In 2019, Microsoft first invested in OpenAI to advance cutting-edge AI research and democratize AI as a new technology platform. Over the years, OpenAI grew from a small AI research startup into one of the most popular consumer brands on the internet with ChatGPT. To accommodate the growing ambitions of both companies, Microsoft and OpenAI recently made changes to their partnership.
Microsoft remains a major investor in OpenAI, providing funding and compute capacity to support its growth. Microsoft also holds rights to OpenAI IP, which includes models and infrastructure. Also, OpenAI's APIs will be exclusive to the Azure public cloud platform.
Last year, a report claimed Microsoft was developing an in-house LLM called MAI-1 to compete with OpenAI's models. Today, The Information reported that Microsoft has developed its own AI “reasoning” models similar to OpenAI’s o1 and o3-mini. The report also mentioned that the OpenAI team refused to share technical details about o1 with Microsoft, causing tensions between the companies. However, it is unclear if this actually occurred since OpenAI has a legal obligation to share its IP with Microsoft.
Bloomberg today reported that internal test results of Microsoft's new in-house developed models were competitive with state-of-the-art models from OpenAI and Anthropic. Microsoft may begin using these models to power its Copilot assistant for both consumers and enterprise customers. Microsoft also has plans to offer its models via APIs to developers later this year.
A Microsoft spokesperson offered the following comment regarding the stories from Bloomberg and The Information:
As we’ve said previously, we are using a mix of models, which includes continuing our deep partnership with OpenAI, along with models from Microsoft AI and open source models
Microsoft's dual strategy of partnering with OpenAI and developing its own models positions it as a key player in the future of AI.
2 Comments - Add comment