Microsoft’s $3.3 billion Mount Pleasant AI data center campus in Wisconsin is now fully operational, the company confirmed this week. The first building, which began limited operations in April, marks a rare milestone in the hyperscaler AI buildout: a megaproject that actually delivered compute capacity.
Key facts
- $3.3 billion facility now fully operational in Mount Pleasant, Wisconsin.
- First building began limited operations in April 2026.
- 550 full-time employees currently support campus operations.
- Dry-cooling system allows water-free operation 90% of the year.
- Site originally intended for Foxconn electronics factory.
Microsoft has completed construction of the first data center facility at its Mount Pleasant campus in Wisconsin, bringing online a project that has become one of the company’s highest-profile AI infrastructure developments According to Data Center Knowledge.
The $3.3 billion facility, which began limited operations in April, is the first completed building on a campus that Microsoft is developing on a site originally intended for a large-scale Foxconn electronics manufacturing project. The data center is now fully operational, Microsoft said this week.
For data center operators, this milestone signifies more than just another hyperscale facility. Microsoft’s Wisconsin campus showcases how hyperscalers are transforming multibillion-dollar AI infrastructure plans into operational reality.
“This is an important milestone because it moves Microsoft’s Wisconsin AI campus from promise to production,” said Sameh Boujelbene, vice president at research firm Dell’Oro Group.
“The AI infrastructure race is no longer just about announcing multi-billion-dollar projects; it is about who can actually bring capacity online, connect it, power it, cool it, and put it to work,” Boujelbene told Data Center Knowledge. “Microsoft completing and operating the first Mount Pleasant facility gives it tangible new AI capacity at a time when compute remains one of the biggest constraints in the market.”
Key Takeaways
- Microsoft’s $3.3B Wisconsin AI data center campus is now operational, delivering compute capacity.
- The site, originally planned for Foxconn, uses dry cooling to cut water use 90%.
From Foxconn Ghost to AI Hub
Microsoft acquired portions of the Wisconsin site after Foxconn dramatically scaled back plans for a manufacturing complex that state officials once projected would create thousands of jobs. Since then, the location has emerged as a major hub in Microsoft’s expanding AI infrastructure footprint. The company said roughly 550 full-time employees currently support operations at the campus.
The conversion of the Foxconn site into AI infrastructure is emblematic of a broader shift: the same public subsidies and industrial land once earmarked for hardware assembly are now being redirected to data centers. The Wisconsin campus sits on land that benefited from state and local tax incentives originally negotiated for Foxconn.
Dry Cooling as a Differentiator
The cooling architecture may be the facility’s most significant operational feature. Microsoft installed a dry-cooling system that relies on evaporative cooling only during the hottest periods of the year, allowing the data center to operate without water consumption for more than 90% of the year, according to Jay Dietrich, research director of sustainability at Uptime Institute.
The Wisconsin facility incorporates a closed-loop cooling system designed to minimize water use, low-carbon construction materials, and habitat restoration efforts across the campus. The company has also highlighted the use of cross-laminated timber in portions of the project as part of broader sustainability initiatives.
The Operational Capacity Gap
The opening comes as hyperscalers race to convert AI investment announcements into operating infrastructure. While billions of dollars in AI data center projects have been announced across the US, far fewer have reached the point of delivering compute capacity for AI training and inference workloads.
Microsoft has committed over $50 billion in AI infrastructure by 2026, per public statements. The Mount Pleasant campus is one of the first visible returns on that spend. Competitors including Google and Amazon are also racing to bring capacity online, but the gap between announced and operational remains wide.
The Wisconsin campus’s delivery of compute capacity matters because AI model training and inference remain compute-constrained. Every new megawatt of operational capacity directly affects training timelines for frontier models and inference costs for enterprise customers.
What to watch
Watch for Microsoft’s next quarterly Azure AI revenue disclosure to see if Wisconsin capacity shows up as incremental inference or training revenue. Also track the buildout pace of the remaining campus buildings — the next facility’s operational date will signal whether Microsoft can replicate this delivery cadence.
Source: datacenterknowledge.com
