SandboxAQ’s peer-reviewed AQCat25 study, published in npj Computational Materials, brings the magnetic behavior of catalysts into large-scale AI modeling and is now generally available to researchers.
Palo Alto, California, June 16, 2026 /PRNewswire/ — SandboxAQ has published peer-reviewed research in the Nature Portfolio journal npj Computational Materials, introducing a machine learning approach that allows industry to simulate, screen, and optimize catalysts with physics-based precision. This research addresses a long-standing gap in computational chemistry: dealing with the magnetism of earth-abundant metals that drive industrial catalysis.
At the heart of this research is AQCat25, a dataset of 13.5 million density functional theory calculations across 47,000 catalyst systems. This is the first large-scale catalyst dataset to incorporate spin polarization and was generated using approximately 400,000 GPU hours on the NVIDIA DGX cloud.
“We’re proud of our team for publishing a new paper in the peer-reviewed journal Nature Computational Materials,” SandboxAQ CEO Jack Hidary wrote on LinkedIn. “The SandboxAQ team brings breakthrough advances in catalyst discovery and computational chemistry.”
Catalysts power more than 80% of manufactured goods such as fertilizers, fuels, and chemicals. Many rely on iron, cobalt, and nickel, whose magnetic behavior greatly influences how molecules bind to surfaces. These effects are expensive to compute and were often omitted in early datasets, limiting their accuracy for industrially important materials. AQCat25 fills that gap. Researchers can explore material discoveries and request access at sandboxaq.com.
“Catalysis powers the global economy, from the fuels that power our world to the materials that shape it,” Hidary wrote. “Using our AQCat model, industry can now simulate, screen and optimize catalysts with physics-based precision, enabling performance and sustainability breakthroughs at an unprecedented scale.”
Main highlights:
- 13.5 million DFT calculations across 47,000 catalyst systems
- Spin polarization of 12 magnetic elements and 6 new elements (barium, cerium, fluorine, lithium, lanthanum, and magnesium)
- SandboxAQ says it’s up to 20,000 times faster than first-principles simulations, making high-throughput virtual screening practical
- Datasets and models published on Hugging Face under a Creative Commons license
About SandboxAQ
SandboxAQ provides solutions at the intersection of AI and quantum technologies. The company’s Large-Scale Quantitative Models (LQM) apply physics-based AI to deliver advances in life sciences, chemicals and materials, financial services, navigation, and cybersecurity. For more information, please visit sandboxaq.com.
The full text of the peer-reviewed paper is available at npj Computational Materials.
Source Sandbox AQ
