The enterprise data center was supposed to be dead by now, but the reality is on-premises workloads are getting new life courtesy of AI workloads.
That's a key take away from Constellation Research analyst Holger Mueller. In a report, "The Case for Data Centers Is Alive and Well in 2025," Mueller acknowledged the reality for on-premises infrastructure and said "workloads are not moving to the public cloud as fast as the prognosis looked a decade ago, so enterprises need to operate in a hybrid cloud environment, with critical enterprise workloads operating both in the public cloud and on-premises, in the private cloud."
Here are a few reasons why hybrid cloud is here to stay:
- CxOs have to support a range of workloads. Yes, AI agents get the headlines, but enterprises are taking AI all the way to the edge. Those edge use cases are more on-prem.
- Data center utilization needs make on-premises more cost effective. Performance in many regions is still better with on-prem data centers even with cloud advances.
- IT operations are complex and require more heterogeneity. Data centers are a nice hedge for shifting business priorities, leverage vs. public cloud providers and budgeting (capital expenses vs. operating expenses).
- Compliance is more of an issue due to large software portfolios, data residency and regulations that make data centers more attractive.
- AI automation will need on-prem platforms and predictable workloads run better on data centers.
Mueller noted that every enterprise will have a different take on on-premises computing. Variables include where SAP workloads will reside, data gravity and innovation from cloud vendors vs. hardware vendors.
Bottom line: Data centers have been declared dead for years, but CxOs can see lower cost of ownership and perks as long as they keep options open with a cloudlike consumption model and can leverage high performance computing.
Related:
- There Is Only ‘One’ (1) IT Strategy in 2025 and Beyond
- AI infrastructure becomes one big leveraged bet
- AI data center building boom: Four themes to know
- Is AI data center buildout a case of irrational exuberance?
- The generative AI buildout, overcapacity and what history tells us
- On-premises AI enterprise workloads? Infrastructure, budgets starting to align
