Dell expands AI Data Platform with knowledge graph

Dell has expanded its AI Data Platform with new data orchestration, processing and storage features for AI agents and applications. The update adds a semantic layer and knowledge graph intended to give systems a shared view of enterprise data.

It also includes faster data processing tools, new security features for PowerScale storage and additional services aimed at helping customers move AI projects into production.

A central part of the expansion is a Unified Semantic Layer, which applies common rules, definitions and a searchable glossary across structured and unstructured data. The feature is intended to ensure business terms are interpreted consistently across different systems, including when separate applications use different labels for the same concept.

The layer also supports imported ontologies and classification taxonomies. NVIDIA Auto-Ontology, an open-source library, can be used with it to build knowledge graphs from enterprise data.

Dell has also added an Enterprise Knowledge Graph to map relationships across datasets using metadata, lineage and query history. It is intended to help applications and agents locate relevant context across tables, data products, multimodal data and vector indexes.

Alongside this, Dell introduced Knowledge Agents tied to defined sections of the Enterprise Knowledge Graph. Customers can set the rules for each agent, including what guidance it follows, what data it may access, what quality threshold it must meet and how much it is allowed to spend.

These elements will remain within the customer's own data centre, reflecting the sensitivity of the information involved and the need for governance over internal datasets.

Processing changes

Dell also outlined changes to the data processing layer. In internal testing, the Dell Data Processing Engine, using NVIDIA cuDF on NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs, processed data nearly four times faster on average than CPU-only Apache Spark runs. For batch processing workloads, the performance increase reached up to 20 times.

Apache Arrow is being used to let jobs query data in place, reducing the need to move or reformat information before analysis. Dell said this should shorten the preparation stage between raw enterprise data and AI use.

These changes reflect a broader focus among infrastructure suppliers on the data pipeline around AI systems rather than on models alone. As companies build internal AI tools, organising, governing and moving data has become a larger part of deployment efforts.

Storage security

PowerScale storage is also receiving security and multitenancy updates. A single cluster will support up to 500 tenants, while mTLS over NFS will add encryption and authentication for file traffic.

The update also adds more granular role-based access controls for tenant management. The changes are aimed at businesses and service providers running shared AI environments for multiple teams or customers on the same infrastructure.

Dell has also made the Dell Storage Performance Tool available. It is intended to help customers test S3-compatible object storage performance across AI training, inference and checkpointing workloads, and compare infrastructure options before deployment.

Services push

Beyond software and infrastructure updates, Dell is extending implementation services around the AI Data Platform. The services cover analytics, processing, search and orchestration, with the aim of helping organisations turn deployed systems into operational ones.

This places the product update within a wider market trend in which technology suppliers increasingly bundle consulting and implementation work with AI infrastructure. Many companies have found that deploying hardware and software is only one step, with integration into existing data environments proving more difficult.

Arthur Lewis, President, Infrastructure Solutions Group, Dell, said: "Data without context is just noise. Most enterprises have spent years making their data accessible. That's not the same as making it usable. An agent that can find a customer record but doesn't know what it means, how it connects to everything else, or whether it can be trusted isn't intelligent. It's just fast. The companies that solve it won't just deploy AI faster or run more agents. They'll get more out of the data they already have."

Dell said the PowerScale security and multitenancy enhancements are due in November, while the initial Data Processing Engine enhancements are expected in December. The Unified Semantic Layer, Enterprise Knowledge Graph and Knowledge Agents are scheduled for release in the first half of 2027, while the Storage Performance Tool and AI-ready data services are already available.