Altrove Demonstrates an End-to-End, High-Throughput Workflow for Next-Generation Inorganic Materials using Synopsys QuantumATK

Collaboration delivers a scalable pipeline from 200,000+ computational candidates to experimental validation

Paris, France — Altrove today announced the results of a year-long collaboration with Synopsys that showcases one of the first fully integrated, industrial-scale pipelines for discovering next‑generation inorganic materials. The joint workflow unites AI-driven candidate selection, high-throughput first‑principles screening via Synopsys’ QuantumATK®, recipe inference, and automated experimental validation. Synopsys QuantumATK™ is a comprehensive, high‑performance computational platform for atomistic simulations, supporting scalable first‑principles and force‑field modeling of electronic, magnetic, thermal, optical, mechanical, and other material properties.

The collaboration addresses growing global challenges around critical inorganic materials, which face increasing pressure due to supply‑chain concentration, regulatory constraints, sustainability demands, and sovereignty considerations. Traditional materials R&D, which relies on slow and manual iteration, cannot navigate the vast chemical and structural design spaces required for innovation. By combining AI, scalable density-functional theory (DFT), and robotized experimentation, Synopsys and altrove demonstrate that discovery cycles historically measured in years can now be reduced to mere months.

From AI Hypotheses to Experimental Reality

Reinventing critical inorganic materials—such as magnetic, piezoelectric, or thermoelectric compounds—is particularly complex because their properties depend on intricate crystal structures, subtle dopant chemistries, and narrow performance tolerances. The relevant design space spans tens of millions of potential compositions, dopants, and microstructures. To make early-stage exploration tractable, altrove employs a combination of AI-driven candidate selection, solid-state recipe inference, and a robotized experimental platform capable of validating materials efficiently and at scale.

At the same time, many key material properties, including magnetic anisotropy, are absent from public datasets and cannot be reliably predicted by existing foundational AI models. High-fidelity first‑principles calculations are therefore essential. Synopsys’ QuantumATK provides a scriptable, HPC-ready platform that enables high-throughput DFT screening of complex inorganic and magnetic systems. This environment allows systematic filtering based on stability, magnetism, electronic structure, and ground‑state verification, forming a physics‑grounded backbone for a modern industrial materials discovery stack.

An Integrated, Four-Stage Workflow

Over the course of the collaboration, Synopsys and altrove implemented a fully reproducible, end-to-end workflow. In the first stage, altrove used foundational models, AI methods, and domain-specific heuristics to evaluate more than 200,000 potential materials, identifying approximately 30,000 with promising characteristics. In the second stage, QuantumATK performed a large-scale physics screening process designed to evaluate magnetic and structural properties with increasing precision. This involved approximately 30,000 magnetocrystalline anisotropy calculations for coarse filtering, followed by around 1,000 calculations assessing Heisenberg exchange interactions and magnetic ordering tendencies, and ultimately roughly 200 high‑accuracy ground‑state calculations to confirm stability and target performance. This sequence enabled a roughly 99% reduction in candidate materials, distilling AI-generated hypotheses into a small set of physically viable structures.

In the third stage, altrove applied its models to infer solid‑state synthesis routes for the shortlisted materials, including dopant strategies and thermal processing schedules. In the final stage, altrove synthesized and experimentally validated X of these candidates using its robotized laboratory platform, further optimizing them for magnetic performance. The result is a closed-loop workflow that seamlessly connects hypotheses, physics evaluation, recipe inference, and experimental validation—designed explicitly for industrial relevance rather than isolated demonstration.

Beyond Structure Prediction

The outcomes of the collaboration extend significantly beyond conventional structure‑prediction workflows. The work demonstrates that high‑throughput DFT for complex and magnetic inorganic systems is now practical at near‑industrial scale, and that critical magnetic properties must be generated systematically rather than assumed. It also shows that physics-based screening can dramatically reduce experimental burden and cost, and that the resulting DFT data can be reused to train machine-learning models capable of accelerating pre‑screening of magnetic properties beyond the systems explicitly computed. Altrove is actively developing such models based on the dataset produced through this effort. More broadly, the work confirms that large‑scale computational screening can robustly identify promising candidates for important functional properties.

Impact for Science and Industry

Together, Synopsys and altrove provide a template for modernizing materials R&D. The workflow illustrates how organizations can compress multi‑year discovery programs into months, build traceable and physics‑grounded pipelines suited for regulated and strategic industries, and develop reliable, sovereign alternatives to constrained critical materials.

About altrove

Altrove is a deep‑tech company developing AI-driven and automated platforms for discovering next-generation inorganic materials, with an emphasis on alternatives to critical and supply-constrained compounds. By integrating physics, machine learning, and robotized experimentation, altrove enables faster and more reliable materials innovation for industrial applications.