MATERIALSDECODEDA referenced map of the materials and AI landscape: the open databases, the tools, the companies and the software they sell, the firms that make the materials, and the people behind it all. Every entry links to its own sources. The rule here is the same as everywhere on this site, no claim without a reference.
One honest caveat before the list. AI materials-discovery startups have collectively raised over $1.3B in the past two years, but commercialising a new material is slow and uncertain, and much of the startup work is still unpublished. Treat bold claims, including the funded ones below, with care.Source: PitchBook.
Where the data comes from. Computed and experimental materials data, most of it free to query, spanning the US, Europe and China.
Lawrence Berkeley National Laboratory (US)
The best-known open database of computed properties for ~150,000+ inorganic materials: band gaps, formation energies, elasticity, batteries and more, with a free API. The backbone of much materials informatics.
Duke University (US), Stefano Curtarolo
A high-throughput database of millions of computed compounds with automated workflows and property prediction.
Northwestern University (US), Chris Wolverton
DFT-calculated thermodynamic and structural properties, shared freely, widely used for stability and phase-diagram work.
NIST (US)
An integrated infrastructure of DFT, force-field and machine-learning datasets and tools for materials design.
Meta FAIR + Carnegie Mellon (US)
Large datasets and models for catalysis and inorganic materials, including OMat24, an open dataset of over 100 million DFT calculations.
Open community
An open aggregator of battery cycling and degradation datasets in a common format, for data-driven battery research.
FAIRmat / Max Planck (Germany)
A large open repository of computational materials science data, storing raw and normalised results from many codes under FAIR principles.
EPFL / MARVEL (Switzerland)
An AiiDA-powered platform for sharing curated materials datasets, workflows and interactive tools with full provenance.
Ruhr University Bochum (Germany)
An open, high-throughput DFT database of over 5 million calculations for 3D, 2D and 1D compounds, with convex hulls, phonons and generative models.
Open community
An open collection of experimental crystal structures of organic, inorganic and metal-organic compounds.
FIZ Karlsruhe (Germany)
The main reference database of experimentally determined inorganic crystal structures (subscription).
Institute of Physics, Chinese Academy of Sciences (China)
A large DFT materials database from China, supporting thermodynamic stability assessment, reaction-path inference and AI property prediction.
The open-source code that does the work: analysis libraries, graph models, and machine-learning interatomic potentials.
The core Python library for materials analysis: structures, symmetry, phase diagrams, and access to the Materials Project API.
A widely used Python toolkit for setting up, running and analysing atomistic simulations across many calculators.
A Python library for materials data mining: featurisation, datasets and machine-learning feature engineering.
An equivariant machine-learning interatomic potential, and the basis of foundation models (MACE-MP) approaching quantum accuracy at simulation speed.
A universal machine-learning interatomic potential from the Ceder group, trained across the periodic table.
Universal graph-network interatomic potentials and property models from the Materials Virtual Lab.
Crystal Graph Convolutional Neural Network, the original graph model for predicting properties from crystal structure.
Atomistic Line Graph Neural Network from NIST, a strong structure-based property predictor.
An E(3)-equivariant neural network interatomic potential, highly data-efficient.
Community materials data, packaged for machine learning.
An open, unified materials dataset initiative (LeMat-Bulk) from Entalpic and Hugging Face, harmonising data from Materials Project, OQMD and Alexandria.
The venture-backed wave building AI systems for materials discovery, from Cambridge to San Francisco. Funding figures are as reported.
San Francisco, USA · founded 2025
Founders: Liam Fedus (ex-OpenAI VP Research), Ekin Dogus Cubuk (ex-Google Brain / DeepMind)
Building an 'AI scientist', a foundation lab for atoms that connects large language models to the physical world and runs closed-loop physical experiments, using LLMs as an orchestration layer alongside specialised neural networks.
Funding: About $300M seed round, valuing the company near $1.5B (Sept 2025).
Cambridge, Massachusetts, USA · founded 2023
Founders: Geoffrey von Maltzahn (CEO), Andrew Beam (CTO); backed by Flagship Pioneering
Pursuing 'scientific superintelligence': autonomous AI scientists that design and run their own experiments across materials and life sciences. Claims early results in carbon-capture materials and antibodies (much not yet peer-reviewed).
Funding: About $550M raised; investors include Flagship Pioneering, General Catalyst and Braidwell.
El Segundo, California, USA · founded 2022
Founders: Jonathan Godwin (ex-Google DeepMind)
AI-driven materials discovery focused on materials for the data-centre buildout, such as carbon capture and cooling. Develops its own AI models and validates candidates experimentally.
Funding: About $21M raised.
Redwood City, California, USA · founded 2013
Founders: Bryce Meredig, Greg Mulholland
One of the earliest AI materials-informatics platforms, applying machine learning to materials and chemicals R&D data to speed up product development for industrial clients.
Funding: Venture-backed; an early pioneer of the AI-for-materials category.
New York, USA · founded 2024
Founders: Radical AI team
Building 'artificial general intelligence for science', including autonomous, self-driving laboratories for materials discovery.
