Microsoft appears to be on the verge of competing with partner company OpenAI with its own home-grown AI models called “MAI-1.”

Microsoft has invested billions in OpenAI, giving it access to the company’s industry-leading AI technology. Despite building Copilot around the AI firm’s tech, as well as incorporating it in Bing, The Information, via Reuters, is reporting that Microsoft is preparing to deploy its own large-scale AI model that will compete with ChatGPT.

The Redmond company has already released a number of AI models, but they have been very small in comparison to ChatGPT, which boasts more than 1 trillion parameters. The Information reports that competing models from Meta and Mistral only have 70 billion parameters. While not exactly on par with ChatGPT, Microsoft’s MAI-1 will be much closer to OpenAI’s model with some 500 billion parameters.

Interestingly, the outlet reports that MAI-1’s development is being led by Mustafa Suleyman, a DeepMind cofounder and CEO of Inflection AI, before joining Microsoft in March 2024.

Fallout From the OpenAI Board Debacle?

Microsoft’s decision to create a competing AI model is an interesting one, given the amount of money it has invested in OpenAI, as well as the close relationship the two companies have. One can’t help but wonder if Microsoft is at least partially motivated by the OpenAI board firing CEO Sam Altman last year, a move that caused significant damage to the OpenAI’s reputation—both inside and outside the company.

At the time, Microsoft was quick to hire Altman and OpenAI President and cofounder Greg Brockman, as well as make an open offer to any and all OpenAI employees who wanted to join them at Microsoft. While the situation was ultimately resolved, with Altman and Brock returning to their previous positions, it’s telling that Microsoft CEO Satya Nadella made clear at the time that Microsoft could continue innovating on its own…without OpenAI if necessary.

Nadella’s statement would seem to indicate that whatever agreement exists between Microsoft and OpenAI may give the former the ability to use the latter’s tech as the foundation for its own research rather than being limited to using OpenAI’s models in finished products.

Companies like Microsoft are practically allergic to the kind of drama that OpenAI stirred up last year, and certainly don’t want critical technology they rely on at the mercy of an unreliable partner. It’s entirely possible that the OpenAI board’s antics may have provided additional motivation for Microsoft to limit its dependence on the company.