Enterprise Data Centers Get Creative to Handle Rising AI Workloads

The massive data center buildout by hyperscalers gets the attention, but there’s a lot of investment and innovation happening in traditional enterprise data centers as well. These are the data centers that run banks, universities, shipping companies, hospitals, and pretty much every other business you can imagine. And they’re also facing pressure to deliver more computing power, support AI workloads, and improve efficiency.

Northwestern University offers a great example. A top-tier research university, Northwestern’s data center team retrofitted its data center (which already was a converted former radio factory) to accommodate up to 100 kW per server rack, including liquid cooling. The project highlights a key challenge for enterprise data centers: expanding how to increase computing capacity and capabilities, while also keeping the lights on for critical operations and research.

“For your community of data center users, they need to not realize what's going on. You need to be as completely in the background as possible while still trying to move forward,” says Michael Korby, Northwestern’s Service Operation Center Manager. “That's one of those challenges when you're not doing a greenfield project.”

Data center operators are feeling the capacity squeeze. The 2026 AFCOM State of the Data Center report reveals that 72% of operators expect AI workloads to increase capacity requirements, and 48% anticipate the impact from rising AI will be significant. The shift is forcing enterprises to rethink traditional infrastructure strategies. To cope with rising rack density and energy efficiency needs, for example, 36% have already deployed liquid cooling, while another 28% plan to adopt it within the next 12–24 months.

Enterprise Data Center Innovation

Korby will be speaking at Data Center World about the challenges of enterprise data center innovation. He’ll be joined by Akshay Viradiya, LinkedIn Senior Staff Data Center Engineer, to share their real-world stories of bringing state-of-the-art data center capabilities into existing facilities. Data Center World runs April 20-23, in Washington D.C.

Data Center World also has a conference session on “Enterprise Data Center Capacity & Capability Planning in the AI Era,” led by Kirk Killian, President of Partners National Mission Critical Facilities. Killian will share case studies from his decades of work advising businesses from banks to airlines to retailers, and provide a practical framework for enterprises that find their data centers can’t accommodate anticipated AI programs.

For many enterprise data center operators, the pressure to expand capacity will rise as AI workloads mature and become embedded into more critical operations. As companies run more AI inference workloads, data center teams will likely find certain tasks require proximity to data and low, predictable latency that are best served by on-premises capacity. The State of the Data Center report finds 67% of respondents are migrating workloads back to on-premises or colocation environments. For many enterprise data centers, this trend means they can no longer defer data center upgrades without limiting their ability to support workloads that are central to operations, whether research, patient care, financial services, or logistics.

Facing similar capacity challenges? Join data center leaders at Data Center World, April 20-23 in Washington, D.C., to explore practical strategies for expanding existing facilities, planning for AI workloads, and bringing new technology into your infrastructure. Secure your seat to Data Center World 2026.