Dell Technologies World 2026 opened today in Las Vegas with a joint keynote from Dell CEO Michael Dell and NVIDIA CEO Jensen Huang. The announcement at the centre of the four-day conference — running May 18 to 21 — is Dell AI Factory 2.0, a significant expansion of the companies’ joint enterprise AI infrastructure initiative.

For business leaders, this is not a story about GPU specs. It is a story about what kind of AI infrastructure is now available to run serious agentic workloads on-premise, at speed, without routing sensitive data through public cloud APIs.

What Was Announced

Dell AI Factory 2.0 is built around NVIDIA’s Blackwell Ultra GPU architecture — the successor to Hopper — and represents a step change in what organisations can run on their own infrastructure.

The new hardware line includes four new PowerEdge servers:

• XE9780 and XE9785 — air-cooled configurations
• XE9780L and XE9785L — liquid-cooled configurations

Each system supports up to 192 NVIDIA Blackwell Ultra GPUs. When configured with a Dell IR7000 rack, that scales to 256 GPUs per rack. Dell claims up to four times faster large language model training compared to the previous XE9680 generation.

On the software side, NVIDIA AI Enterprise is now available directly through Dell, packaged with managed services for the NVIDIA stack. That means enterprises can procure the hardware, software, and ongoing management from a single vendor rather than assembling pieces from multiple suppliers.

The Agentic AI Play

The more strategically interesting part of Dell Technologies World 2026 is the focus on agentic AI at the deskside — AI agents that run locally on workstations powered by NVIDIA Grace Blackwell, rather than in a data centre or the cloud.

The implication is that agentic workflows — where AI doesn’t just respond but takes action, coordinates tasks, and executes multi-step processes — no longer have to live in a cloud environment. Enterprises with strict data residency requirements, regulated industries, or simply those who want lower latency can now run this infrastructure on-premise or on individual workstations.

This aligns with a broader push at Dell Technologies World 2026 toward what Deloitte, HCLTech, and others attending the conference are calling “sovereign AI” — the ability to run AI models on infrastructure you control, with data that stays where you need it to stay.

Why This Matters Right Now

Enterprise AI adoption is moving fast. Research published earlier this year found that 72% of enterprises now have agentic AI in some form of production deployment. The bottleneck is no longer whether to adopt AI — it is how to run it at the quality, speed, and governance level that real business operations require.

The limiting factor for many organisations has been infrastructure: cloud AI APIs are fast to start with but create cost, latency, and compliance headaches at scale. On-premise GPU infrastructure has historically been expensive to procure and complex to manage.

Dell AI Factory 2.0 is a direct response to that bottleneck. By packaging Blackwell servers with NVIDIA AI Enterprise software and Dell managed services, the barrier to on-premise enterprise AI drops considerably.

Three groups of businesses should pay attention:

Regulated industries: Finance, healthcare, legal, and government organisations where data cannot leave the organisation’s control. On-premise Blackwell infrastructure means you can run frontier-level AI without sending data to third-party cloud APIs.

High-volume agentic deployments: Organisations running AI agents at scale — customer service automation, operations agents, internal knowledge systems — where cloud inference costs compound quickly. Local inference at the rack level changes the economics significantly.

Organisations preparing for the next wave: If you are still in pilot mode with AI, the infrastructure ecosystem is becoming mature enough that planning for production-scale deployment is no longer premature. The hardware is here, the software is productised, and the managed services exist to run it.

What This Means for Business

The Dell + NVIDIA partnership has been building toward this for several years, but Dell Technologies World 2026 marks the point where the infrastructure story becomes genuinely enterprise-ready at scale.

A few practical takeaways:

The AI infrastructure race is driving costs down. Competition between Dell, HPE, Supermicro, and cloud providers means that compute is getting more powerful and more accessible every quarter. Businesses that have been waiting for the right moment to invest in AI infrastructure will find more options and better economics in 2026 than they had in 2024.

On-premise AI is a real alternative now. For organisations that have avoided cloud-based AI due to data concerns, the case for on-premise deployment just got stronger. Blackwell-powered local inference is fast enough to run serious production workloads.

Agentic AI infrastructure is the next procurement decision. Most enterprise AI discussions have focused on which model or platform to use. The next decision is about where to run it and how to govern it. Infrastructure decisions made this year will shape what AI an organisation can realistically deploy in 2027 and beyond.

NVIDIA CEO Jensen Huang has framed AI factories as the new data centres — the fundamental infrastructure investment of this era. Today’s keynote at Dell Technologies World 2026 is the clearest signal yet that this framing is becoming enterprise reality.

For teams evaluating when and how to deploy AI across their operations, the infrastructure question is no longer a barrier. It is a decision.

---

If this is the kind of problem agents can help with, the free Working With Claude field guide is the practical next step. Thirty-two pages, no fluff. Get the free guide.