Matlantis – Universal Atomistic Simulator

Matlantis Corporation is a Tokyo-based company, founded on 2021-06-01 as Preferred Computational Chemistry and renamed Matlantis in July 2025. Shareholders are Preferred Networks, ENEOS and Mitsubishi Corporation. It offers an atomic-level AI simulator that predicts material behavior before experiments.

Matlantis is a cloud-based SaaS built on the PFP machine-learning potential, which supports 96 elements without restriction on composition or arrangement. Modules include LightPFP, crystal structure prediction (CSP), reaction path search and PFP descriptors. The site reports more than 150 companies and institutions using it, including Volkswagen, AkzoNobel, Panasonic and Toyota Battery.

• PFP universal potential: AI model covering 96 elements for atomistic simulation.
• LightPFP: lighter model for large-scale molecular dynamics.
• Matlantis CSP: crystal structure prediction module.
• Reaction search: RestScan/ReactionString and GRRM20 integration.
• PFP Descriptors: descriptor outputs for downstream machine-learning models.
• AI agent integration: coding-agent workflows presented in webinars and videos.

A university or institute in Vietnam could screen catalyst or battery-electrolyte candidates by simulation. A semiconductor or electronics materials firm could study thin-film cracking and surface reactions. A chemical producer could test reaction pathways before pilot runs.

Screens materials and reactions by simulation before costly lab work. PFP v9 adds the higher-accuracy r2SCAN mode as default, validated against experimental values for stability, melting point and water viscosity. Holds ISO 27001 and ISO 27017 certification. Site also highlights integration with AI coding agents.

Machine-learning interatomic potentials are replacing slow first-principles calculations in battery, catalyst and semiconductor research. Vendors are adding agent interfaces and GPU toolkits. Vietnamese universities and materials or semiconductor firms may use cloud simulation to avoid building in-house HPC.

Joint venture of Preferred Networks (51% at founding) and ENEOS, with Mitsubishi Corporation joining in 2024-06. Offices in Tokyo and Cambridge, Massachusetts (Matlantis Inc.). Released PFP v9 on 2026-07-16.