AI infrastructure is often framed as a race for power, land, chips, and capital. But a deeper shift is underway: the end user is no longer only human. As AI systems become more autonomous, infrastructure must support new demands around deployment, identity, trust, and control — while still being financed on long-lived assumptions. This panel brings together perspectives from capital, enterprise deployment, identity infrastructure, and strategy to examine what changes when agentic systems become the workload, the operator, or the customer.