Dell COO: Agentic AI Reshaping Data Center Economics

Dell Technologies says agentic AI is fundamentally changing data center economics and architecture. Token consumption for reasoning has surged 320 times, even as token costs fell 80% in a single year. What's interesting is that Dell believes cloud-only enterprise strategies will become unsustainable due to the continuous demand for inference. Jeff Clarke, Dell's COO, notes that autonomous systems generate constant token requests, unlike human-driven tasks. The company highlights that 83% of the world's data is on-premises, not in the cloud. Moving this data to the cloud for AI agents creates latency and high costs. Dell's solution is a "deskside to data center" strategy, running AI where the data lives. Dell's AI-optimized server revenue was $9 billion in Q4 FY2026, up 342% year-over-year, and they project about $50 billion for FY2027. This growth has also improved their Infrastructure Solutions Group’s operating margins. The bottom line is that enterprises face an AI execution problem, as 90% of their data is unstructured and not ready for agentic AI. The company says the correct approach is to move AI to the data, not the other way around. This could impact how businesses plan their AI infrastructure.