Data - AI Chemicals - Materials

AI and Materials: How Artificial Intelligence Is Turning Data into Revolutionary Smart Materials

Published on 11 September 2026 Read 25 min

ai materials experts consulting consutlancy agency

In the last years, Artificial Intelligence (AI) has established itself as an indispensable tool that is revolutionizing many different domains of human knowledge, from medicine to psychology or engineering. However, the material science is a domain that is not as commonly associated with AI as those mentioned above, but over which it can bring changes that are both profound and far-reaching.

Conceived as the science that focuses on understanding and manipulating the properties and applications of materials to develop innovative solutions, this field is at a pivotal juncture. While traditional methods based on manual trial-and-error methodologies are effective, they are often time-consuming and resource-intensive. These issues can be tackled by adopting a more efficient AI-based strategy during the R&D process, representing not only a merely enhancement on the field, but a paradigm shift. This transformative capability is reshaping how materials’ R&D is approached, offering unprecedented speed and precision in the development of innovative materials. More recently, the emergence of generative models for materials design has propelled AI beyond the role of a tool, into an autonomous pillar of materials R&D.

In this article, Alcimed explores the dynamic intersection of AI and materials science through the examples of enterprises and successful projects.

Improving the materials discovery process through artificial intelligence

The materials discovery process is a comprehensive approach used to identify and develop new materials with desirable properties for a wide range of applications. This process, which can be broken down into several stages, has historically faced certain limitations due to the reliance on traditional methods based on manual trials that slowed down the discovery cycle.

With the advent of AI, however, the entire cycle is being accelerated and enhanced at every stage, driving an unprecedented speed-up of the discovery process. Since 2024, AI’s role in improving this process has reached a new milestone with the emergence of generative models for materials design. Far from merely supporting research, AI has become a driver of design itself, capable of proposing, evaluating and optimizing new materials autonomously.

AI for identifying future trends

Using algorithms, natural language processing (NLP), text mining and techniques such as latent Dirichlet allocation (LDA), AI tools can analyze market trends, competitor activity, customer feedback, economic indicators and more, in order to identify emerging requirements, future trends and new needs-based questions.

AI for literature review and foundational research

Using NLP or machine learning algorithms, AI-powered tools such as ChemDataExtractor, tmChem and IBM DeepSearch can work with extremely large databases fed by historical scientific knowledge (patents, articles, reports, etc.) that are continuously updated.

These tools can ingest and analyze all this data to extract and identify relevant information, key concepts, patterns, correlations and more, and even summarize the results to provide an overview of existing knowledge. To date, this stage has been one of the most rewarding, and two distinct approaches can be identified:

Open source: AI-powered tools search exclusively for information within the existing public literature (e.g. Citrine Informatics with HRL Laboratories).
Open source + internal information: AI-powered tools are supplemented with internal data provided by the end client (e.g. Chemintelligence).

AI for hypothesis formulation

Based on the previously analyzed existing data and on theoretical frameworks, AI-driven analysis can identify plausible relationships between materials, properties and other variables, and propose hypotheses that expand the discovery space.

AI for experimenting with new materials

Prediction: AI-trained models, such as neural networks and regression algorithms, can predict the properties and behavior of new materials based on several variables such as their composition or structure. AI can then rapidly screen vast libraries of potential material compositions, predicting their properties and identifying promising candidates for further study.

Simulation: Running AI-driven materials simulations at various scales (atomic, molecular, macro) can challenge previously established predictions by modeling their behavior under a range of conditions. AI can then retrieve and analyze the simulation data and, using generative models such as GANs (Generative Adversarial Networks) or VAEs (Variational Autoencoders), propose new material structures that are likely to exhibit the desired properties. This creates a closed loop that can be fed until a desired number of solutions meeting the established requirements are predicted.

Testing: Once the predicted solution matches the sought-after characteristics, AI can carry out virtual experiments to test the material’s properties under various simulated conditions. This makes it possible to optimize the design of real-world physical experiments by selecting the most informative and efficient set of evaluations to run, thereby maximizing the insights obtained.

In recent years, the use of AI has ramped up in the modeling and testing stages thanks to the integration of digital twins, digital models that virtually represent a physical object or system, , originally used in industrial settings to simulate equipment.

Nevertheless, strong predictive performance is not enough to produce scientific knowledge. Indeed, a model may accurately predict a property without correctly representing the physical or chemical mechanism behind it.

AI for analyzing performance data on new materials

AI-powered tools enable data preprocessing and cleaning. These tools can perform various types of analysis (statistical, multivariate, real-time, etc.) and can even be integrated into simulation and experimentation models, providing immediate feedback and insights that make it possible to optimize experimental conditions dynamically.

AI for creating customized results reports

Using NLP, AI can generate customized reports that summarize large volumes of experimental data, include advanced interactive visualizations (3D plots, heat maps, etc.), and even incorporate information from textual data such as lab notes or research papers.

Finally, since AI’s potential can be maximized with the knowledge of a human domain expert who defines certain necessary characteristics and fine-tunes specific parameters, researchers remain at the heart of the innovation process. The presence of the expert increases the likelihood that AI-powered generative models will propose a range of possible materials to be evaluated and synthesized in the laboratory.

