Dell pushes on‑prem AI stack
- Dell and Nvidia expanded Dell AI Factory on May 18, adding deskside agentic AI systems, rack-scale infrastructure and new partner integrations for on-prem deployments. - Dell said AI Factory now has more than 5,000 customers after adding 1,000 in the last quarter, while promoting “time to token.” - Dell Technologies World continues through May 22 in Las Vegas, where Dell is detailing AI Factory products and partner rollouts.
Dell Technologies and Nvidia used Dell Technologies World in Las Vegas on May 18 to widen their pitch for enterprise AI that runs on customer-owned infrastructure rather than in public clouds. The updates to Dell AI Factory with Nvidia span deskside workstations, data platforms, liquid-cooled racks, networking and software integrations, according to Dell and Nvidia. Dell framed the package as a full-stack system for companies moving AI from pilots into production. Nvidia said Michael Dell and Jensen Huang presented the platform as covering “deskside to data center” deployments for autonomous agents and other enterprise workloads. (dell.com) ### What did Dell actually announce? Dell said the new releases include a deskside system for building and running secure agentic AI locally, plus OpenShell integration intended to let workloads scale from deskside machines into the data center. The company also announced new data, storage, networking and rack infrastructure, including turnkey rack deployment and liquid-cooling-related upgrades. (dell.com) Dell’s product pages describe AI Factory as a combined stack of compute, storage, networking, software and services built with Nvidia. Dell has used that framing since the platform’s 2024 launch, but the 2026 event adds more emphasis on agentic AI and on-prem control. ### Why is Dell stressing on-prem and “infrastructure they control”? Dell said customers including Eli Lilly, Honeywell and Samsung are choosing AI infrastructure deployments “running workloads on hardware they own and control rather than in the cloud.” Its event materials repeatedly tied that message to security, governance and data location. (dell.com 1) (dell.com 2) Nvidia’s event coverage also described the offering as a platform for enterprises that want secure, local deployment of autonomous agents. Dell said the open ecosystem now includes integrations or solutions with Google, Hugging Face, OpenAI, Palantir, Reflection, ServiceNow and SpaceXAI. (dell.com) ### How big is the business Dell says it has built? Dell said on May 18 that more than 5,000 customers are now deploying Dell AI Factory. A Dell event post said the company added 1,000 new customers in the last quarter alone. That marks an increase from the 4,000-plus customer figure Dell cited on March 16, when it announced earlier AI Factory updates and said the business had reached its two-year anniversary. (dell.com) At that time, Dell said early adopters had seen up to 2.6x return on investment within the first year. ### Why are “time to token” and “cost per token” showing up now? Dell’s event coverage said enterprises are increasingly evaluating AI systems by production metrics rather than by model access alone. Forbes reported that Dell is pushing “time to token” and “cost per token” as measures for AI infrastructure decisions, linking them to latency and unit economics. (dell.com) Those terms fit Dell’s broader argument that AI is becoming an operational asset. In its March 16 release, Dell said customers need data platforms, infrastructure and services that shorten the path from pilot to production and deliver measurable returns. ### What comes next from here? (dell.com) Dell Technologies World runs through May 22 in Las Vegas, and Dell said it will continue posting product and partner details from the event on its AI announcements page. The next public milestones are likely to be availability dates and customer deployments tied to the newly announced deskside, rack and ecosystem offerings. (dell.com) (dell.com)