Microsoft wants approval for 15 massive data centers on the former Foxconn site in Mount Pleasant, on top of two already under construction, turning the area into a large-scale campus.
Each building is roughly 580,000 square feet, so we’re talking multi-million-square-foot capacity that can easily support current and future GPU-heavy AI workloads.
The location along I‑94 signals a classic hyperscaler play: plenty of land, regional connectivity, and room to bolt on more power and possibly on-site energy infrastructure later.
Brad Smith is pushing a “community-first” AI story with higher utility rates, water minimization, and replenishment, which reads as a preemptive move to manage local pushback over power and water use.
Scrapping prior Caledonia plans and consolidating here suggests Microsoft is optimizing for a denser, easier-to-serve power and network footprint rather than scattering smaller sites.
Hints that Microsoft could expand into power-related facilities underscore how central energy control is becoming to hyperscale AI buildouts.
Worth a read if you track where the next big AI data center clusters, and their supporting power footprint, are forming.
Source: Microsoft in Wisconsin; Mount Pleasant may become data center hub | FOX6 Milwaukee
