Dell Technologies is expanding its Dell AI Data Platform with new capabilities designed to help enterprise AI applications and agents interpret information across different business systems. Alongside improvements to data processing, storage security and infrastructure management, the updates introduce a shared semantic layer, an enterprise knowledge graph and specialised Knowledge Agents.
Serving as the data foundation of the Dell AI Factory, the platform brings together storage, data orchestration, analytics and search to make enterprise information accessible to AI systems. By giving applications consistent business definitions and access to related information, Dell aims to reduce the need to reconstruct context for individual queries. The new capabilities will become available in phases through the first half of 2027.
Connecting enterprise data through shared definitions
The Unified Semantic Layer is designed to give structured and unstructured information consistent business meaning across an organisation. By applying common definitions, rules and a searchable glossary, it helps AI applications interpret terminology that may differ between systems. For example, where one system refers to a customer as a client while another uses the term account, the semantic layer establishes that these references describe the same information, allowing applications to interpret them consistently.
Organisations can also import existing ontologies and classification taxonomies, including industry-standard and internally developed structures. In addition, Dell is enabling NVIDIA Auto-Ontology, an open-source library that builds knowledge graphs from enterprise data, to extend the semantic layer.
To link distributed records, the Enterprise Knowledge Graph maps relationships between information held across different systems, drawing on metadata, data lineage and query history to refine those connections as business activity changes. When an AI agent receives a question, the platform can retrieve related tables, data products, multimodal information and vector indexes from different locations, subject to the agent’s access permissions.
Dell illustrates this with a manufacturing scenario in which an unusual sensor reading could be traced to the affected machine, its repair history, a supplier batch and orders potentially at risk. By connecting these records, the knowledge graph is intended to help applications examine operational problems using information that would otherwise sit across separate systems.
Knowledge Agents use this connected information to provide guidance on specific subjects. Each operates within a defined portion of the Enterprise Knowledge Graph, with organisations controlling the guidance it follows, which data it can access, the quality standards it must meet and how much it is permitted to spend.
Reasoning and visual understanding for these agents are provided by NVIDIA Nemotron Retriever models. Elsewhere in the platform, the models support document parsing, embeddings and reranking, while NVIDIA cuVS accelerates vector indexing and search. Together, these technologies are intended to help the platform prepare, organise and retrieve enterprise information for AI applications.
Dell says the Unified Semantic Layer, Enterprise Knowledge Graph and Knowledge Agents will remain inside the customer’s data centre, where they can work with governed enterprise information. The platform is also designed to share this context across applications and agents without requiring organisations to commit to a single AI model, data provider or storage vendor.
Accelerated processing and stronger storage controls
Dell is also introducing GPU acceleration into its Data Processing Engine to improve how enterprise information is prepared and analysed for AI workloads. Developed in collaboration with NVIDIA, the changes are intended to reduce processing time before stored information can be used by AI applications.
In internal testing conducted in September 2026, the Data Processing Engine powered by NVIDIA cuDF achieved an average processing speed-up of 3.9 times compared with CPU-only configurations across different workloads, with the highest recorded improvement reaching 20.4 times for a batch data-mining workload. The tests compared GPU-accelerated and CPU-only Apache Spark processing on a Dell PowerEdge R770 equipped with NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs. These results were obtained using default configurations without additional performance tuning, and Dell notes that actual results may vary.
Apache Arrow forms another component of the processing improvements, enabling more efficient movement of data between Dell storage and processing systems by allowing jobs to query information in place. According to Dell, this could reduce the preparation required before enterprise information becomes usable by AI applications.
The storage updates also address organisations running shared AI infrastructure. Dell is introducing expanded multitenancy capabilities for PowerScale, with enhancements designed to support up to 500 tenants within a single cluster. Scheduled for availability in November 2026, these capabilities are intended to allow enterprises and AI service providers to accommodate multiple teams or customers while keeping their information separate.
Planned security improvements include mutual Transport Layer Security (mTLS) over Network File System (NFS), which encrypts and authenticates file traffic. More granular role-based access controls are also designed to give administrators greater control over permissions for individual tenants within the shared storage environment.
Additionally, Dell has introduced the Dell Storage Performance Tool for testing S3-compatible object storage. It measures performance across AI training, inference and checkpointing workloads, helping organisations assess infrastructure capacity and compare storage systems according to their ability to supply data to GPU-based computing environments.
Rollout schedule and implementation services
The platform updates follow a phased rollout schedule. Enhancements to the Dell Data Processing Engine using the NVIDIA acceleration stack will be available in December 2026, with further acceleration using Apache Arrow expected during the first half of 2027. The Unified Semantic Layer, Enterprise Knowledge Graph and Knowledge Agents are scheduled for release in the first half of 2027. Meanwhile, the Dell Storage Performance Tool and the company’s AI-ready data services are already available.
Alongside the technology updates, Dell is expanding its implementation services across the platform’s data and storage engines. These services cover the configuration and activation of analytics, processing, search and orchestration capabilities, together with adjustments to support the requirements of AI workloads. They are intended to help organisations establish a production-ready data environment and maintain the platform as workload requirements change.
