Microsoft is making significant strides in the artificial intelligence landscape, focusing on both infrastructure and application development. CEO Satya Nadella has emphasized cost optimization, aiming for a tenfold performance increase for the same computing expenses by refining algorithms within Microsoft's Azure cloud computing services, which are experiencing growing demand. This strategy includes keeping revenue-generating AI services within Microsoft's own data centers for better cost management and using external services like CoreWeave only for short bursts of extra computing power.

Ahead of its Build 2025 conference, Microsoft CTO Kevin Scott outlined advancements in AI agent collaboration and memory capabilities. The company is supporting the Model Context Protocol (MCP) to foster cross-company AI interoperability, envisioning an "agentic web" where AI systems can interact regardless of their developer. Microsoft is also developing “structured retrieval augmentation” to improve AI memory, allowing AI agents to retain essential elements of past conversations, enhancing continuity and reducing repetitive input. This move signals a shift towards context-aware digital coworkers.

Microsoft is also venturing into AI for scientific discovery with Microsoft Discovery, a platform designed to accelerate research by transforming the entire discovery process using agentic AI. This platform assists in scientific knowledge reasoning, hypothesis formulation, candidate generation, and simulation and analysis, enabling scientists and researchers to collaborate with specialized AI agents. This initiative reflects a broader trend of AI labs exploring AI's potential in science, although some researchers remain skeptical about AI's current reliability and problem-solving capabilities.

Furthermore, Microsoft is introducing new APIs for its Edge browser, allowing developers to integrate AI functionality into web apps using models built into Edge. These APIs, mirroring functionality in Google Chrome, provide access to Phi 4 mini, an AI model excelling at math problems, and offer writing assistance APIs for generating, summarizing, and editing text. Microsoft plans to release a translator API for AI-powered language translations via Edge. These APIs prioritize privacy and security by enabling on-device processing, eliminating the need to send data to external cloud services. Additionally, Edge is gaining a translation tool for PDFs, supporting over 70 languages.