As companies respond to rising cloud costs and concerns over data sovereignty by adjusting where AI computing is done, business PCs are drawing attention as endpoints for local inference.

SiliconANGLE reported on Thursday that Dell Technologies said the spread of agentic AI is changing the role and criteria for AI PCs.

Dell cited token costs as the first factor behind renewed growth in demand for AI PCs. Agentic AI workflows involve agents repeatedly asking questions, making plans and requesting again, a structure that can quickly raise cloud API costs. Dell said its "desk-side agentic AI" is focused on easing that burden.

Rob Bruckner (롭 브루크너), president of Dell's Commercial Client Solutions Group, said at Dell Technologies World 2026 that "inference does not necessarily require top-tier inference performance" and that "a right-sized model is often enough." He added that AI PCs allow companies to control token budgets more finely and operate more flexibly, such as using high-performance tokens only when needed.

Data sovereignty has also emerged as an important factor.

Dell said companies are looking for alternatives that let them keep and process sensitive data and intellectual property on devices. Bruckner said local inference lets developers test and improve workflows while being less constrained by token budget limits. He said, "Developers need enough room to experiment, and if they are overly conscious of token costs, attempts at innovation itself can be curtailed."