Funding: About $65M raised.
San Francisco, USA · founded 2025
Founders: Edison Scientific team
Building an AI platform to enable discovery across the sciences, including materials.
Funding: About $70M raised.
Cambridge, United Kingdom · founded 2024
Founders: Dr Chad Edwards (CEO), Prof. Max Welling (CTO)
Building an AI system for materials discovery. Combines generative AI models, scientific data, computational simulation and lab validation to design materials with target properties for semiconductors, energy storage, climate tech and advanced manufacturing. Runs the 'AI Materials Foundry', a network of 45+ partners including NVIDIA and Meta.
Funding: Over $650M raised (a $100M Series A, then a $450M Series B at a $2.6B valuation). Investors include Kleiner Perkins, NEA, Temasek, NVIDIA's NVentures and Bezos Expeditions.
Paris, France · founded 2024
Founders: Team of ML researchers (alumni of Meta, Google, Applied Materials, Intel, Air Liquide, Amazon)
AI-driven engineering of materials at the atomic scale, for surface-driven industrial processes, using predictive and generative models, atomistic simulation and experimental feedback loops. Co-creator (with Hugging Face) of the open LeMaterial dataset initiative.
Funding: Seed-stage; focused on the energy transition and net-zero industrial processes.
Commercial platforms that sell materials informatics, AI, and simulation as a product or service, the tools companies actually buy to run materials R&D.
Redwood City, California, USA
One of the longest-running commercial materials informatics platforms. Applies AI to a company's own materials and chemicals data (via its DataManager and VirtualLab tools) to recommend new formulations and speed up product development.
United States
An AI platform for R&D, quality control and product-lifecycle data, used by advanced-materials and chemicals teams (clients include Clariant, Repsol and Carbon) to design and optimise formulations.
United States
A materials R&D cloud that runs quantum and atomistic simulations (Quantum ESPRESSO, VASP, LAMMPS) with high-throughput workflows and machine learning, all in the browser.
New York, USA
A public software company whose materials science division offers physics-based molecular simulation for designing materials from the molecular level up. Partnered with Ansys since 2023 for multiscale simulation.
United States
Granta MI is the widely used enterprise system for materials data management, keeping traceable, engineering-grade materials data for simulation and design. It integrates the Intellegens Alchemite machine-learning engine.
Cambridge, United Kingdom
A Cambridge machine-learning company. Its Alchemite platform learns from sparse, noisy experimental data to guide R&D, and is embedded inside the Ansys Granta MI materials data system.
France
BIOVIA Materials Studio is a long-established, multi-scale modelling and simulation environment for materials and chemistry, now offered on the 3DEXPERIENCE cloud platform.
Israel
A cloud platform for materials informatics that captures scattered R&D data and applies AI to guide experiments, from formulation through to production.
Companies that make and sell physical materials: nanomaterials, graphene and carbon nanotubes, quantum dots, and the instruments to grow and characterise them.
United States
A large manufacturer of nanoparticles, nanopowders and advanced engineered materials across the periodic table, supplying both research and industry.
United States
A major supplier of carbon nanotubes, graphene and graphene oxide, plus metal and oxide nanoparticles and ready-made dispersions.
United States
A producer of advanced nanomaterials, including graphene, carbon nanotubes and other 2D materials, alongside related technical services.
United States
A public engineered-nanomaterials maker (NASDAQ: NANX) supplying nanoscale coatings, surface finishing and personal-care ingredients such as mineral sunscreen actives.
Manchester, United Kingdom
A Manchester nanotechnology company specialising in cadmium-free quantum dots and tuned nanomaterials for sensing, imaging and electronics.
United Kingdom
A public high-technology firm making tools and systems to fabricate, analyse and manipulate matter at the atomic and molecular scale, including nanomaterial growth and characterisation (ALD, CVD, microscopy).
A few of the researchers whose work built this field. Not a ranking, just names worth knowing.
Director of the Materials Project, at UC Berkeley and Lawrence Berkeley National Laboratory. A leader of open materials data infrastructure.
UC Berkeley and Berkeley Lab; co-founder of the Materials Project and a pioneer of computational battery-materials design and high-throughput methods.
Northwestern University; creator of the Open Quantum Materials Database (OQMD) and a leader in high-throughput DFT and ML for materials.
Duke University; founder of the AFLOW high-throughput materials framework and database.
Lawrence Berkeley National Laboratory; a leader in materials informatics and a driving force behind pymatgen, matminer and automated workflows.
University of Toronto; pioneer of AI for chemistry and materials, and of self-driving (autonomous) laboratories.
Co-founder and CTO of CuspAI, and a foundational researcher in equivariant deep learning, the geometry-respecting architectures behind modern materials models.
Microsoft Research; author of CGCNN (crystal graph networks) and a lead on MatterGen, a generative model for inorganic materials.
Missing something, or spot an error? This directory grows over time. Tell me on LinkedInor via the contact page. For the ideas behind these tools, read the blog.