Summary

• Poste Italiane is developing a 70MW datacentre in Lombardy.
• The facility could support large-scale AI training and inference.
• The project links state-backed infrastructure, financial institutions, and Italy’s regional datacentre planning framework.

Poste Italiane is developing a 70MW datacentre in Lombardy, moving Italy’s state-controlled postal operator into one of Europe’s most capital-intensive technology infrastructure markets.

The company’s chief executive, Matteo Del Fante, said during a recent earnings call that the facility could be used to train and run AI models. The likely location has not been disclosed, although land and an energy supplier have already been identified.

“The land has already been identified, the energy supplier has also been identified,” Del Fante said. “We have the support of all the major Italian financial institutions and, for that matter, even corporates that could join forces and create a large hub where we could run both training models and inference models on a large scale.”

Poste Italiane is not a conventional datacentre developer. It is a national infrastructure operator with deep roles in logistics, payments, savings, digital identity, and public service delivery. Its move into compute therefore sits closer to industrial strategy than ordinary real estate expansion.

The plan also lands in a Lombardy market that is growing quickly but becoming more politically sensitive. The region hosts dozens of datacentres, while large new developments have attracted opposition over land use, energy demand, and environmental impact. Lombardy has also enacted a legal framework for datacentre development, encouraging brownfield sites and sustainability considerations through the approval process.

The Italian angle is important because sovereign compute is not only an EU-level concern. National institutions, banks, telecoms providers, public bodies, and large industrial groups are all examining how much high performance AI infrastructure they want to control domestically. A state-linked operator with access to financial institutions and corporate partners could become a useful anchor for that kind of capacity.

Execution will not be simple. Training and inference at scale require more than a power allocation. Poste Italiane and any partners would need cooling design, chips, connectivity, operations teams, software orchestration, security controls, customers, and a commercial model that can compete with established hyperscale and colocation providers.

The possible involvement of TIM, Intesa Sanpaolo, Coima, and other partners points to a consortium-style model, where national infrastructure capacity is assembled from telecoms, finance, property, and public-adjacent operators. That structure reflects the cost and risk of AI infrastructure: few organisations can justify building strategic compute alone, but many want some role in the systems that may carry future workloads.

If the project moves ahead, Poste Italiane will become part of a broader European pattern in which AI infrastructure is treated less like a specialised technology facility and more like a strategic asset class. Italy’s question will be whether the institutional backing can be matched by the engineering and operational maturity needed to run the workloads that justified the project in the first place.