- Dell kicked off its annual partner conference with a series of product updates
- On-prem and sovereign AI offerings are a particular area of focus for the company
- Mistral, Samsung Electronics and Eli Lilly are some of the customers using its tech stack
DELL TECHNOLOGIES WORLD, LAS VEGAS — Dell Technologies launched a broad set of advancements for its on-premises stack to help enterprises and their telco partners build sovereign AI infrastructure.
Kicking off Dell's customer and partner conference Monday, the company announced enhancements to the Dell AI Factory with Nvidia, spanning agentic AI, data orchestration and next-generation rack-scale infrastructure.
"On-prem matters more than ever. It's being driven by sovereignty, security, performance, increasing costs and how you take advantage and take control of token economics," said Sam Grocott, Dell SVP of products, speaking at a briefing for technology press and analysts last week.
Frontier models, on-premises
Dell is partnering with the developers of frontier AI models to advance sovereign AI.
European LLM provider Mistral AI is using the Dell AI Factory with Nvidia for its own large language model training and deployment, Dell said. Mistral's platform runs on liquid-cooled Dell PowerRack with Dell PowerEdge XE9712 servers and Nvidia GB200 NVL72 rack-scale systems.
Beyond Mistral, Dell announced that Reflection's open-source frontier AI models are coming on-premises on the AI Factory, offering a U.S.-based option for regulated industries, governments and sovereign entities. And Dell Enterprise Hub on Hugging Face gives enterprises on-premises access to a curated collection of open-weight models including DeepSeek and others, optimized for AI Factory infrastructure.
Additionally, Google and Dell are collaborating to bring Gemini 3 Flash models to fully on-premises confidential computing environments. The models run on Google Distributed Cloud on Dell PowerEdge XE9780 servers. The integration supports the sovereignty, security attestation and data residency requirements that European and Asia-Pacific operators face as they build sovereign cloud offerings for government and regulated enterprise customers.
Bringing frontier AI models to on-premises deployment is important, and it's particularly significant that enterprises have a domestically sourced option — not controlled by a foreign government or company, said Caitlin Gordon, Dell VP of private cloud and AI solutions.
The current round of AI Factory upgrades comes two months after previous enhancements to boost scalability.
Customers putting it to work
Two major customer announcements show what on-premises AI at scale looks like in practice.
Samsung Electronics is using Dell AI solutions across its semiconductor design, manufacturing and automation operations globally. Inside Samsung's fabs, AI agents analyze equipment telemetry and inspection outputs to support digital twins and yield optimization, with Dell providing the compute, storage and data movement infrastructure to run those agents continuously without interruption.
Eli Lilly has been running on Dell infrastructure for 15 years and is now using it for AI. Dell's storage infrastructure feeds LillyPod, Lilly's AI supercomputer, at nearly two terabytes per second of read bandwidth across the cluster, keeping more than 1,000 GPUs fully utilized for large-scale AI model training.
Lilly, like other leading enterprises, is extending its existing infrastructure to implement AI effectively.
Sovereign AI is becoming an enterprise priority
Worldwide spending on AI is forecast to total $2.52 trillion in 2026, a 44% increase year-over-year, according to Gartner, with AI-optimized server end-user spending expected to reach $421 billion this year.
Sovereign AI in particular is becoming more important to enterprises, and an increasingly lucrative opportunity for vendors and telcos. Sovereign cloud infrastructure spending is forecast to total $80 billion this year, up 35.6% year-over-year, according to Gartner.
Some 95% of organizations say private and sovereign AI are important, but only 29% are making sovereign AI a concrete near-term priority, according to NTT Data's 2026 Global AI Report. Roughly 60% of AI leaders cite cross-border data restrictions as a major challenge, and only 38% say they have high confidence in their cloud security posture. That gap between recognition and action is the infrastructure problem Dell is positioning itself to solve — and an opportunity for telcos as well.
AI goes deskside
Dell also launched Deskside Agentic AI, a new solution powered by Dell's high-performance workstations and Nvidia NemoClaw, to allow enterprises to build and run agentic AI locally with data that never leaves the device.
The solution can save money for enterprises: organizations can expect up to 87% reduced spend versus public cloud APIs over two years, with break-even in as little as three months, said Jon Siegal, Dell SVP of client solutions, at the briefing last week.
Finally, Dell debuted a new PowerRack for compute, networking and storage engineered as one system, deployable in under six and a half hours after delivery, managed through a single control plane. The rack-scale approach gives enterprises building AI factory infrastructure a repeatable unit they can deploy and manage without custom integration overhead.
This article was updated on Tuesday, May 19, 2026, at 12:15 p.m., with a correction: Eli Lilly's machine learning models run on the automated visual inspection machines, not on a Dell server. The digital twin work was run on cloud computing.
For more about how Lilly and other leading enterprises are extending their IT infrastructure to implement AI, see our recent report: "AI doesn't replace. It extends: A practical guide to retooling enterprise infrastructure and teams for the AI age."