Microsoft and OpenAI have announced a new agreement to restructure their long-term tech partnership. Under the new terms, Microsoft gives up its status as the sole cloud provider infrastructure for training the AI model. OpenAI has instead developed a new AI infrastructure agreement called Stargate with SoftBank, Oracle, and others.
The new terms allow Microsoft to enjoy the ‘right of first refusal’ on OpenAI's cloud computing capacity. Essentially, that means Microsoft has the right to choose the first preference for AI workloads. It can then assign these if it is not in a position to meet their demands.
Microsoft's blog post confirmed that OpenAI had made a significant commitment to Azure. Thus ensuring that all OpenAI products and model training are supported. The deal also allows OpenAI to build additional infrastructure primarily for research and training purposes. Hence addressing the company's increasing need for computing power.
The new deal has been reached amid the two companies' cloud computing capacity tensions. OpenAI has cited the lack of available computing as a major reason for product delays. In June, Microsoft allowed OpenAI to seek additional capacity from Oracle after facing shareholder pressure.
The partnership between Microsoft and OpenAI remains strong, with key terms staying in place through 2030. These include continued access to OpenAI's intellectual property, revenue sharing arrangements, and exclusivity for OpenAI’s APIs. However, this may change if OpenAI achieves artificial general intelligence (AGI), as Microsoft would lose access to OpenAI’s technology if that happens.
In a surprising twist, OpenAI is reportedly considering nullifying its agreement with Microsoft to secure additional funding. This potential move highlights the growing competition in the AI space and the evolving nature of their collaboration.