Zayo enables AI agents to execute live network actions

Zayo has introduced a new capability allowing approved artificial intelligence agents to interact directly with its live network infrastructure. The company’s DynamicLink platform now exposes network functions through a production Model Context Protocol (MCP) server, enabling software to gather real-time conditions and execute authorized actions without human intervention. This development marks a shift from traditional Network-as-a-Service (NaaS) models, which rely on human operators using dashboards, toward machine-driven network operations.

What changed

The integration allows AI agents to inspect network states and perform tasks such as rerouting traffic, adjusting bandwidth allocations, or diagnosing issues. Zayo’s implementation focuses on enterprise use cases, where agents can act within predefined parameters set by network operators. The company has emphasized that these agents operate under strict authorization controls, though the specifics of how permissions are enforced remain undisclosed.

While AI-driven network monitoring is not new, Zayo’s approach differs by exposing operational functions rather than just providing data for human review. The move reflects broader industry trends where infrastructure providers are testing the boundaries of automation in critical systems. However, the practical adoption of such capabilities may depend on how enterprises balance efficiency gains against concerns about relinquishing control to autonomous software.

Industry context

Zayo’s announcement arrives amid a period of rapid expansion for its fiber network, driven by demand from artificial intelligence and cloud workloads. Over the past year, the company has secured long-term fiber supply agreements, completed acquisitions to extend its metro fiber reach, and built new long-haul routes optimized for AI traffic. These investments highlight a growing recognition that network capacity, rather than compute hardware alone, is becoming a bottleneck for large-scale AI deployments.

The company’s recent activities include:

• A 15,000-mile network expansion through 2030, supported by a fiber supply deal with Corning.
• The acquisition of Crown Castle’s Fiber Solutions unit in May 2026, adding 90,000 route miles and 40,000 enterprise locations.
• Completion of a 622-mile AI-optimized fiber route across the western United States in late 2025.

These developments underscore how infrastructure providers are adapting to the distributed nature of AI workloads, which require high-capacity connections between training clusters, inference systems, and end users.

Challenges and considerations

The integration of AI agents into live network operations introduces several technical and operational questions. Interoperability remains a key concern, as enterprises may use different AI platforms or proprietary tools that must interface with Zayo’s MCP server. Additionally, accountability frameworks for machine-initiated actions are still evolving, particularly in scenarios where automated decisions could impact service reliability or security.

The broader industry will likely watch how enterprises respond to this model. While automation can improve efficiency, the degree of autonomy that organizations are willing to grant to AI agents in production networks remains uncertain. Zayo’s implementation may serve as a test case for whether such systems can balance innovation with the need for human oversight.