Microsoft used its Build 2026 developer conference to unveil a broad range of artificial intelligence and computing innovations, including seven new in-house AI models, a next-generation quantum chip and a dedicated search platform designed specifically for AI agents.

The announcements signal Microsoft's growing push to build an independent AI ecosystem while reducing its reliance on external model providers. The company is positioning itself to compete more aggressively with rivals such as Google, Anthropic and Meta, particularly in enterprise AI, software development and autonomous agent systems.

The updates also reflect Microsoft's broader ambition to transform Windows and its cloud ecosystem into a platform where AI agents can not only assist users but independently perform tasks across applications and services.

Microsoft launches seven new AI models

At the centre of the announcements was the introduction of seven new models within Microsoft's MAI (Microsoft AI) family.

The models span reasoning, coding, vision, speech and multimodal capabilities, forming a comprehensive AI stack designed to support next-generation agent systems. Microsoft says the models are intended to move beyond traditional chatbot interactions by enabling planning, reasoning, execution and task completion.

Among the most notable releases is MAI-Thinking-1, Microsoft's first flagship reasoning model. Unlike conventional conversational models, reasoning systems are designed to break complex problems into logical steps, perform advanced planning and execute multi-stage tasks autonomously.

Reasoning models have become one of the most competitive areas in artificial intelligence as companies race to develop systems capable of supporting increasingly sophisticated AI agents.

Microsoft says MAI-Thinking-1 performs competitively on key software engineering benchmarks and approaches the capabilities of some of the industry's leading reasoning models.

The company also introduced MAI-Code-1-Flash, a coding model specifically optimised for GitHub workflows. The model is being rolled out to GitHub Copilot users and is designed to assist developers with programming tasks while maintaining high efficiency and lower operating costs.

According to Mustafa Suleyman, Chief Executive Officer of Microsoft AI, the company's focus is increasingly centred on enterprise applications, software development and practical business use cases rather than pursuing model scale alone.

Majorana 2 advances Microsoft's quantum computing ambitions

Alongside its AI announcements, Microsoft unveiled Majorana 2, the latest version of its experimental quantum computing chip.

The chip forms part of Microsoft's long-term investment in topological quantum computing, a more challenging but potentially more stable alternative to approaches being pursued by competitors including Google and IBM.

While the number of qubits increased from eight to 12, Microsoft says the most significant breakthrough lies in stability. The company reported that some qubits on Majorana 2 remained stable for more than a minute, compared with less than 12 milliseconds in the previous generation.

According to Microsoft, the improvement represents more than a 1,000-fold increase in reliability and could significantly accelerate the development of practical quantum computers.

The company also revealed that artificial intelligence played a role in designing the chip. Using its Microsoft Discovery platform, researchers leveraged AI to accelerate material selection and optimise chip architecture, shortening development timelines that traditionally take years.

Microsoft now believes practical quantum computing could arrive as early as 2029, several years ahead of previous projections.



Web IQ targets the growing AI agent economy

Microsoft also introduced Web IQ, a new search and information platform built specifically for AI agents.

Unlike traditional search engines that return ranked results for human users, Web IQ is designed to provide structured, machine-readable information that AI systems can quickly process and act upon.

The platform leverages decades of search expertise from Bing while focusing on the needs of autonomous agents that require real-time access to information, documentation, pricing data and web services.

According to Microsoft, Web IQ can respond to most requests in approximately 165 milliseconds and reduces token consumption by as much as 60% compared to traditional search APIs.

The company says the platform's grounding technology helps AI systems access more reliable information while reducing hallucinations and improving accuracy.

More significantly, Web IQ is designed to support direct interaction between AI agents and online services, allowing systems to complete transactions, access external tools and interact with digital platforms without requiring human intervention.

The development aligns closely with Microsoft's broader vision of an internet increasingly navigated by AI agents rather than traditional web browsers.

Microsoft accelerates its independent AI strategy

The announcements come as Microsoft continues to expand its in-house AI capabilities despite maintaining a close partnership with OpenAI.

Over the past several years, Microsoft's AI products have relied heavily on OpenAI's technology. However, growing competition and the strategic importance of artificial intelligence have pushed the company to invest more aggressively in proprietary models and infrastructure.

Microsoft executives have acknowledged the risks of depending too heavily on a single AI provider and have increasingly emphasised the importance of maintaining technological independence.

The Build 2026 announcements demonstrate that strategy in action. From reasoning models and coding assistants to quantum computing and AI-native search infrastructure, Microsoft is building the foundations of a broader ecosystem designed to compete across every layer of the AI market.

As competition intensifies among technology giants, the company's latest releases highlight how the race for AI leadership is expanding beyond chatbots into reasoning systems, autonomous agents, enterprise platforms and the next generation of computing itself.