OpenAI and Microsoft have revised the terms of their partnership, introducing changes to cloud exclusivity, licensing, and revenue-sharing arrangements as both companies scale their artificial intelligence operations.Under the amended agreement, Microsoft will remain OpenAI’s primary cloud partner, and OpenAI’s products will continue to launch first on Azure. However, OpenAI will now be able to distribute its products through other cloud providers when Microsoft is unable or chooses not to support specific requirements. This marks a shift from the earlier, more tightly coupled cloud arrangement.Microsoft will continue to hold a licence to OpenAI’s intellectual property for its models and products through 2032, though the licence is now non-exclusive. The change allows OpenAI to extend similar access to other partners.

The financial terms have also been modified, which means that Microsoft will no longer pay a revenue share to OpenAI, favouring the former. Instead, OpenAI will continue to share a portion of its revenues with Microsoft through 2030. These payments will remain at the existing percentage rate but will be subject to a total cap.

Microsoft will retain its position as a major shareholder in OpenAI, continuing to participate in the company’s growth.

The AI company says that this amendment simplifies the partnership, but the work done together remains ambitious. “From scaling gigawatts of new datacentre capacity, to collaborating on next- generation silicon, to applying AI to advance cybersecurity, and more, we’re excited to keep partnering to advance and scale AI for people and organisations around the world,” the statement read. This is in line with the other deals OpenAI has signed with companies like Amazon. 

Multi-cloud shift widens OpenAI’s compute access

The revised agreement allows OpenAI to work across cloud providers beyond Azure, enabling it to tap infrastructure from players such as Amazon and Google to meet rising compute requirements. The change comes as model training and deployment demand continues to scale, requiring access to larger pools of specialised hardware across geographies.A key trigger for this shift has been OpenAI’s deepening engagement with Amazon. The arrangement, which includes access to Amazon’s cloud infrastructure and is linked to a commitment reportedly valued at up to $50 billion over time, significantly expands OpenAI’s compute capacity beyond a single-provider setup. The earlier structure with Microsoft had limited such flexibility, particularly for large enterprise deployments.

Tensions over Amazon deal prompted renegotiation

The Amazon engagement had also emerged as a point of friction within the existing partnership. According to reports, Microsoft had raised concerns over its own exclusivity clauses amid OpenAI’s move to deepen ties with a competing cloud provider. A Financial Times report noted that Microsoft had explored legal recourse in response to OpenAI’s expanding Amazon relationship, highlighting the strain the deal had placed on the partnership.The revised terms appear to resolve that impasse by formally allowing OpenAI to operate in a multi-cloud environment while retaining Microsoft as its primary partner.  

Positioning alongside peers in a multi-cloud race

The shift also brings OpenAI’s infrastructure strategy closer to that of rivals such as Anthropic, which has already diversified its compute partnerships across multiple cloud providers. By moving away from exclusivity, OpenAI gains the ability to balance workloads, negotiate capacity, and align deployments with availability across providers.With access to infrastructure from Microsoft, Amazon, and Google, OpenAI is now positioned to scale its models across a broader compute base, reflecting a wider industry trend toward multi-cloud AI development.