SandboxAQ's AQCat AI Model Accelerates Catalyst Screening 20,000x with Natural Language Interface — SandboxAQ

Event summary

SandboxAQ made its AQCat AI model generally available on Claude Science on August 19, 2026, enabling natural language-based catalyst screening.
AQCat performs adsorption energy calculations up to 20,000 times faster than traditional methods while maintaining near-DFT accuracy.
The model is 'spin-aware,' accounting for magnetic behavior in earth-abundant metals like iron, cobalt, and nickel.
AQCat was trained on AQCat25, a dataset of 13.5 million high-fidelity DFT calculations across 47,000 intermediate-catalyst systems.

The big picture

SandboxAQ's AQCat addresses a critical bottleneck in catalyst research, where traditional methods limit progress to incremental gains. By democratizing access to high-throughput screening through natural language interfaces, the company is positioning itself as a key enabler in industries reliant on chemical manufacturing. The strategic move aligns with broader trends in AI-driven materials science and quantum computing applications, potentially reshaping R&D timelines and cost structures across multiple sectors.

What we're watching

Adoption Pace: How quickly research teams across industries will integrate AQCat into their workflows, particularly in fields like green hydrogen and sustainable aviation fuel.
Competitive Response: Whether existing catalyst screening providers will develop comparable AI solutions or form partnerships to keep pace with SandboxAQ's technological advantage.
Commercialization Impact: The extent to which AQCat's speed and accuracy will reduce R&D costs and accelerate the development of new catalysts for industrial applications.