SandboxAQ AQChemSim – Quantitative AI for Materials

SandboxAQ is a science-first technology company, spun out of Alphabet in 2022 according to its materials page, with press releases datelined Palo Alto, California. It applies Large Quantitative Models (LQMs) that combine AI with physics-based simulation to materials and chemistry problems such as catalysts, battery materials, PFAS alternatives and semiconductor substrates.

Models include AQCat for heterogeneous catalysis and AQVolt for battery materials, offered as LQM-enabled LLM tools through the Model Context Protocol, as enterprise licenses deployed in a customer's own environment with fine-tuning on proprietary data, and as milestone-based Frontier Partnerships. Partners named include NOVONIX, AWS, Intel, Accenture, DIC and NVIDIA.

• AQCat Adsorption Spin: predicts spin-polarized adsorption energies and geometries on catalytic surfaces.
• AQCat25-EV2: model family for heterogeneous catalysis.
• AQVolt: AI model for battery materials.
• MCP integration: LQMs exposed to LLM clients such as Claude.
• Enterprise licensing: deployment in the customer environment with fine-tuning.
• Frontier Partnerships: long-term, shared-risk, milestone-based engagements.

A Vietnamese battery or energy-storage developer could screen electrode and electrolyte candidates. A chemical or catalyst producer could screen catalysts for adsorption behavior before lab tests. A semiconductor materials supplier could study PFAS-free process chemicals through a licensed model.

Near-DFT accuracy at much higher speed for high-throughput screening. Site reports a 95% reduction in battery testing time with NOVONIX and an 80x speedup in complex chemistry simulations with NVIDIA. Enterprise licenses allow deployment in the customer's own environment with fine-tuning on proprietary data. Models connect to MCP-compatible clients.

Materials supply chains for semiconductors, batteries and PFAS replacement are attracting large public and private funding, including the US CHIPS R&D award to SandboxAQ. AI simulation models are replacing slow quantum chemistry in screening. Vietnam's semiconductor, battery and chemical sectors may engage through licensing or joint research.

Press release datelined Palo Alto, California on 2026-06-17 announced a definitive agreement with the US CHIPS R&D Office for a 500M USD award, with the Department of Commerce taking a minority non-voting equity stake. Customers and partners named on the page include NOVONIX, AWS, Intel, Accenture, DIC and NVIDIA.