Microsoft and OpenAI, have signed a new definitive agreement that alters key terms of their relationship and frees Microsoft to pursue artificial general intelligence (AGI) independently. Under the previous deal, Microsoft was reportedly restricted from pursuing A(AG)I on its own through 2030, but that clause has now been amended.
Following the new agreement, Microsoft AI CEO Mustafa Suleyman has signaled a major strategic shift. “Microsoft needs to be self-sufficient in AI,” Suleyman said, adding that the company must “train frontier models of all scales with our own data and compute at the state-of-the-art level.” To support this, Microsoft has created a new “MAI Superintelligence” team to develop a “world-class, frontier-grade research capability in-house.”
This new posture is a significant change from Suleyman’s comments earlier in the year, when he stated Microsoft’s strategy was to “play a very tight second” to OpenAI and develop models that were “3 to 6 months behind.”
The new agreement, which follows a period of reported tension over OpenAI’s plans to become a for-profit venture, includes other new clauses. OpenAI can no longer prematurely declare it has achieved AGI; instead, an independent expert panel will be appointed to verify such a claim. Microsoft, which has an approximately $13 billion stake in the AI lab, will also retain its IP rights to all OpenAI models and products, including post-AGI models, through 2032.
