Microsoft Shifts More AI Workloads To In-House MAI Models To Cut Costs

Microsoft is increasingly relying on its in-house MAI family of artificial intelligence models instead of the most advanced models from OpenAI and Anthropic, as the company looks to reduce the rising cost of AI operations, according to a Bloomberg report.

Citing a person familiar with Microsoft's AI strategy, the report said that tens of thousands of prompts generated through applications such as Excel and Outlook, which were previously handled by OpenAI and Anthropic models, are now being processed by Microsoft's own MAI models.

The aim is to lower inference costs, although it currently represents only a small portion of the millions of prompts processed weekly by Microsoft Copilot.

The move follows the launch of seven new MAI models, including MAI-Thinking 1, Microsoft's first reasoning model. The company has positioned the 35-billion-parameter model as a cost-efficient alternative capable of delivering strong performance across reasoning and coding tasks. Microsoft has also introduced MAI models for image generation, transcription, voice recognition, and coding.

The report highlights a broader industry trend as enterprises seek to lower AI spending by adopting more efficient models. Several technology companies, including Amazon, Accenture, Meta, and Uber, are reportedly taking steps to reduce AI infrastructure costs.

“Anthropic is extremely expensive and I think many people are urgently looking for alternatives,” Microsoft AI Chief Executive Mustafa Suleyman told Bloomberg in an interview last month. “We pay a lot of money to Anthropic, so our goal is to reduce and ultimately eliminate that cost.”

While OpenAI's models are less expensive than Anthropic's and Microsoft benefits from discounted pricing through its partnership with OpenAI, the company is increasingly investing in its proprietary AI models as it seeks greater cost efficiency and long-term independence.