E2E multi-agent smart digital infrastructure CapEx – Phase IV

E2E Multi-Agent Smart Digital Infrastructure CapEx – Phase IV reimagines how CSPs plan investment as they evolve from network operators into digital infrastructure companies. Building on the award-winning Phase III solution, Phase IV expands end-to-end, intent-based multi-agent AI beyond RAN, FTTH/FWA, Transport and Core to include Data Centers. Instead of network and data center investments being planned separately, the project creates a unified approach to planning infrastructure around real demand, available capital, power, energy, sustainability and business priorities.

At the heart of the innovation is a collaborative ecosystem of specialized AI agents. New Data Center agents forecast compute, storage and interconnect demand, including AI/GPU and edge workloads, then assess site selection, power, cooling, latency and build, lease or partner options. These agents negotiate directly with network planning agents through TM Forum intents, enabling decisions to be optimized across domains. For example, the solution can determine whether deploying an edge data center could avoid or defer a Transport or Core upgrade, rather than assessing each investment independently.

The result is a single, explainable digital infrastructure investment plan that ranks network and Data Center projects together according to financial and sustainability outcomes. By incorporating NPV, IRR, payback, energy costs and CO₂ alongside traditional capacity requirements, the solution is designed to help CSPs direct capital towards the investments delivering the greatest overall value. Target outcomes include at least 15% greater CapEx efficiency, 15% lower energy cost and CO₂ per unit of capacity, reduced stranded capacity, and planning cycles shortened from months to weeks, while retaining human approval of investment decisions.

Phase IV demonstrates how autonomous, standards-based planning can turn increasingly complex infrastructure choices into coordinated and auditable investment decisions. Using TM Forum assets including TMF921, TIO, TMF785 and TMF939, the project will validate the approach against real Champion data in at least two CSP markets. By bringing network, Data Center, energy and sustainability considerations into one intelligent planning framework, the Catalyst aims to help CSPs accelerate AI and edge service monetization, avoid unnecessary infrastructure build, and create a clearer link between infrastructure investment, customer demand and business return.