Materials science
5 labs · 11 models and products Theme: Companies · Generated from this page's data. Click a name to open its panel on the Labs page. Records resting on one source are tagged "one source".Materials science AI predicts how arrangements of atoms will behave, so that new materials for chips, energy, cooling and climate uses can be screened on a computer before anyone makes them. Google DeepMind's GNoME predicts which new crystals are stable, and DeepMind plans its first UK automated materials lab, using robots and Gemini, in 2026. Orbital Industries builds Orb, an open model that simulates atoms; CuspAI designs materials for semiconductors, energy and climate; Materials Nexus combines machine learning and quantum mechanics to predict how new compositions perform (per the site).
Context
- The AI for Science Strategy says there is no AlphaFold-style breakthrough yet in materials science. 20 November 2025 gov.uk/government/publications/ai-for-science-st
We don't have AlphaFold equivalents in materials science, fluid dynamics or neuroscience
- The strategy cites Liverpool's Materials Innovation Factory, an £81 million facility co-founded with Unilever. 20 November 2025 gov.uk/government/publications/ai-for-science-st
The Materials Innovation Factory (MIF) is an £81 million facility co-founded by the University of Liverpool and Unilever.
- Google DeepMind's GNoME, published in Nature, expanded the number of known stable materials roughly tenfold, its authors report. 2023 (Nature vol. 624) nature.com/articles/s41586-023-06735-9
Our work represents an order-of-magnitude expansion in stable materials known to humanity.
- Google DeepMind said it would set up its first automated laboratory in the UK in 2026, focused on materials science. 10 December 2025 deepmind.google/blog/strengthening-our-partnersh
establish Google DeepMind's first automated laboratory in the UK in 2026
- The government's one-year-on report names Cambridge's MACE, a materials foundation model, under its home-grown AI section. 29 January 2026 gov.uk/government/publications/ai-opportunities-
the University of Cambridge's MACE (a materials foundation model for atomistic simulations and discovery)
- Orbital Materials released Orb, a family of AI models that simulate how atoms in materials behave. 29 October 2024 arxiv.org/abs/2410.22570
We introduce Orb, a family of universal interatomic potentials for atomistic modelling of materials.
Key papers
- Scaling deep learning for materials discovery (GNoME) Nature · 2023
- Orb: A Fast, Scalable Neural Network Potential arXiv preprint · 29 October 2024
- Orb-v3: atomistic simulation at scale arXiv preprint · 8 April 2025
Labs Round 8 (31 Oct 2026): introduction and facts as the researchers wrote them; "(per the site)" marks a sentence resting on this site's own description of that lab.
| Record | What it is | Place | Links | Confidence |
|---|---|---|---|---|
| AI Materials Foundry (CuspAI)Chips, models and products | CuspAI's network of industry, lab and data partners that uses its AI to design and test new materials. live (partner network)Key facts Stated as a global network of over 45 partners, including NVIDIA and Meta as core partners; industry list includes 3M, AMD, Applied Materials and Samsung; five-step approach: define, generate, simulate, synthesise, scale; access: partnershipMade by CuspAI · — (launch date not stated) | Cambridge | cusp.ai/ | one source |
| Alchemite (Intellegens)Chips, models and products | Intellegens' machine-learning product suite, used to design experiments and formulations for materials, chemicals and life sciences. liveKey facts Product lines: Alchemite Suite, for DOE, for formulations, for R&D insights, for oligonucleotides; uses uncertainty quantification; MCP server added 2026-07-02Made by Intellegens · 2026-07-02 | Cambridge (via Intellegens) | intellegens.com/ · intellegens.com/new-alchemite-mcp-feature-delive | lab's own pages |
| CurieOSChips, models and products | Orbital's software for AI-driven materials discovery, used in chemistry, hardware and engineering. released (deployed by BASF ECMS; date not published)Key facts Research page (read 30 Oct 2026) cites a press item: "BASF ECMS deploys CurieOS for AI-driven materials discovery." No date shown.Made by Orbital Industries (Orbital Materials Ltd) | London (WC1N 2EB) (via Orbital Industries (Orbital Materials Ltd)) | orbitalindustries.com/research | one source |
