AI in Materials Discovery
With a new generation of cloud HPC and AI tools, computational chemistry and materials science is scaling and accelerating R&D efforts that can address, some of the world’s toughest challenges. While there are few concrete examples of computational discoveries that have been validated experimentally, a shift is finally starting to take place.
By leveraging specialized AI models for material property prediction combined with HPC, researchers can rapidly screen vast material databases. Microsoft, in a collaboration with Pacific Northwest National Laboratory, winnowed down 32.6 million potential inorganic materials to just 18 promising candidates, for solid-state battery development—all accomplished in a remarkable 80 hours. AI can accelerate material properties estimations without the need for extensive experiments.
Additionally, atomic and molecular simulations using cloud platforms can provide the needed resolution and confidence for finer filtering. By simulating material behavior at an atomic level, researchers can screen vast search spaces for materials with previously unimaginable properties, including novel alloys and catalysts.
We will explore how advancements in HPC and AI are accelerating the next frontier of scientific breakthroughs, in turn compressing the next 250 years of chemistry into the next 25.
BIOGRAPHY
Ken Kaneko is the principal program lead for Azure Quantum Elements at Microsoft where he is focused on accelerating molecular simulation in the cloud that enables new applications for computational chemists and materials scientists. Ken brings in-depth insights and experience in the semiconductor industry. Previously, he worked at Intel in the Portland Technology Development group where ran R&D process engineering group. He holds a bachelor’s degree from Stanford University and a PhD from Carnegie Mellon University
He will share how real progress can be made today by combining high performance computing (HPC), state-of-the-art machine learning, and quantum knowledge to fundamentally transform our ability to model and predict the outcome of chemical processes with application to the semiconductor Industry.