Enterprise AI transformation does not stall because organizations lack capable models, sufficient adoption, or technology investment. It stalls at the moment a successful deployment is expected to change who—or what—has the authority to decide. This article identifies that moment as the handoff boundary and argues that enterprise AI value depends on the depth of decision-right transfer rather than the breadth of deployment. By distinguishing between systems that merely augment human judgment and those that are genuinely delegated authority, it offers boards a practical framework for evaluating AI portfolios—and a clear sequencing principle: reallocate authority before scaling spend.