According to a fresh report, Microsoft is planning to reduce its dependence on OpenAI models that power its Microsoft 365 Copilot. Notably, the company is aiming to diversify the AI models used in the 365 Copilot and incorporate its own AI models and those from other providers. This report comes following the news of a major Microsoft 365 UI update. This shift in its AI strategy looks to address the cost and performance challenges for Microsoft 365 Copilot users.
When Microsoft announced 365 Copilot in March 2023, its integration with OpenAI's GPT-4 was a key selling point. However, the high costs and slow speeds associated with the model have reportedly prompted Microsoft to reduce its reliance on OpenAI and look for other AI alternatives to improve the performance and affordability for enterprise users. Microsoft has confirmed that its partnership with OpenAI remains intact, it added that the company "incorporates various models from OpenAI and Microsoft depending on the product and experience."
Microsoft is also training its models such as the newly developed Phi-4, and its other open-weight models to make 365 Copilot faster and more efficient. The approach aims to reduce operational costs and potentially pass the same savings benefits to its customers. Not only this, but Microsoft has been applying a similar strategy in other areas of its business. GitHub, acquired by Microsoft in 2018, recently integrated AI models from Anthropic and Google on top of OpenAI's GPT-4.
Additionally, Microsoft's chatbot Copilot also now makes use of an in-house AI model alongside OpenAI's technology. According to sources close to the matter, Microsoft CEO Satya Nadella and other Microsoft leadership are closely monitoring the efforts to strengthen the company's AI offerings, improve cost-effectiveness, and remain competitive in the market. All this signals their importance to the company's long-term strategy.
Source: Reuters
5 Comments - Add comment