Microsoft Reworks AI Compute Sourcing as Oracle Capacity Winds Down and IREN Project Slips

Microsoft is reassessing how it secures computing capacity for artificial-intelligence workloads, winding down an arrangement to rent AI infrastructure from Oracle while requesting network changes that delayed a major IREN project.

Reuters reported on November 6, citing Bloomberg News and people familiar with the matter, that Microsoft had been turning off rented computing capacity from Oracle in recent months and declined to renew the contract. The capacity reportedly became less useful as Microsoft’s requirements evolved, with the Oracle hardware relying on an older generation of Nvidia chips.

The development highlights the infrastructure challenges facing Microsoft as it continues expanding Azure and AI services. Rather than simply increasing external capacity, the company appears to be reassessing whether individual facilities and hardware configurations match its changing workload requirements.

IREN delays Microsoft infrastructure target

Separately, AI data-center company IREN said on November 5 that it had delayed some infrastructure deliveries after Microsoft requested changes to the network design. IREN moved its target for bringing more than $4 billion in annualized revenue capacity online from the end of 2026 to January, with a possible further delay until late February.

IREN’s update does not establish that Microsoft’s Oracle decision and the delayed project are part of the same contract or represent a single strategic move. Together, however, the developments show that Microsoft’s AI buildout remains dependent not only on securing chips and data-center space but also on completing technical and network specifications that meet its operating needs.

What it means for MSFT investors

Microsoft shares closed at $530.61 on November 6, 2026, up 0.94%. The reported infrastructure changes and the stock’s trading performance should be viewed separately because the available reporting does not establish that the Oracle or IREN developments caused the move.

For investors, the more significant issue is how Microsoft balances AI capacity, performance and capital efficiency. Ending or reducing reliance on infrastructure that no longer fits Microsoft’s needs could improve flexibility over time, but project redesigns and delayed deployments may also affect the timing of capacity coming online.

Future updates on Azure demand, Microsoft’s data-center expansion and the timing of IREN’s revised deployment schedule will help clarify whether the changes represent routine optimization or a broader shift in how Microsoft funds and operates its AI infrastructure.