AI’s potential may feel limitless—but the infrastructure powering it is not. As always-on agents drive exponential demand for compute, businesses are colliding with hard constraints in chips, energy, and cost. Leaders also have to make more deliberate choices about when to deploy AI, when to augment human roles, and when work should remain human. This session challenges the assumption that everything should be automated and instead asks: When is AI actually the right choice? Bringing together an enterprise TMT leader and an AI-native founder, we explore a new decision framework for deploying human versus digital labor in a resource-constrained world.