SandboxAQ lands $500M CHIPS Act R&D award for AI-accelerated materials discovery

The U.S. Commerce Department's CHIPS Research & Development Office signed a definitive agreement with SandboxAQ for a $500 million award to develop AI-aided materials discovery for semiconductor manufacturing. The Palo Alto-based NVIDIA-backed company, valued at $5.75 billion in April 2025, will deploy its proprietary ReAQT simulation platform and Large Quantitative Models to accelerate discovery of alternatives to PFAS forever chemicals, catalysts, rare earth-free magnets, and backup power battery chemistries—critical but less visible fab supply-chain components.

SandboxAQ uses physics-first methodology: Large Quantitative Models are trained directly on fundamental laws of physics, chemistry, and biology, not human text. The platform generates training sets via high-fidelity quantum chemistry simulations (Density Functional Theory and Molecular Dynamics), establishing reliable predictive maps of molecular behavior before committing to physical laboratory synthesis. This enables automated Design-Make-Test-Learn loops that screen millions of untested chemical candidates, compressing discovery timelines from decades into weeks—a critical acceleration for chipmakers solving supply-chain bottlenecks.

The Commerce Department will receive a minority equity stake in SandboxAQ and royalties if successful formulas are licensed to industrial partners—an unusual return mechanism for CHIPS Act awards that signals confidence in the commercial potential of materials research. CEO Jack Hidary emphasized opportunities to choose different chemicals and, where PFAS cannot be avoided, break it down on site before leaving the fab.

For manufacturing and supply-chain teams, this targets one of the least visible but most critical bottlenecks: fab materials. While chip architectures and power delivery dominate headlines, the catalysts, rare-earth magnets, and PFAS replacements that enable production are often in shorter supply than silicon itself. AI-accelerated materials discovery could unlock alternatives to constrained inputs before physical capacity limits hit the industry's real constraint.