In a significant change to its AI strategy, Microsoft is reportedly reducing reliance on OpenAI’s and Anthropic’s top-tier AI models, opting instead for its own Microsoft AI (MAI) family to manage increasing cost pressures.
- Microsoft shifts AI processing to cost-effective MAI models
- MAI-Thinking 1 reported to match quality of high-end competitors
- Industry-wide pressure to cut AI spending amid rising costs
What happened
Microsoft has begun transitioning a portion of its AI processing workload away from advanced models developed by OpenAI and Anthropic, opting instead to use its internally developed Microsoft AI (MAI) models. This shift reportedly started with rerouting tens of thousands of AI prompts in widely used software like Excel and Outlook. While still a fraction of its overall AI usage, the change indicates a deliberate move to reduce third-party dependency.
This move follows Microsoft's recent launch of seven new MAI models, including MAI-Thinking 1, a mid-sized reasoning model designed to optimize performance and operational costs. According to tests, MAI-Thinking 1 delivers coding capabilities on par with Anthropic's Claude Opus 4.6 while operating more efficiently, highlighting Microsoft's intent to balance capability with cost.
Why it matters
The decision to adopt proprietary AI models reflects growing concern over the escalating costs of accessing top-tier AI systems from third-party providers. Although Microsoft benefits from discounted rates due to its partnership with OpenAI, the expenses associated with high-volume AI usage remain substantial, prompting the company to seek more sustainable alternatives internally.
This cost-conscious approach mirrors broader industry trends, with other major technology firms such as Amazon, Meta, and Uber also exploring ways to lower AI spend. Some have even turned to more affordable options from Chinese AI providers despite potential security considerations. Microsoft's strategy highlights the financial pressures pushing leading companies to innovate internally rather than rely solely on external providers.
What to watch next
Microsoft's continued development and deployment of the MAI model family will be key to its AI strategy and cost management in the coming years. Observers should monitor how effectively these in-house models scale across Microsoft's extensive AI offerings and whether they can fully replace or significantly reduce reliance on OpenAI and Anthropic models.
Additionally, the approaching expiration of Microsoft's deal with OpenAI in 2032 adds a longer-term dimension to this shift. How that relationship evolves, especially in terms of pricing and technology integration, could influence Microsoft's future AI direction as well as competitive dynamics in the AI market.