Microsoft and OpenAI have signed a new definitive agreement, marking the next phase of their partnership that began in 2019. The tech giant's stake in the AI research lab is now worth approximately $135 billion, translating to roughly 27% ownership on a fully diluted basis. Previously, Microsoft held a 32.5% stake before recent funding rounds. The agreement supports OpenAI's board forming a public benefit corporation while preserving Microsoft's position as their exclusive frontier model partner.

Key changes include extending Microsoft's intellectual property rights through 2032, with an independent expert panel now verifying when artificial general intelligence is achieved. Microsoft's IP rights cover both models and products, including those developed after AGI, with appropriate safety guardrails. Research IP—defined as confidential methods used in model and system development—remains protected until either AGI is verified or through 2030, whichever comes first.

OpenAI gains new flexibility under the agreement: the ability to jointly develop products with third parties, release open weight models meeting capability criteria, and serve US government national security customers regardless of cloud provider. The company has contracted to purchase $250 billion in Azure services, while Microsoft no longer holds right of first refusal for OpenAI's compute needs. Microsoft can now independently pursue AGI development or partner with others, though any resulting models would face compute thresholds significantly larger than today's leading systems.