Microsoft ends cloud contract with Oracle for AI and re-evaluates infrastructure providers

1–2 minutes

Reason for the change

Microsoft is gradually ending a cloud capacity rental agreement with Oracle, used for artificial intelligence processing, according to Bloomberg. The contract involved third-party servers to provide processing power.

The service became obsolete because it relied on older-generation Nvidia chips that no longer met Microsoft’s specifications. The change is part of a broader re-evaluation of partnerships and spending on AI infrastructure.

Oracle’s perspective

Oracle says demand for its infrastructure remains strong, including for older processors. In September, the company said that GPU capacity whose contracts had reached renewal time was recontracted at prices 20% higher than before.

In the quarter ended August, Oracle’s cloud infrastructure revenue grew 121% year-over-year to $7.4 billion. The company made more than 300,000 GPUs available to AI cloud customers since the end of the previous quarter. Most of these chips were four years old or older.

Previous partnership with OpenAI

In June 2024, Microsoft, Oracle, and OpenAI announced a collaboration to extend Azure’s AI platform to Oracle Cloud Infrastructure, with the goal of providing additional capacity to the creator of ChatGPT.

Re-evaluation of other suppliers

Microsoft is also stepping up oversight of data center projects from suppliers such as Nebius, IREN, and Nscale, known as neoclouds because they focus their offerings on AI infrastructure. The company says its strategy combines its own facilities, leased real estate, and partnerships with third parties, and that it invests in its own chips to control costs and performance.

In the quarter ended September, Microsoft’s capital investments, including leases, totaled $45.2 billion, reflecting the continued need for processing power to meet AI demand.

Conclusion

The review of suppliers is taking place amid the expansion of AI capacity, with equipment choices and delivery times at the center of decision-making.