Microsoft AI Chief Wants to Deliver State-of-the-Art AI Models by 2027: Report

Microsoft's AI chief has reportedly set his sights on developing cutting-edge artificial intelligence (AI) models by 2027. As per the report, the executive wants the Redmond-based tech giant to become self-sustaining in the AI space and handle all the different layers of the technology, from consumer-end implementation to development of foundational models, in-house. The company's ambition to become a leader in developing state-of-the-art (SOTA) models began after the collapse of the previous deal with OpenAI last year.

Microsoft Wants to Develop SOTA AI Models by 2027

Among the big tech companies, Microsoft was particularly late to the AI party. While the initial chatbot (known as Bing Chat in 2023) came early, it was essentially a ChatGPT wrapper, which came as part of the partnership between OpenAI and the Windows maker. The first real implementation of the chatbot, later named Copilot, did not come until September 2023, when the company began rolling out the AI assistant to Windows 11 as a preview.

Factors were holding the tech giant back. Microsoft was one of the early backers of OpenAI, spending a massive $1 billion (roughly Rs. 9,330 crore) on the ChatGPT maker. As part of the deal, while the Windows maker received early access to all the new AI models developed by the ChatGPT maker, it barred Microsoft from developing AI models of its own.

While it appears as a blunder in hindsight, at the time, it likely appeared as a safe bet for the company, which was benefitting from OpenAI's research and development, and focusing on just the integration and experience-building front. However, in 2026, that strategy means Microsoft will always be one step behind its rivals, who are capitalising on building AI models with new and advanced capabilities.

In late 2025, things changed for Microsoft, however. The collapse of the original deal and the subsequent new deal made the rules more flexible and allowed the tech giant to both partner with other AI providers and develop in-house models. Soon after, Microsoft AI's Chief Executive Officer, Mustafa Suleyman, gave an interview highlighting the importance of becoming self-sustaining in the AI space.

Now, in a new interview, Suleyman said (via Bloomberg), “We must deliver the absolute frontier.” The executive added that the company will start releasing state-of-the-art (SOTA) AI models that can both respond to and generate text, images, and audio.

SOTA, in the world of AI, means a large language model (LLM) which outperforms rival models of the same size across one or more parameters. The performance is typically judged via various third-party and internal benchmark evaluations conducted by the company, as well as independent researchers.

Microsoft's ambitions are big, but bridging the gap will not be easy, given that rivals, such as Google, Anthropic, OpenAI, and Meta, have now spent years on AI research. The company is reportedly feeling the pressure. As per Bloomberg, Suleyman, who was hired in 2024 to oversee the company's integration of AI into consumer products, witnessed his role shrink to just model development recently. Part of the tech giant's reorganisation effort, former Snap SVP Jacob Andreou was handed the charge of Copilot assistant for both end consumers as well as enterprise clients.