SandboxAQ & Nvidia: AI Molecule Dataset for Drug Discovery
- — SandboxAQ, an AI startup originating from Alphabet and supported by Nvidia, has launched a vast dataset of 5.2 million synthetic 3D molecules.
- The dataset, called the Structurally Augmented IC50 Repository (SAIR), is publicly available.
- SandboxAQ's approach combines physics-based modeling with machine learning, potentially reshaping early-stage pharmaceutical research and advancement.
SandboxAQ, backed by Nvidia, has unleashed a groundbreaking AI dataset of 5.2 million synthetic molecules, revolutionizing drug discovery. This massive repository, called SAIR, aims to drastically accelerate teh identification of potential drug candidates by simulating drug-protein interactions with exceptional speed and precision. The new AI dataset combines physics-based modeling and machine learning, perhaps cutting down months of lab work into a single prediction. SandboxAQ plans to monetize its AI models, offering virtual labs as a service, signifying a major shift in pharmaceutical research. This technology also extends to cybersecurity, and medical diagnostics. News Directory 3 is proud to highlight this advancement. Discover what’s next as SandboxAQ pioneers the future of scientific exploration.
SandboxAQ’s Massive AI Dataset Accelerates Drug Discovery
Updated June 19, 2025
Palo Alto, Calif. — SandboxAQ, an AI startup originating from Alphabet and supported by Nvidia, has launched a vast dataset of 5.2 million synthetic 3D molecules. The goal is to speed up the process of drug discovery. These molecules,created using Nvidia’s chips,simulate how drugs interact with proteins,a crucial step in developing effective treatments.
The dataset, called the Structurally Augmented IC50 Repository (SAIR), is publicly available. It is designed to train AI models to predict drug-protein interactions much faster than conventional lab methods. This AI dataset seeks to replicate lab results virtually with high accuracy, possibly reducing months of research to a single AI prediction.
SandboxAQ’s approach combines physics-based modeling with machine learning, potentially reshaping early-stage pharmaceutical research and advancement. The company intends to monetize its trained models, offering virtual labs as a service.
Headquartered in Silicon Valley, SandboxAQ blends quantum technology and AI. Spun out of Alphabet’s “moonshot factory,” the company has nearly $1 billion in venture capital, including backing from Nvidia.
SandboxAQ’s core technology involves Large Quantitative Models, AI systems grounded in physics, chemistry, and biology. These models simulate the real world, autonomously exploring millions of chemical pathways to discover novel molecules.
Beyond drug discovery, SandboxAQ is also addressing cybersecurity, navigation in GPS-denied environments, medical diagnostics using AI to analyse cardiac signals, and materials science by predicting atomic-level properties.
Jack Hidary, a tech entrepreneur with a background in quantum computing and neuroscience, founded SandboxAQ. Hidary, who studied beliefs and neuroscience at Columbia University, previously led a quantum initiative at alphabet before spinning out SandboxAQ in 2022. He also authored a book on quantum computing.
Eric Schmidt, former CEO of Google, serves as chairman of the board.
What’s next
SandboxAQ plans to further develop its AI models trained on the SAIR dataset, offering them as a service to pharmaceutical companies and research institutions, potentially transforming the landscape of early-stage drug development.