Enterprises are accelerating AI pilots but face challenges moving them to production, primarily due to networking infrastructure limitations. IDC's survey reveals that security concerns, automation difficulties, and staff talent restrictions are significant obstacles. Agentic AI, with its distributed and dynamic interactions, exacerbates these issues by expanding operational scope and security vulnerabilities. Networking plays a crucial role in enabling these agentic interactions by providing connectivity, security, and consistent governance. It acts as a control plane that applies policy-based controls and supports observability across diverse environments. An infrastructure-level approach to networking is vital for consistent policy implementation and to prevent fragmented visibility and unmanaged agent activities. The choice between best-of-breed solutions and integrated platforms for AI workloads is a strategic consideration. While best-of-breed offers specific capabilities, it can lead to policy inconsistencies and operational complexity. Survey respondents favoring platforms cited stronger security, reduced complexity, and faster deployment as key benefits. A balanced approach combining extensible platforms with modularity to integrate best-of-breed functions is recommended. This allows for flexibility, openness, and adaptability in managing dynamic agentic AI systems. Ultimately, operationalizing agentic AI at scale requires robust networking platforms that ensure cohesive security and governance frameworks.