Together with NVIDIA, Dell Technologies rolled out a broad update to its Dell AI Factory, shifting its focus to make AI actually usable at scale, not just something that gets stuck in pilot mode.
One of the big parts is the introduction to the Dell Deskside Agentic AI that runs locally on high-performance workstations using NVIDIA’s NemoClaw stack and OpenShell runtime. That means sensitive data stays on-device, latency drops, and companies don’t get hit with unpredictable cloud token costs. Dell says businesses can break even versus cloud APIs in as little as three months, and even cut costs by up to 87% over 2 years, depending on usage.
This new platform also scales, since it uses the same OpenShell runtime that works across the entire stack, from deskside systems like Dell Pro Max and Precision workstations all the way up to PowerEdge XE servers in the data center. So teams can prototype locally and then scale without rebuilding everything from scratch.
Speaking of workstations, a new range of hardware has been announced too, from compact systems for light workloads, enterprise towers that come with NVIDIA RTX PRO Blackwell GPUs, as well as the very best setups powered by GB300 Grace Blackwell Ultra Desktop Superchip.
At the same time, Dell is tackling another bottleneck: data. The company is upgrading its AI Data Platform to handle massive datasets more efficiently, including indexing billions of unstructured files and accelerating SQL queries by up to six times using NVIDIA Blackwell GPUs. There’s also tighter integration with NVIDIA Omniverse, which opens the door for things like digital twins and simulation-driven AI workflows using enterprise data.
The infrastructure layer gets a new addition in the form of PowerRack, a fully integrated rack-scale system that combines compute, storage, and networking into a single optimized platform, and is made to remove the complexity of stitching components together while improving performance, power efficiency, and cooling. Additions like Exascale storage, unified rack management, and a new cooling system capable of handling next-gen AI hardware all point to Dell preparing for much heavier AI workloads.
Meanwhile, partners are all joining in, with the likes of Google, Hugging Face, OpenAI, Palantir, and ServiceNow all part of this, bringing everything from Gemini models running on-prem to curated open-weight models and enterprise-grade AI assistants.
Related to this, there’s also the new Dell AI Ecosystem Program, giving software vendors a structured way to validate their solutions on Dell’s infrastructure. For enterprises, that translates into less risk when moving from proof-of-concept to production, which has been a major pain point.
All of this ties back to a bigger shift – AI workloads are moving toward agentic systems that act autonomously, and that changes the economics completely. Cloud-only setups start to fall apart at scale due to cost and data concerns, so Dell is positioning its stack as a more controlled alternative that spans from individual workstations to full data centers.
Availability
Most of the things are rolling out in phases, with Deskside Agentic AI and OpenShell support available now, while other pieces like data platform upgrades, PowerRack expansions, and cooling systems will continue launching through 2026 and into 2027.
The post Dell Technologies advances agentic AI ranging from deskside to data center appeared first on Nasi Lemak Tech.