Microsoft Unveils 7 AI Models as Suleyman Pushes Independent Superintelligence Strategy
Updated
Updated · The Verge · Jun 8
Microsoft Unveils 7 AI Models as Suleyman Pushes Independent Superintelligence Strategy
3 articles · Updated · The Verge · Jun 8
Summary
Seven new Microsoft AI models debuted at Build, where Mustafa Suleyman said the company is now building frontier systems across text, image, audio and coding under an independent superintelligence effort.
October’s revised OpenAI contract gave Microsoft room to keep licensing OpenAI models while pursuing its own stack, and Suleyman said that shift has driven new model training, chip deployment and team hiring over the past 15 to 18 months.
MAI-Thinking-1 is Microsoft’s flagship reasoning model, which Suleyman said scores 97% on AIME and runs on Maia 200 chips that are 30% cheaper than Nvidia GB200 systems, with 1.4x better performance per watt after co-optimization.
Suleyman said Microsoft avoided distilling rival models because it wants to exceed existing systems rather than copy them, while still using OpenAI, Anthropic and more than 11,000 models in Foundry where they best fit customer needs.
He argued superintelligence is nearing but will automate tasks rather than eliminate whole jobs, and said AI must prove its value through practical uses such as enterprise tools and a new Mayo Clinic healthcare partnership.
Is Microsoft building its own AI future, or just racing to catch up after losing control of OpenAI?
As user trust in its AI plummets, can Microsoft's $190 billion bet on superintelligence actually succeed?
Microsoft says AI isn't conscious, but can it prevent us from falling for the convincing illusion it's creating?
Microsoft’s MAI Family and Maia 200 Chip: The 2026 Push for Proprietary, Trusted AI at Scale
Overview
In June 2026, Microsoft unveiled a new family of seven proprietary AI models under the 'MAI' brand, marking a major strategic shift led by Mustafa Suleyman. By developing these models from scratch and building them internally, Microsoft is moving towards greater self-sufficiency in AI, aiming to reduce reliance on external partners like OpenAI and Anthropic. This approach is designed to foster trusted, in-house AI capabilities, enhance efficiency, and ensure seamless integration across Microsoft’s products and services. The company’s commitment to owning the foundational layers of its AI infrastructure highlights its focus on control, transparency, and innovation.