| CuspAICompanies | Builds AI systems that search for and design new materials for semiconductors, energy and climate uses.Profile
Labs Round 7 deep profile (30 Oct 2026), as the researchers wrote it; confidence: single_source (rounds trade-only; total from company site). Sources Models and products (2)
Key facts and news (7)
Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked. | 20 Station Road, Cambridge (CB1 2JD) | Companies House · tech.eu/2026/07/20/jeff-bezos-and-sovereign-ai-b · cusp.ai/ · find-and-update.company-information.service.gov. · sovereignai.gov.uk/post/sovereign-ai-invests-in- | one source |
| FermiNet (Fermionic Neural Network)Chips, models and products | Neural network that calculates the quantum behaviour of electrons in molecules from first principles. released (open code; papers)Key facts First published 19 Oct 2020; update 22 Aug 2024 covers a Science paper on excited states; code from the original study and the excited-states work open-sourced on GitHub; access: open codeMade by Google DeepMind · 2020-10-19 (first post; updated 22 Aug 2024 for excited-states Science paper) | London | deepmind.google/blog/ferminet-quantum-physics-an · deepmind.google/blog/page/8/ | lab's own pages |
| GNoME (Graph Networks for Materials Exploration)Chips, models and products | A deep-learning tool that predicts which new crystal materials are stable. released (paper and data)Key facts Theme: Materials science. Nature paper 29 Nov 2023. Figures (29 Nov 2023): 2.2 million new crystals; 380,000 stable materials contributed to the Materials Project; 736 made independently by outside labs.Made by Google DeepMind · 2023-11-29Read the paper | London (EC2M 4AG) (via Google DeepMind) | deepmind.google/blog/millions-of-new-materials-d · deepmind.google/blog/page/9/ | one source |
| Google DeepMind UK automated materials science labData centres and compute sites | Google DeepMind's first automated lab in the UK, announced for 2026, using robots and Gemini to make and test materials. announced (opening not verified)Key facts Planned 2026; robots to "synthesize and characterize hundreds of materials per day"; fully integrated with Gemini (announced 10 Dec 2025). Location in the UK not published on the page read | deepmind.google/blog/strengthening-our-partnersh · uktech.news/ai/google-deepmind-to-open-automated | one source | |
| IntellegensCompanies | Cambridge spin-out whose Alchemite deep-learning software predicts materials and chemicals from sparse, noisy lab data.Key facts Co-founded 2017; technology originated at the Cavendish Laboratory (Gareth Conduit); 2023 Cambridge Independent "AI Company of the Year"; 2026-08-11 selected from over 600 applications for an Innovate UK Frontier AI grant ("Adaptive World Models for Experimental Design"); 2026-07-02 Alchemite MCP feature links Alchemite to LLMs and agents; named users include Johnson Matthey, voestalpine, NASA, Boeing, Genentech (with Optibrium)Models and products (1)
| Cambridge | Companies House · intellegens.com/ · intellegens.com/intellegens-wins-frontier-ai-gra | verified |
| LINUS (Orbital foundation model)Chips, models and products | Orbital's in-house foundation model for materials, on which Orb is built. internal (date not published)Key facts Named in the Orb press release (3 Sep 2024); no separate release date published.Made by Orbital Industries (Orbital Materials Ltd) | London (WC1N 2EB) (via Orbital Industries (Orbital Materials Ltd)) | orbitalindustries.com/posts/press-release-orb | one source |
| MatNex (Materials Nexus)Companies · also on Start-ups | Machine-learning and quantum-mechanics software that predicts how new material compositions will perform, to cut lab experiments.Profile
Labs Round 8 deep profile (31 Oct 2026), as the researchers wrote it; confidence: verified (2023 seed: trade plus company news list); trading name and UKAEA contract one source. Sources Key facts and news (3)
Labs Round 8 (31 Oct 2026), from the researchers' rows; one-source items are marked. | Cambridge (CB1 2LA) | Companies House · uktech.news/ai/materials-nexus-2m-funding-202307 · find-and-update.company-information.service.gov. · find-and-update.company-information.service.gov. · assets.publishing.service.gov.uk/media/64bf8d179 | verified |
