Why AI’s future is on-premises: Dell Tech World 2026 insights

At this year’s Dell Technologies World, the conversation around artificial intelligence took a distinctly practical turn. While nearly every tech conference now highlights AI, what set this event apart was its emphasis on how businesses can actually execute on AI by embedding more capabilities directly into their infrastructure.

Three major trends are driving companies to boost their on-premises AI capabilities: the rising demand for data and AI sovereignty, the need for tighter governance especially around agents, and the desire for more direct control over critical systems. As Michael Dell, chairman and CEO of Dell Technologies, put it in his opening keynote, “Abundant intelligence is here. Intelligence is becoming infrastructure.”

The shift from cloud-based pilots to full-scale production is revealing significant hurdles. Starting an AI project through a public cloud API is straightforward, but scaling it requires dedicated internal server and compute resources. Without on-premises or hybrid architecture, enterprises face bottlenecks in data capacity and latency, particularly as they move from traditional AI to more complex agentic systems.

One of the most pressing issues is the escalating cost of using cloud-based large language models (LLMs). Dell introduced the concept of “tokenomics” to describe this challenge. Jeff Clarke, Dell’s vice chairman and COO, noted during his Day 2 keynote that token usage for AI has surged by 320-fold, and global token consumption is predicted to grow by 3,400% by 2030. The primary way businesses can reduce these costs is by moving AI workloads from the cloud to on-premises compute, whether that means local workstations, massive data center racks, or edge devices.

Cost isn’t the only driver. Research from Aberdeen shows that companies across all sectors and regions are prioritizing keeping data and AI training out of the cloud and secured in their own data centers. This push for data and AI sovereignty was a major theme at the conference, highlighted by the introduction of the new Dell AI Data Platform.

The need for sovereignty becomes even more critical as businesses adopt agents and agentic systems. With agents, token costs can skyrocket. One case study at the show revealed that a company burned through its entire annual token budget by March after introducing agents. More importantly, agents demand strict security, governance, and control to prevent unintended consequences. “When an agent takes an action on your behalf, you need to know what it did, why it did it, and how to undo it if it got it wrong,” Clarke said.

To address these concerns, Dell announced Dell Deskside Agentic AI, a development offering that combines workstations, Nvidia NemoClaw software, and Dell services. They also introduced support for Nvidia OpenShell, a sandboxed environment designed for building agents while enforcing corporate governance and privacy policies.

Throughout the conference, a central tension emerged: the conflict between moving fast and going slow. Speakers often urged businesses to “move fast” and not be left behind, while practical sessions emphasized starting small, ensuring security, and prioritizing governance. Many of the software solutions touted as answers to AI and agent challenges are still in beta or alpha, and often come with warnings against use in production. This forces companies to carefully assess whether these tools are mature and secure enough for their compliance needs.

Despite these tensions, Dell Technologies World 2026 offered a wealth of practical resources for businesses seeking measured, actionable steps toward building an AI enterprise. From product announcements to detailed demonstrations, the conference provided the tools and guidance needed to navigate the complex landscape of modern AI infrastructure.