AI for autonomous laboratories

Thanks to the integration of artificial intelligence and robotics, autonomous laboratories, or “self-driving labs”, represent a major operational breakthrough. These platforms automate the entire research and experimentation process, delivering significant gains in the speed at which high-performance materials can be identified. In time, this innovation could shorten materials innovation cycles from 10–20 years to just a few years¹. Nevertheless, numerous obstacles are slowing the development of self-driving labs, particularly economic ones: the infrastructure is complex and costly.

AI for materials design

Initially, AI applications operated in a screening mode: evaluating millions of existing candidates to identify the most relevant ones. Today, generative models such as MatterGen and AtomGPT mark a conceptual break. Indeed, AI now directly proposes molecular or crystal structures optimized for a set of target properties. This paradigm shift makes it possible to extend the scope of R&D beyond materials that are already known and documented. Graph Networks for Materials Exploration (GNoME), a machine learning model developed by Google DeepMind, identified 2.2 million new crystal structures, 380,000 of which are stable. These results are equivalent to 800 years of experimental discoveries, demonstrating the power of applying AI to materials science².

Find out how our team can meet your needs for artificial intelligence consulting >

3 examples of initiatives using AI to discover new materials

Initiative #1: A-Lab, an innovative laboratory powered by Berkeley and focused on the synthesis of new materials

Berkeley’s Materials Science and Engineering department has developed the A-Lab, an autonomous laboratory that harnesses artificial intelligence (AI) to dramatically accelerate the synthesis of new materials and bridge the gap between computational screening rates and the experimental realization of new materials.

Designed to speed up the synthesis of new inorganic materials, the A-Lab synthesized 41 new compounds from a set of 58 targets in just 17 days³. Using data from the Materials Project and Google DeepMind, along with natural language models trained on historical literature, the autonomous lab generated and optimized synthesis recipes through thermodynamics-based active learning.

The A-Lab demonstrated the transformative impact of AI in materials science, enabling faster discovery, continuous optimization, and high-throughput scalability. The integration of AI and robotics made possible autonomous, efficient, and precise synthesis processes, rendering experiments reproducible with minimal human intervention.

Initiative #2: Altrove, a French start-up aiming to create alternatives to materials containing rare earths

Altrove is a French start-up that relies on artificial intelligence (AI) to accelerate the discovery and creation of new materials essential to technologies such as electric vehicles and advanced electronics. The company focuses in particular on creating alternatives to materials containing rare earths, given that rare-earth elements are currently surrounded by numerous geopolitical and economic challenges.

The company tackles the complex challenges of materials science not only by predicting potential new materials, but also by devising precise recipes for their synthesis. Its approach involves using AI models to predict stable inorganic materials, followed by automated laboratory processes to synthesize, test, and optimize these materials.

So far, Altrove has forged around twenty partnerships with players in aerospace, automotive, robotics, electronics, and defense. Following a co-development phase, the start-up offers partners temporary exclusivity over the critical material in their market, while retaining the intellectual property on both the material and the recipe.

In 2 years, the company has raised €15 million, which it will use to move from proof of concept to industrial production. Altrove indeed plans to bring its first products to market within 2 to 3 years, in the form of a catalog of materials aimed at industrial users⁴.

Initiative #3: Citrine Informatics, an American company that uses machine learning algorithms to help scientists discover materials

Citrine Informatics is an American company specializing in the materials industry that applies artificial intelligence and data-driven methods to accelerate materials discovery and development. It uses advanced machine learning algorithms to analyze vast datasets on the properties, composition, and performance of materials. Through its four different products, it can create an AI-powered ecosystem that enables scientists and engineers to make data-driven decisions more effectively and efficiently.

Citrine’s AI-driven platforms have already been successfully implemented by several entities in the materials industry to optimize their R&D efforts.

Examples of case studies:

New materials: HRL Laboratories partnered with Citrine Informatics to accelerate the development of a 3D-printable, aerospace-grade alloy. HRL Laboratories wanted to find nanoparticles that would nucleate a microstructure less prone to hot cracking. Given the specific properties sought, the Citrine AI software drew on classical nucleation theory, lattice-spacing rules, thermodynamic stability, and materials informatics to efficiently explore 11.5 million powder and nanoparticle combinations, from which 100 promising candidates were identified. Ultimately, the resulting material, AL 7A77, was commercialized within two years of the project’s start—a considerable reduction in time compared with conventional R&D procedures.
Enhanced properties: A global leader in specialty chemicals and plastics turned to Citrine Informatics to improve its ability to respond dynamically to a customer’s specific requirements. The challenge was to increase the mechanical properties of a glass-fiber-reinforced polymer while preserving its overall property profile. The Citrine platform was updated with the customer’s test data and recipe information from its portfolio. The AI models were retrained through a process called sequential learning. In the end, out of trillions of potential candidates, 10 experimental candidates were proposed, improving the previous material’s performance by an average of 21%.

In conclusion, the integration of artificial intelligence (AI) into materials science is revolutionizing the R&D landscape for new materials. AI-based methodologies facilitate the rapid analysis of vast datasets, the predictive modeling of material properties, and the optimization of experimental protocols, thereby accelerating the innovation cycle.

However, despite these advances, the path to fully harnessing AI’s potential in R&D is still under way. Challenges remain, particularly in terms of data integration, algorithm development, and translating AI-derived knowledge into practical applications, among others. If you have an AI-related project and would like to discuss it with our team, don’t hesitate to contact us!

About the author,

Vincent, Director of Alcimed’s Chemistry and Materials team in France.

Juliette, Senior Consultant within Alcimed’s Energy-Environment-Mobility team in France.