After investing $13 billion in OpenAI, Microsoft unveils cheaper in-house AI models

Microsoft is trying to turn its massive AI investments into lower-cost models of its own, CNBC reported.

What's happening?

According to CNBC, the company introduced new in-house AI tools at its Build developer conference in San Francisco. Microsoft hopes they will write code and solve complex problems more efficiently than pricier third-party systems.

Microsoft unveiled MAI-Code-1-Flash, its first model designed to turn plain language prompts into source code for apps and websites. It also introduced MAI-Thinking-1, a reasoning model that it says is built to deliver strong performance at a lower token cost, a key pricing metric for developers using AI.

The announcement marks a notable shift for Microsoft, which has been one of the biggest backers of outside AI firms.

CNBC stated that the company has invested $13 billion in OpenAI and another $5 billion in Anthropic. It's also offering those companies' models through its Azure cloud platform.

Microsoft now appears to be pushing deeper into the AI stack with alternatives of its own. MAI-Thinking-1 is entering private preview through Microsoft Foundry. MAI-Code-1-Flash will be a part of GitHub Copilot and Visual Studio Code.

Why do Microsoft's AI models matter?

Cost is a central issue. As demand for AI coding tools rises, so do the expenses tied to using top-tier proprietary models.

By running its own models on Azure, Microsoft can avoid paying outside providers and potentially offer customers cheaper options.

CNBC reported that Microsoft AI CEO Mustafa Suleyman said that the company tailored its models for consulting firm McKinsey. After doing so, the company achieved 10 times better cost efficiency than OpenAI's GPT 5-5.

While AI can help optimize clean energy systems and accelerate research, it can also increase electricity demand and strain local resources.

What's next?

CNBC stated that Microsoft is framing these new models as more efficient tools.

The company is also giving customers ways to tailor the systems. Microsoft said businesses could make MAI-Thinking-1 more accurate by adding their own data. That could make the model more useful for company-specific tasks without requiring a fully custom AI system.

Microsoft also announced updated speech, voice, and image generation models. Smaller Aion models will be able to run locally on Windows devices, potentially reducing reliance on cloud computing for some tasks.

"What you just saw is a pretty significant shift," Microsoft CEO Satya Nadella said at the conference. "We believe the time has come for every company to just move from consuming a frontier model to fully participating at the frontier in the frontier ecosystem."

Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.