Key Notes
- Microsoft is reportedly winding down the Oracle capacity deal after older Nvidia chips stopped meeting its requirements.
- The original agreement supplied computing for Bing search and was distinct from the companies’ database partnership.
- Contract renewals are testing whether cloud providers can keep installed AI hardware earning attractive returns as customer requirements change.
Microsoft is winding down an agreement to rent AI computing capacity from Oracle as older Nvidia chips no longer meet its requirements. Bloomberg reported on November 6 that Microsoft has been switching off rented capacity in recent months after declining to renew the contract.
The report cites people familiar with the arrangement. Microsoft declined to comment, and the deal’s full value was not established.
The Computing Deal Began With AI-Powered Bing
Oracle announced the multiyear agreement in November 2023, when Microsoft was expanding conversational search in Bing. The plan combined Oracle Cloud Infrastructure with Microsoft’s own Azure infrastructure to run AI inference: the computation needed to produce answers from an already trained model.
The companies linked the two clouds through Oracle Interconnect for Microsoft Azure. Microsoft could use Azure Kubernetes Service to coordinate computing resources on Oracle’s infrastructure, extending the capacity available for Bing searches without having to supply every server itself.
That arrangement was separate from Oracle Database@Azure, which the companies introduced in September 2023. The database partnership put Oracle database services on infrastructure deployed in Azure data centers. A decision about a particular GPU rental contract therefore should not be read as an announcement ending every Microsoft–Oracle relationship.
Oracle Says Older GPUs Still Find Buyers
Oracle told Bloomberg that demand remains strong, including for earlier-generation GPUs. The company said in September that contracts for four-year-old chips were renewing at a 20% premium to their original prices.
Those accounts can coexist. Hardware that no longer suits one customer’s requirements may still be attractive for another application. Oracle’s original Bing announcement itself described inference as a workload requiring coordinated computing, storage and networking; counting chips alone does not capture how a system performs.
Microsoft’s own infrastructure plans illustrate the importance of the surrounding system. In November 2025, it described its Fairwater data centers as connected AI facilities using Nvidia GB200 NVL72 systems, advanced cooling and dedicated networking. Microsoft emphasized the ability to move data quickly between GPUs and across sites, rather than treating accelerator purchases as the only measure of capacity.
AI Cloud Deals Face a Renewal Test
The business question is what happens when an early AI capacity agreement reaches its renewal date. For a cloud operator, replacing a departing customer can preserve the value of installed equipment. For the buyer, renewing a contract makes sense only if the performance and economics still fit the work it needs done.
That distinction matters as investors examine the financing behind the AI buildout. We recently reported on ING offering about $500 million of data-center loans for sale, including an exposure to the New Mexico Stargate project. A proposed loan sale and an expiring computing contract involve different risks, but both bring attention to the commitments supporting new infrastructure.
Microsoft’s reported pullback makes customer retention and the useful life of AI hardware more consequential. Oracle’s test is whether capacity released by one major buyer can continue earning attractive returns elsewhere.
Disclaimer: AIstify is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, technology, software, and digital innovation sectors. These relationships do not influence AIstify’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.