| MIRA (CuspAI)Chips, models and products | CuspAI's own AI platform that runs the whole cycle of designing, simulating and testing new materials. liveKey facts At the centre of the AI Materials Foundry launched 2026-07-20; covers generative materials design, simulation, synthesis route planning and experimental validationMade by CuspAI · 2026-07-20 | 20 Station Road, Cambridge (CB1 2JD) (via CuspAI) | uktech.news/ai/cuspai-raises-334m-and-launches-a | one source |
| Orb (universal interatomic potential)Chips, models and products | Orbital's open-source AI model that simulates how atoms in advanced materials behave, faster than older methods. released (open source)Key facts Technical introduction 9 Mar 2024; open-sourced 2 Sep 2024 (code at github.com/orbital-materials/orb-models; permissive licence free for non-commercial use and start-ups); press release 3 Sep 2024 says it is built on Orbital's LINUS foundation model. Figures (Sep 2024, as claimed): "5x faster for large-scale simulations"; more accurate than comparable Google and Microsoft models. Version history: Orb (2024) → OrbMol (27 Aug 2025) → Orb-v3 (4 Sep 2025) → OrbMol-v2 (2 Jun 2026).Made by Orbital Industries (Orbital Materials Ltd) · 2024-09-02Versions Orb (2024) → OrbMol (27 Aug 2025) → Orb-v3 (4 Sep 2025) → OrbMol-v2 (2 Jun 2026)Read the paper | London (WC1N 2EB) (via Orbital Industries (Orbital Materials Ltd)) | orbitalindustries.com/posts/introducing-orb · orbitalindustries.com/posts/press-release-orb | lab's own pages |
| Orb-v3Chips, models and products | A family of Orb models that simulate materials atom by atom at large scale, trading precision against computing cost. released (open source, Apache 2.0)Key facts Released 4 Sep 2025; code and weights on GitHub under Apache 2.0; launch partner Rowan. Figures (4 Sep 2025): tens or hundreds of thousands of atoms; "10x faster and 5x more memory efficient than MACE-MPA-0"; "40x faster and 40x more memory efficient than Sevennet".Made by Orbital Industries (Orbital Materials Ltd) · 2025-09-04Read the paper | London (WC1N 2EB) (via Orbital Industries (Orbital Materials Ltd)) | orbitalindustries.com/news/blog/orb-v3-atomistic · orbitalindustries.com/research | lab's own pages |
| Orbital Industries (Orbital Materials Ltd)Companies | London company that uses its own atom-level AI simulation engine, Orb, to design materials and build cooling and other hardware for AI data centres. renamedProfile
Labs Round 7 deep profile (30 Oct 2026), as the researchers wrote it; confidence: single_source (two trade sources; company site could not be opened); sources are listed in the record panel. Models and products (6)Version chainsOrb (2024) → OrbMol (27 Aug 2025) → Orb-v3 (4 Sep 2025) → OrbMol-v2 (2 Jun 2026)
Selected papers (2)
| London (WC1N 2EB) | Companies House · tech.eu/2026/05/28/orbital-industries-secures-50 · find-and-update.company-information.service.gov. · find-and-update.company-information.service.gov. · find-and-update.company-information.service.gov. | verified |
| OrbMolChips, models and products | An Orb model extended to molecules, including metal compounds, biological molecules and battery electrolytes. released (open source, Apache 2.0)Key facts Announced 27 Aug 2025; Apache 2.0. Figures (27 Aug 2025): trained on over 100M high-accuracy DFT calculations from OMol25; "at least 2x faster" than comparable models; simulated an enzyme of over 20,000 atoms. Successor OrbMol-v2 (2 Jun 2026).Made by Orbital Industries (Orbital Materials Ltd) · 2025-08-27 | London (WC1N 2EB) (via Orbital Industries (Orbital Materials Ltd)) | orbitalindustries.com/posts/orbmol-extending-orb · orbitalindustries.com/research | lab's own pages |
| OrbMol-v2Chips, models and products | Orbital's second molecular model, which it says outperforms the most-cited method in chemistry. released (blog)Key facts Blog dated 2 Jun 2026 (research index). Post not opened; figures not recorded.Made by Orbital Industries (Orbital Materials Ltd) · 2026-06-02 | London (WC1N 2EB) (via Orbital Industries (Orbital Materials Ltd)) | orbitalindustries.com/research | one source |