Microsoft announced Microsoft Discovery, a new enterprise agentic AI platform for research and development acceleration, at Microsoft Build 2025 on May 19th. Vice President Aseem Datar revealed Microsoft researchers used the platform's advanced AI models and high-performance computing simulation tools to discover a novel coolant prototype with promising properties for immersion cooling in datacentres in about 200 hours—a process that otherwise would have taken months, if not years.
The platform enables researchers to collaborate with specialised AI agents combined with a graph-based knowledge engine to drive scientific outcomes with speed, scale, and accuracy. Built on Microsoft Azure infrastructure, the platform emphasises trust, compliance, transparency, and governance as key design principles for responsible innovation while keeping researchers in control.
Microsoft Discovery implements a continuous and iterative R&D cycle where researchers guide and orchestrate specialised AI agents that learn and adapt over time through natural language definitions. The platform features Microsoft Copilot as a scientific AI assistant that orchestrates specialised agents based on researcher prompts, identifying which agents to leverage and setting up end-to-end workflows covering the full discovery process.
The platform's graph-based knowledge engine builds nuanced relationships between proprietary data and external scientific research, providing contextual understanding of conflicting theories, diverse experimental results, and underlying assumptions across disciplines. Rather than outputting monolithic answers, the system maintains transparency through detailed source tracking and reasoning.
Microsoft successfully synthesised the discovered coolant prototype in under four months after digital discovery, with initial property tests aligning to AI predictions. The material represents a non-PFAS alternative addressing global efforts to ban "forever chemicals" in favour of environmentally friendly options.
Pacific Northwest National Laboratory scientists are using Microsoft Discovery's advanced generative AI and HPC capabilities to develop machine learning models predicting and optimising complex chemical separations critical for nuclear science, aiming to reduce time scientists spend in hazardous radioactive environments while improving yields and purity.
GSK plans to advance generative platforms for parallel prediction and testing to create medicines with greater speed and precision. The Estée Lauder Companies seeks to leverage proprietary R&D data from nearly 80 years of research to drive breakthrough innovation and high-quality, personalised products. Microsoft plans to integrate NVIDIA ALCHEMI and NVIDIA BioNeMo NIM microservices for materials and life sciences acceleration, plus Synopsys industry solutions for semiconductor engineering enhancement.
The platform enables research teams without computational expertise to drive impact through flexible agent orchestration replacing hard-coded behaviours of specialised digital simulation tools. Integration partnerships with Accenture and Capgemini support custom platform deployments for R&D-intensive sectors. The extensible architecture allows researchers to integrate latest Microsoft innovations with their own models, tools, and datasets alongside partner and open-source solutions, positioning Microsoft Discovery as a comprehensive scientific ecosystem.