
Key Points
- 01Microsoft (MSFT) is winding down an AI cloud capacity deal with Oracle (ORCL) after declining to renew the contract
- 02The Oracle (ORCL) agreement relied on older Nvidia (NVDA) chips that no longer match Microsoft (MSFT)’s AI requirements
- 03Microsoft (MSFT) is reviewing more than $80 billion of cloud capacity commitments with outside providers
- 04Oracle (ORCL) says demand for its AI infrastructure remains strong and reports renewing or reselling older GPUs at a 20% premium
Microsoft pulls back from Oracle AI cloud capacity
Microsoft Corp. is winding down an arrangement to rent AI computing capacity from Oracle Corp., after deciding not to renew a contract that used older Nvidia (NVDA) chips. In recent months, Microsoft has been switching off rented Oracle capacity as the company reassesses billions of dollars in cloud partnerships and updates its hardware specifications for AI workloads.
The rented resources had become less useful as Microsoft’s requirements changed, prompting the company to let the agreement lapse rather than extend it. The arrangement dates to 2023, when Oracle agreed to supply servers supporting Microsoft’s AI-powered Bing search engine.
As news of the wind-down circulated on Nov. 6, Oracle shares were down nearly 3% in afternoon trading, while Microsoft’s stock was little changed. Both companies did not immediately respond to requests for comment at the time, and the equipment involved has been described as older-generation Nvidia (NVDA) chips without reference to specific models.
Oracle defends demand for older GPU infrastructure
Oracle has pushed back on the idea that capacity tied to older graphics processing units has lost its appeal. In a Nov. 6 statement highlighting comments from its latest earnings call, the company said AI infrastructure demand remains strong, including for earlier-generation GPUs.
Oracle said GPU capacity that came up for renewal during its fiscal first quarter was either renewed or sold to other customers under long-term contracts at prices 20% above previous agreements. Most of the GPUs covered by those renewals were at least four years old, which Oracle presented as evidence that customers continue to accept older hardware at higher prices.
The company’s response focused on overall demand for Oracle Cloud Infrastructure AI capacity and did not disclose the value of Microsoft’s departing contract or identify the customers taking on capacity that Microsoft is leaving. Oracle has also stressed that Microsoft remains both a partner and a customer across broader cloud collaborations.
Microsoft reviews $80 billion-plus cloud commitments
The reduction in Oracle capacity comes as Microsoft reviews a wide range of external cloud deals. Microsoft has committed more than $80 billion to renting computing capacity from outside providers, including firms such as CoreWeave (CRWV), Nscale and Nebius (NBIS), to support its AI operations.
Several executives have grown concerned about supplier costs and delivery delays, leading Microsoft to tighten oversight of data center construction and withdraw from projects considered risky. The company has declined an option to expand its existing Texas agreement with Nscale, a move that does not affect the current contract or Nscale’s disclosed backlog.
Microsoft has said its infrastructure strategy combines company-owned data centers, leased facilities and outside partnerships, a mix it describes as providing flexibility in bringing computing capacity online reliably and efficiently. Within that framework, the shift away from Oracle’s older Nvidia-based capacity reflects an effort to align external commitments more closely with its current AI hardware standards.
Key Takeaways
- 01Microsoft’s wind-down of the Oracle AI cloud deal is part of a broader effort to realign massive third-party cloud commitments with its evolving AI hardware needs.
- 02Oracle is using renewed and resold contracts at a 20% price premium on older GPUs to argue that its AI infrastructure, even on four-year-old chips, remains commercially attractive.
- 03The developments highlight growing scrutiny of cloud infrastructure costs, delivery performance and regulatory requirements as hyperscalers balance in-house buildouts with leased capacity.
References
- https://www.bloomberg.com/news/articles/2026-11-06/microsoft-is-winding-down-oracle-deal-to-rent-ai-cloud-capacity
- https://www.channelnewsasia.com/business/microsoft-winding-down-oracle-deal-rent-ai-cloud-capacity-bloomberg-news-reports-6532071
- https://cryptobriefing.com/microsoft-winds-down-oracle-ai-cloud-deal/
- https://crypto.news/microsoft-oracle-ai-cloud-capacity-deal/