scale, privacy-by-design, and zero-copy data sharing, and enables brands to activate first-party data, optimise campaigns in real-time, and drive ROI with full transparency.
…
The Ultra Ethernet Consortium (UEC) announced the release of UEC Specification 1.0, a comprehensive, Ethernet-based communication stack. It delivers RDMA for Ethernet and IP, open standards and interoperability to avoid vendor lock-in and end-to-end scalability to millions of endpoints.
…
TechCrunch reports VAST Data is looking to raise more cash at a valuation of around $25 billion, up from its $9 billion Series E valuation in December 2023. VAST did not comment.
…
SW RAID supplier Xinnor says its xiRAID technology has just achieved a #3 position globally and #1 among Lustre-based systems in the IO500 production rankings announced at ISC High Performance 2025, “demonstrating world-class storage performance with just a ½ rack deployment.”
