Knowledge on the intensity and extension of current subsurface urban heat islands (SUHI) is not only based on the availability of spatiotemporal high-resolution and long-term groundwater monitoring data but also in-depth investigations on the role of single natural and anthropogenic factors. A holistic city-scale 3D FEM model is presented to introduce possible thermal management applications in the Milan metropolitan area such as: (1) understanding the hydro-thermal regime of the urban aquifer disentangling the thermal contribution of natural and anthropogenic heat sources, (2) quantifying the geothermal potential and (3) investigating the effects of urbanization and climate change scenarios. Focusing on the most relevant heat sources (boundaries) and transport mechanisms (parameters), this work deals with (I) the reconstruction of large-scale aquifer heterogeneities to consider the advective dominated heat transport, (II) the accurate definition of the upper thermal boundary by a coupled analytical solution and (III) the integration of natural and human-related fluid/heat sources as transient boundary conditions. The model was calibrated against 15 groundwater head and temperature time series and validated in space and time by temperature profiles at 40 additional observation wells. Thus, a fluid and heat budget analysis revealed the most relevant natural and anthropogenic sources at the city-scale. The heat flow from buildings, surface infrastructures and tunnels contribute to 85% of the net annual heat accumulation in the subsurface which totals to 1.4 PJ/y. The results of the simulations were used to evaluate the geothermal potential of the shallow aquifer and to localize promising and critical areas that should be further investigated for an effective thermal management. Finally, it was demonstrated that possible future climate change and city expansion scenarios could lead to the highest thermal energy increment in the subsurface compared to shallow geothermics development which, for this reason, should be highly supported.