Oracle Explores Flexible Data Center Models, Including Customer-Owned Hardware

Oracle Explores Flexible Data Center Models, Including Customer-Owned Hardware

December 15, 2025

In a strategic shift to manage capital intensity and cater to diverse client demands in the competitive cloud and AI infrastructure market, Oracle Corporation is considering new, flexible business models for its global data center operations. The move signals a potential departure from the traditional capital-heavy approach of hyperscale cloud providers, offering customers more control over their hardware investments.

During the company's recent quarterly earnings call, Oracle co-CEO Clay Magouryk outlined several deployment models under consideration. One notable possibility is allowing enterprise customers to bring their own servers and hardware into Oracle's data center facilities. "In some cases, customers actually want to bring their own hardware, in which case, we don't have any capital expense," Magouryk explained. He noted that Oracle's costs in such a scenario would be limited to the data center shell, potential networking gear, and associated labor.

Magouryk detailed other models, including arrangements where third-party vendors, such as chip manufacturers, would lease capacity back from Oracle after their hardware is installed. This model, which has gained traction among so-called "neocloud" providers, was highlighted by a report from The Information. It mirrors recent high-value industry deals, such as Nvidia's agreement to lease back approximately $6.3 billion in compute capacity from CoreWeave and a separate $1.5 billion deal with Lambda.

The executive also shed light on the dynamic nature of current AI infrastructure demand, revealing that Oracle frequently adjusts capacity for large-scale clients. He stated that the company often handles requests from customers who "might sign up for a few thousand of one type of GPU, and then they'll come back and say, 'Well, actually, I'd like to get even more capacity somewhere else. Will you take this back?'" According to Magouryk, provisioning such capacity for major AI model providers typically takes two to three days.

This exploration of hybrid ownership models could significantly impact the cloud and colocation industry landscape. If implemented, it would provide large enterprises and specialized AI firms with a middle ground between full colocation and managed cloud services, potentially improving cost efficiency and hardware flexibility. It also represents a strategic move by Oracle to accelerate its data center expansion with varied financing options, reducing upfront capital burdens while capturing demand from clients seeking bespoke infrastructure solutions.

Source: datacenterdynamics

Read Also
Lubbock City Council to Decide Fate of Proposed AI-Powered Hyperscale Data Center
McDuffie County Considers Proposal for Major New Data Center Development
Data center issuers get cracking on bumper year with $1.45bn of deals

Research