Year: 2026
Type: preprint
Abstract: Materials science has hit an inflection point with AI, drawing an intense flurry of interest and enormous scales of capital across academia, tech, and traditional industry. To look forward we first take a tour through the development of modern technologies powering this new age, from shared databases to foundation machine learned interaction potentials to agentic laboratories. We conceptualize the technology stack as 3 tiers (data foundations, tools, and orchestration), trace their rapid maturation into infrastructure, and map the modern industrial ecosystem and how prominent parties are making different bets. We use our battery research programme to illustrate how these AI tools can come together in an industrial setting, and point out emerging areas stemming from challenges in materials development and translation. Specifically, we observe that AI progress has to date concentrated on candidate generation and low latency measurements that may or may not correlate with longer time horizon materials development. Drawing on lessons from pharmaceuticals, specialty chemicals design, and semiconductor design automation, we argue that the potential of AI will only be materialized with key contributions from materials scientists: richer mechanistic characterization and judicious encoding of downstream constraints.
Source: ChemRxiv
Language: English
Subfield: Materials Chemistry
Field: Materials Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7202368673