Large Quantitative Models — AI grounded in physics, chemistry, and biology for defense, biotech, and materials
HQ: Palo Alto, CA, United States | Founded: 2022 | Employees: ~200 (est.) | Stage: Late Stage VC ($5.75B valuation) | Website: https://www.sandboxaq.com
SandboxAQ started inside Alphabet around 2016 as 'Sandbox,' an AI-and-quantum moonshot, and spun out as an independent company in March 2022 with Eric Schmidt as chairman and Jack Hidary as CEO. Its thesis: the next wave of AI value lies not in language models trained on internet text but in Large Quantitative Models — neural systems trained on equations, simulations, and experimental data from physics, chemistry, and biology.
The product portfolio makes the thesis concrete. Synthesis generates AI-based synthetic data and manages cryptographic inventory for post-quantum migration — it is used by the U.S. Navy. AQNav pairs quantum magnetometers with AI to read Earth's magnetic crust for GPS-denied navigation, and has been trialed with the U.S. Air Force and aerospace partners including Airbus and Boeing. Cardiac Diffusion applies AI to cardiac MRI for clinical heart modeling. In June 2026, the U.S. government awarded SandboxAQ a $500 million program to discover new chipmaking materials — the clearest signal yet that LQMs have first-tier government demand.
Funding has tracked the milestones: a $300M+ first close at a $5.6B valuation (December 2024, Bloomberg), a $450M+ Series E at $5.75B (April 2025) with participation from Google and NVIDIA, and a $95M follow-on (July 2025) led by Rizvi Traverse, Forge Global, and the Ava Family Office — more than $1B raised in total per Reuters.
SandboxAQ is Alphabet's most successful scientific-AI spinout: rather than competing with ChatGPT, it applies AI to equations — navigation, cryptography, chemistry, materials — selling to governments and enterprises where the physics is the moat. The June 2026 $500M U.S. award for chipmaking-materials discovery is the strongest proof yet that the LQM thesis has real government demand, and a $5.75B valuation plus NVIDIA and Google backing gives it runway. The risks are classic deep-tech: long government sales cycles, lumpy revenue concentration, and the brand tax of explaining why 'AQ' is mostly physics-based classical AI with quantum at the edges.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki