Dell has expanded its Dell AI Factory with NVIDIA to help businesses move agentic AI projects into production on their own infrastructure. Adoption has risen to more than 6,500 customers, up from 5,000 in the previous quarter.
The update covers validated software, deskside systems, managed services, security architecture, and new server and modular data centre designs for organisations building and running AI agents in-house.
Dell is targeting a market it expects to grow quickly as companies look beyond pilot projects. It cited forecasts that agentic AI will become the largest enterprise workload by 2028, while token consumption is expected to rise sharply and inference costs to fall over the rest of the decade.
At the same time, businesses are still wrestling with the economics of AI deployments. Dell said unclear return on investment remains the main obstacle for 59% of enterprise AI roll-outs and argued that owning infrastructure gives customers more control over spending, data, and operations than public cloud-only models.
Among the additions are newly validated software offerings in the Dell AI Factory with NVIDIA. These include Reflection's Beam open-weight model, Mistral-based tools, Dynatrace for AI observability, Fogsphere for computer vision, and SUSE Linux as an operating system option.
Dell is also adding a layered security approach built with Red Hat AI Factory software and the NVIDIA OpenShell Reference Architecture. It is designed to cover runtime, network, data, and infrastructure layers for organisations running autonomous AI agents at scale.
Managed services form another part of the push. Dell Managed Services for AI is intended to help customers allocate resources, control access, and track AI consumption and costs as projects move from pilot stages into wider deployment.
Deskside expansion
Dell is also extending its deskside AI offering with validated deployment templates for agentic AI workloads. The templates are designed to support installations ranging from a single developer system to more than 1,000 systems across an enterprise.
The blueprints set out the hardware, models, and software stack required for use cases such as agentic coding, agentic VSS, Hermes Agent on NemoClaw, and higher education AI lab environments.
On hardware, a 64GB unified memory version of Dell Pro Precision with GB10 will be available for individual developers and smaller teams. The system can be linked in two- or four-machine clusters to handle larger models and more complex workflows.
Dell Pro Precision with GB300 will also be offered on Windows, bringing up to 748GB of coherent memory and 20 petaflops of FP4 compute to deskside environments already managed by enterprise IT teams.
Infrastructure push
In the data centre, Dell outlined several additions aimed at larger-scale AI workloads. These include a modular data centre reference design with NVIDIA DSX POD that packages compute, storage, networking, rack-level power, and liquid cooling in a containerised system.
Dell also detailed two PowerEdge server lines aimed at CPU-led AI and high-performance computing tasks. The PowerEdge R8822 is designed for standard 19-inch rack environments and supports workloads such as agent orchestration, sandboxing, reinforcement learning, data processing, and analytics, with support for up to four double-wide GPUs.
The PowerEdge M9882 is aimed at CPU-based execution for agentic workloads, including orchestration, tool use, code execution, and evaluation alongside GPU infrastructure for inference. Dell said the design separates CPU compute from GPU clusters so operators can manage power and cooling more efficiently.
For inference, the PowerEdge XE9810 will be Dell's first rack-scale system using Groq 3 language processing units. The liquid-cooled design is intended for production inference workloads and can complement existing GPU clusters.
The expansion also reaches telecoms infrastructure. NVIDIA Aerial RAN Computer on Dell PowerEdge servers will allow communications service providers to run radio access network and AI workloads on a shared compute footprint rather than on separate systems.
"Enterprises used to think about AI infrastructure as an IT budget question, but that's changed. AI infrastructure is no longer a cost center. It's a value creation center. Companies are expanding and reallocating budgets because AI infrastructure drives growth, productivity and competitive advantage. Together with NVIDIA, we are giving enterprises the integrated platforms they need to move agentic AI from experimentation to results, on infrastructure they own and control," said Michael Dell, Chief Executive Officer, Dell.
Industry observers said the latest move reflects a broader shift in customer priorities. Buyers that spent the past two years testing AI tools are now under pressure to show measurable returns and decide whether those systems should sit in the cloud, on-premise, or across both.
"The market is entering a new phase where customers are focused less on building AI infrastructure and more on generating measurable business outcomes. The continued expansion of the Dell AI Factory with NVIDIA reflects a mature approach that combines validated solutions, ecosystem integration, automation and services, helping organizations move from pilots to production with greater confidence and less risk," said Kimball.