SandboxAQ has unveiled a major breakthrough in computational chemistry, leveraging NVIDIA technology to accelerate calculations by an impressive 80x. This collaboration, utilizing SandboxAQ’s Quantitative AI models and NVIDIA’s CUDA-accelerated Density Matrix Renormalization Group (DMRG) algorithm, marks a significant leap in fields such as biopharma, chemicals, and materials science.
By integrating the CUDA-DMRG algorithm with the NVIDIA Quantum platform, scientists can now perform highly accurate simulations of real-life systems, surpassing the capabilities of existing AI models. This advancement doubles the size of computable catalysts and enzyme active sites, fostering precision in computational chemistry.
Dr. Martin Ganahl, senior staff scientist at SandboxAQ, highlighted the transformative impact of this technology. “Advanced computing is opening new frontiers in scientific research. Our collaboration with NVIDIA addresses some of the most challenging problems in chemistry, paving the way for innovations in drug discovery and catalysis to tackle currently untreatable conditions and find safer, cost-effective methods to synthesize molecules and materials,” he stated.
Tim Costa, director of high-performance and quantum computing at NVIDIA, emphasized the broader implications. “AI supercomputing is solving critical problems in the chemical and pharmaceutical industries. SandboxAQ’s use of the NVIDIA Quantum platform enables simulations at an unprecedented scale, redefining what’s possible in computational chemistry,” Costa remarked.
Jim Breyer, Founder and CEO of Breyer Capital and an early investor in SandboxAQ, underscored the significance of these developments. “This work with NVIDIA highlights SandboxAQ’s dedication to scientific discovery and technological innovation. Unlocking new compounds and catalysts heralds a new era of breakthroughs across various industries, with profound implications for improving quality of life and driving economic growth,” Breyer commented.
SandboxAQ’s pioneering efforts extend beyond this collaboration. The company has previously announced AI partnerships with institutions such as the University of California San Francisco (UCSF), NOVONIX, and Riboscience. In 2024, organizations including Flagship Pioneering and SPARK NS joined forces with SandboxAQ to enhance their innovation pipelines.
The applications of Large Quantitative Models (LQMs) are vast, ranging from biopharma to advanced materials. Notably, LQMs can revolutionize biopharma by accurately modeling enzyme Cytochrome P450s, crucial for understanding drug metabolism and toxicity. The CUDA-DMRG algorithm offers a groundbreaking approach to computational toxicity prediction, potentially reducing risks in clinical trials.
Quantitative AI, at the core of SandboxAQ’s innovations, enables large AI models to make precise, physics-based predictions. Unlike traditional Large Language Models (LLMs), which rely on existing data, SandboxAQ’s LQMs generate training data through physics-based simulations, unlocking new possibilities in the physical world.