Microsoft and Dell believe the answer to rising cloud token costs sits on every employee’s desk

Enterprises are adopting an agentic AI PC strategy as Copilot+ PCs shift AI workloads from expensive cloud inference to secure, high-performance on-device processing.

As agentic AI moves from experimentation into production, enterprises are confronting a new economic reality: Cloud token costs are rising faster than budgets can absorb them, and the device on every employee’s desk may be the most underutilized asset in the AI stack. The question is no longer whether to run AI on-premises, but how to govern it when you do, according to Jon Siegel (pictured, right), senior vice president of Dell portfolio marketing at Dell Technologies Inc.

“What was good enough a year ago is no longer good enough in the AI era,” Siegel said. “Our customers need to get to the latest Copilot+ PCs so they can actually run agentic AI workflows, because they are coming fast. The smaller, more performant models are coming really soon — like 13 billion parameter-type models — and those are going to be really powerful assistants for the average user, and it’s critical that they have PCs that can support that.”

Siegel and Mary Ann Anderson (left), worldwide marketing director of Dell partnerships at Microsoft Corp., spoke with theCUBE’s Dave Vellante and Gemma Allen at Dell Technologies World 2026, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the agentic AI PC strategy, the evolving Dell-Microsoft partnership and accelerating enterprise AI readiness. (* Disclosure below.)

Agentic AI PC strategy takes shape

The economics of running agentic workflows in the cloud are beginning to break down at scale. Dell’s new Deskside Agentic AI offering — announced at Dell Technologies World 2026 — lets enterprises run autonomous agents locally on high-performance workstations, with a break-even against public cloud costs projected in as few as three months, Siegel noted. The concept resonates with customers who are watching token consumption compound as agents loop through multi-step workflows.

“The PC is a great place to start,” Siegel said. “When you have a free token generator, you give your employees a free place to experiment, to innovate, and then when they want to run into production, they could then run it in another spot like in the data center or in the cloud.”

The opportunity from Microsoft’s perspective is one of governance as much as infrastructure. Windows is evolving beyond an operating system into  a “canvas” for agentic AI — a platform where control, identity and data sovereignty converge, according to Anderson. Tools such as Windows AI Foundry give developers a structured path to build, package and scale agents inside a governed environment, while Microsoft Intune and Entra provide the identity layer for those agents.

“With Windows, we’re going beyond just it being an operating system — we’re thinking about it as really a canvas for AI,” Anderson said. “Pretty soon there’s going to be a point where individual employees are managing their own agents, maybe a small army of agents. You have to do that in a way that is governed and is very secure.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of Dell Technologies World 2026:

(* Disclosure: TheCUBE is a paid media partner for Dell Technologies World. Neither Dell, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE