US awards SandboxAQ $500M CHIPS grant for AI-driven semiconductor materials discovery

The US Department of Commerce's CHIPS Research and Development Office signed a definitive agreement on June 17 awarding $500 million to SandboxAQ, an Alphabet spinoff, to develop critical semiconductor materials using physics-based AI. The funding targets four strategic areas: PFAS-free process chemicals (replacing 'forever chemicals' used in lithography), catalysts for ultra-pure gas generation, rare earth-free magnets (China controls 90%+ of global neodymium production), and battery chemistries that avoid lithium and cobalt imports. As part of the deal, the Commerce Department receives a minority, non-voting equity stake and future royalties.

SandboxAQ's ReAQT simulation platform and Large Quantitative Models (LQMs)—trained on physics and chemistry rather than human text—will screen millions of candidate materials virtually before laboratory validation. The company claims its AQCat catalyst-screening workflows, built on 13.5 million calculations with NVIDIA, achieve ~20,000x speedup over conventional lab methods, compressing discovery timelines from decades to weeks. This represents one of the largest federal investments to date in physics-based AI for scientific discovery.

For AI architects tracking supply-chain localization: this signals a shift beyond CHIPS funding for fab capacity (Intel, TSMC, Samsung) toward securing upstream materials dependencies. The equity stake and royalty structure mirrors venture investing, suggesting Commerce views semiconductor supply resilience as a multi-decade infrastructure problem requiring aligned incentives—not just grants. PFAS alternatives are deployable at existing fabs; rare-earth magnets and battery tech face longer qualification timelines.