Dell AI Data Platform Expands with Knowledge Graph, Semantic Layer for Enterprise AI Agents

Dell AI Data Platform Tackles Enterprise AI's Core Challenge

Dell Technologies has expanded its Dell AI Data Platform with capabilities designed to solve what Arthur Lewis, president of Dell's Infrastructure Solutions Group, identifies as the primary constraint in Enterprise AI: data, not models1

1

. The company introduced three extensions to the Dell Data Orchestration Engine—the Unified Semantic Layer, Enterprise Knowledge Graph, and Knowledge Agents—alongside NVIDIA-accelerated data preparation that processes workloads up to 20 times faster2

2

.

A recent Dell survey reveals that 53 percent of business and IT decision makers rank establishing and scaling AI capabilities as a top priority, yet half lack a clear roadmap spanning AI, data governance, and security3

3

. This disconnect creates what Dell calls the pilot-to-production gap, where AI use cases succeed in demos but fail at scale due to fragmented data, governance complexities, and data movement inefficiencies.

Unified Semantic Layer Brings Consistent Business Meaning

The Unified Semantic Layer addresses a fundamental problem: enterprise data was never organized for AI Agents to read. When one system labels information as "client" and another as "account," AI systems waste computational resources reconciling these differences on every query4

4

. Dell's solution provides enterprise-wide business definitions and rules, ensuring terms mean the same thing across all applications.

Vrashank Jain, lead product manager for the AI Data Platform at Dell Technologies, explained that unlike competing semantic tools that reach only structured data sources, Dell's approach handles both structured and unstructured data1

1

. The system uses open models to generate entities and definitions at scale, with human-in-the-loop verification. Organizations can import existing ontologies and classification taxonomies, avoiding vendor lock-in while maintaining their unique semantic frameworks.

Enterprise Knowledge Graph Maps Data Relationships

The Enterprise Knowledge Graph tracks how company data connects across systems, using metadata, lineage, and query history to continuously refine its understanding4

4

. When AI Agents query the platform, it pulls every related table, data product, and vector index the agent is authorized to access, regardless of where data lives. For manufacturers, this means tracing a single anomalous sensor reading through machine repair history, supplier batches, and at-risk orders without manual correlation2

2

.

This capability directly addresses the agentic era of AI, where models must reason across enterprise contexts rather than simply answer isolated questions. The Knowledge Graph eliminates the need for AI Agents to reconstruct relationships on every request, reducing token consumption and computational costs while improving response accuracy.

Knowledge Agents Deliver Trusted Context for Enterprise AI

Knowledge Agents operate as specialized advisors, each grounded in a defined slice of the Enterprise Knowledge Graph4

4

. Organizations control what data each agent accesses, what guidance it follows, quality thresholds it must meet, and spending limits. NVIDIA Nemotron Retriever models provide reasoning and visual understanding capabilities, while NVIDIA cuVS accelerates vector indexing and search.

Crucially, the Unified Semantic Layer, Enterprise Knowledge Graph, and Knowledge Agents run inside customer data centers, keeping sensitive information within the organization's security perimeter2

2

. This architecture supports disconnected deployments and data sovereignty requirements while maintaining model choice flexibility—customers aren't locked into any single AI provider.

NVIDIA-Accelerated Data Preparation Speeds Processing

The Data Processing Engine now runs on graphics processing units using NVIDIA's cuDF library, with Apache Arrow transferring data between Dell storage and processing engines3

3

. Dell's September testing on PowerEdge R770 servers with NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs showed Apache Spark jobs ran 3.9 times faster on average than CPU-only processing, with batch data mining jobs achieving 20.4 times speedup using default settings2

2

.

This accelerated data processing tackles a critical bottleneck: data preparation consumes significant time in AI workflows, and inefficient processing creates delays that compound across the pipeline. By bringing GPU acceleration directly into data engines rather than requiring separate proprietary layers, Dell reduces the steps between raw enterprise information and AI-ready context1

1

.

PowerScale Expands Multitenancy and Security

Dell PowerScale now supports up to 500 tenants in a single cluster, with more granular role-based access control for managing each tenant3

3

. The storage platform can encrypt and authenticate file traffic over Network File System protocol with mutual Transport Layer Security, addressing security requirements for AI service providers and enterprises running shared platforms.

PowerScale serves as the foundation for Enterprise AI at scale, supporting up to 16,000 GPUs, while ObjectScale delivers up to 40 gigabytes per second per node for object storage workloads1

1

. Dell's Lightning File System handles extreme throughput requirements for large-scale training, checkpointing, and demanding inference paths.

Storage Performance Tool and Professional Services

Dell introduced an open-source Storage Performance Tool that benchmarks S3-compatible object storage across training, inference, and checkpointing workloads2

2

. The tool measures throughput and latency across writes, high-concurrency reads, mixed environments, and Iceberg queries, helping organizations size AI infrastructure based on realistic workloads rather than vendor benchmarks alone.

Dell is expanding implementation services to move customers from deployment into production, addressing the operational complexity that slows AI scaling3

3

. The Storage Performance Tool and new services are available immediately, while PowerScale's multitenancy and security updates arrive in November. The accelerated Data Processing Engine follows in December, with additional Apache Arrow acceleration expected in the first half of 2027. The Unified Semantic Layer, Enterprise Knowledge Graph, and Knowledge Agents are scheduled for the first half of 2027.

Why This Matters for Enterprise AI Adoption

Dell's platform expansion addresses the core challenge preventing Enterprise AI from reaching production scale: making fragmented, ungoverned data usable for AI systems without creating new silos or dependencies. By integrating semantic understanding, relationship mapping, and GPU utilization optimization into a unified platform, Dell aims to reduce the operational complexity and cost barriers that trap AI projects in pilot purgatory. Organizations running AI workloads on Microsoft Azure will benefit from managed PowerScale integration, extending the platform's reach into hybrid cloud environments3

3

. As theCUBE Research's Dave Vellante noted, enterprises need harmonized information delivered efficiently at low cost with minimal operational overhead—exactly what Dell is targeting with these releases1

1

.