Dell and Nvidia push private AI

Dell Technologies and Nvidia used Dell Technologies World in Las Vegas on May 18 to press a simple argument: more companies want AI systems inside their own walls, not only in public clouds. The companies expanded the Dell AI Factory with Nvidia with new deskside systems, rack-scale infrastructure and networking aimed at enterprises building and running AI locally. Palantir joined that push with Dell-backed messaging around an on-premises AI operating system for customers handling sensitive data. The announcements were aimed at buyers in regulated sectors, government and other organizations that want tighter control over cost, security and data location. ### Why are Dell and Nvidia talking about “private AI” now? Michael Dell and Jensen Huang framed the issue around enterprise deployment, not model novelty. Nvidia said Huang joined Dell on stage to unveil updates to the Dell AI Factory with Nvidia, describing a full-stack platform for autonomous agents “from deskside workstations to data center racks.” Dell said the new systems are meant to give workgroups a local, secure and “cost-predictable” base for production AI. (investors.delltechnologies.com) Dell’s May 18 release said its new Deskside Agentic AI offering lets enterprises deploy and scale agentic AI workflows locally “without the cost, latency and data sovereignty constraints of cloud-only approaches.” The company said organizations can break even against public-cloud API costs in as little as three months for some workloads, while keeping inferencing local. (investors.delltechnologies.com) ### What exactly did Dell put on the table? Dell introduced new infrastructure that stretches from workstations to integrated racks. Nvidia’s event recap said the refresh includes Dell PowerEdge XE9812 built on Nvidia Vera Rubin NVL72, plus other Dell systems based on Nvidia HGX Rubin NVL8, supporting up to 144 GPUs per rack with direct liquid cooling. Dell also rolled out PowerRack, which Nvidia described as a fully integrated system combining compute, networking and storage. (investors.delltechnologies.com) IT Pro reported that Dell said PowerRack can move from delivery to live AI or high-performance computing workloads in as little as six-and-a-half hours. Dell executive Varun Chhabra told reporters the product is designed as scalable units in which thermal design, power management and infrastructure software are engineered together. (blogs.nvidia.com) ### Which customers are these systems aimed at? Dell’s own materials repeatedly pointed to enterprises with security, sovereignty and compliance requirements. The company said its Palantir partnership targets “highly regulated industries and governments,” and listed national security, critical infrastructure, financial services and healthcare among environments where “cloud-first” is not an option. (itpro.com) Nvidia’s event post added a scale marker. The company said 5,000 enterprises, including Lilly, Samsung and Honeywell, are already running AI workloads on Dell AI Factories with Nvidia. Michael Dell said worldwide AI infrastructure spending could reach $3 trillion to $4 trillion by 2030, according to Nvidia’s account of the keynote. ### Where does Palantir fit into this? (dell.com) Dell said on May 18 that it and Palantir were announcing “a full AI operating system” running on Dell AI Factory with Nvidia, bringing Palantir’s Foundry and Ontology platform on premises. Dell described the product as a way to turn fragmented enterprise data into operational AI agents and workflows on infrastructure customers control. (blogs.nvidia.com) Palantir’s own reference architecture page with Nvidia says the system is designed for on-premises, edge and sovereign-cloud environments. Palantir said the stack is tested to run its software suite, including AIP, Foundry, Apollo, Rubix and AIP Hub, on Nvidia infrastructure with Spectrum-X networking and a zero-trust security layer. ### What does this change in the AI spending story? (dell.com) Dell’s announcements widened the focus from chips and public-cloud capacity to the physical systems around them. The new products bundle servers, networking, storage, cooling, power management and deployment services into integrated units, rather than asking customers to assemble those pieces themselves. (palantir.com) The next concrete milestone is customer deployment. Dell said the Palantir on-premises AI operating system is being delivered through Dell AI Factory with Nvidia architectures, while the new Dell AI Factory systems announced at Dell Technologies World are now part of the company’s enterprise AI portfolio. (dell.com) (blogs.nvidia.com)