Dell Pushes Agentic AI to the Desktop to Cut Enterprise Cloud Bills
Dell Technologies announced a suite of AI infrastructure products and services today aimed at moving enterprises beyond experimentation into production-scale deployments, with particular emphasis on running agentic AI systems on-premises rather than in the cloud.
The centerpiece is Dell Deskside Agentic AI, which pairs Dell workstations with Nvidia's NemoClaw software to let organizations develop and run AI agents locally. The initiative addresses a specific cost problem: agentic systems that autonomously execute multi-step workflows can consume inference tokens continuously, creating unexpectedly large cloud bills.
Dell cited an example where a single developer burned through 1 billion tokens in 24 hours, resulting in a $3,400 cloud bill. Running the same workload on deskside systems could reduce two-year costs by as much as 87% compared with public cloud infrastructure.
Sam Grocott, senior vice president of product marketing at Dell, said the real barrier for enterprises isn't ambition. "Most enterprises don't have an AI ambition problem," Grocott said. "They have an AI execution problem."
Hardware for Different Model Sizes
The deskside offering includes multiple hardware configurations. Compact Dell Pro Max systems handle smaller models, while high-end workstation towers support models with up to 1 trillion parameters. The systems keep sensitive data on-premises and allow organizations to run open-weight models without relying on cloud vendors.
Dell is integrating Nvidia OpenShell across its AI Factory portfolio to provide what executives called "a secure sandbox for running, building, testing and fine-tuning agents" locally.
Data Fragmentation Remains a Bottleneck
Beyond agentic AI, Dell announced updates to its AI Data Platform designed to address a different execution barrier: fragmented data across organizational silos.
Varun Chhabra, senior vice president of infrastructure and telecom marketing, said many customers remain stuck in pilot phases because they cannot access or govern their data effectively. "Most customers are not short on AI ideas," Chhabra said. "One of the hidden bottlenecks that we find when we talk to customers is the ability to get their data strategy right."
The platform updates include enhanced orchestration and search capabilities that can index billions of unstructured files and accelerate vector indexing up to 12-fold. Dell also announced GPU-accelerated SQL analytics developed with Nvidia and Starburst Data that promise up to six times faster query performance on Nvidia Blackwell processors.
Integrated Infrastructure and New Partnerships
Dell introduced PowerRack, a pre-engineered system that integrates computing, networking, storage, cooling and management for AI and high-performance computing workloads. Chhabra said customers increasingly want integrated systems instead of assembling components from multiple vendors.
The company also announced new cooling systems, including PowerCool CDU C7000, designed to support Nvidia's upcoming Vera Rubin NVL72 platform in a compact 4U form factor.
Dell announced partnerships with Google, OpenAI, Palantir Technologies, ServiceNow and Hugging Face. Google Gemini models will be available through Google Distributed Cloud on Dell infrastructure, while OpenAI's Codex coding agent will integrate with Dell's AI Data Platform.
A new Dell AI Ecosystem program will validate partner applications on Dell infrastructure and help vendors reach enterprise customers through Dell's sales channels.
Most products will ship throughout 2026, with some offerings available immediately. Dell said its AI Factory with Nvidia initiative now has more than 5,000 customers globally.
Learn more: Generative AI and LLM fundamentals, or explore AI for Product Development strategies.
Your membership also unlocks:
