Driving the Future of Transportation: SES AI's NVIDIA-Powered Breakthrough in Electric Vehicles

 

Accelerating EV Innovation: How SES AI and NVIDIA are Transforming Battery Technology

Electric vehicles (EVs) are reshaping transportation, but widespread adoption still faces challenges such as cost, battery longevity, and driving range. At the core of these issues lies battery technology—specifically, the electrolyte, a crucial component that determines charging speed, power output, stability, and safety.

To overcome these barriers, researchers are leveraging AI-driven approaches to accelerate the discovery of advanced battery materials.

Revolutionizing Battery Innovation with AI

SES AI, a leader in next-generation battery technology, is harnessing the power of NVIDIA’s cutting-edge hardware and software ecosystem to drive breakthroughs in materials discovery. By integrating domain-adapted large language models (LLMs) with AI-driven simulations on GPUs, SES AI condenses decades of research into mere months—unlocking transformative advancements in EV battery performance.

SES AI’s AI-Powered Battery Development Approach

SES AI’s pioneering research follows a structured, AI-enhanced methodology:

Mapping the Molecular Universe

SES AI is constructing an extensive molecular database to catalog the physical and chemical properties of millions of small molecules. Central to this effort is AIMNet2 (Atoms in Molecules Neural Network)—a neural network potential developed by Carnegie Mellon University’s Isayev Lab. AIMNet2 significantly enhances the speed and accuracy of molecular simulations, expediting property predictions.

"Our goal is to map the properties of small molecules to develop superior energy storage solutions—for applications ranging from flying cars to humanoid robots and data centers," says Qichao Hu, CEO of SES AI. "With our collaboration with NVIDIA and the use of AIMNet2, we've accelerated this process from thousands of years to just a few months."

Advancing Chemistry LLMs

By integrating domain-specific Chemistry LLMs, SES AI enhances reasoning capabilities, streamlining the discovery of novel battery materials and providing researchers with powerful tools for next-generation battery development.

Constructing High-Performance Batteries

SES AI rapidly tests and refines candidate materials in prototypes, demonstrating significant performance improvements and paving the way for safer, longer-lasting EV batteries.

Leveraging NVIDIA’s AI Platform for Breakthroughs

SES AI relies on NVIDIA’s powerful AI platform to explore a vast solution space of over 100 million molecules. Their workflow integrates accelerated molecular simulations, optimized data exploration, and enhanced LLM-driven insights, pushing the boundaries of battery innovation.

By combining AI, advanced simulations, and high-performance computing, SES AI and NVIDIA are accelerating the future of EVs—bringing us closer to more efficient, affordable, and sustainable electric transportation.

                                          Figure 1. SES AI’s NVIDIA-enabled workflow

Enhancing Molecular Simulations with NVIDIA ALCHEMI

SES AI leverages NVIDIA ALCHEMI to accelerate molecular simulations, utilizing Batched Geometric Relaxation (BGR) with AIMNet2 NIM microservice and the Batched Density Functional Theory (DFT) microservice. These cutting-edge tools enable rapid molecular structure simulations and the calculation of essential properties, such as the HOMO-LUMO gap, which defines a molecule’s electronic characteristics and energy storage potential.

The BGR AIMNet2 NIM microservice streamlines the search for the most stable molecular configuration, significantly reducing computational time. Meanwhile, the Batched DFT microservice enhances GPU4PySCF—an open-source, GPU-accelerated Python package integrated into the Python-based Simulations of Chemistry Framework (PySCF)—boosting the efficiency of quantum chemistry calculations.

By integrating these innovations, SES AI achieves unprecedented speed and accuracy in molecular simulations, accelerating breakthroughs in battery materials and energy storage technology.

Accelerating Battery Innovation with NVIDIA-Powered AI

Demo: Boosting Simulations with NVIDIA ALCHEMI BGR AIMNet2 NIM

Interactive Data Exploration

SES AI leverages NVIDIA cuML to accelerate UMAP for dimensionality reduction and HDBSCAN for clustering, enabling the creation of interactive molecular maps. These AI-driven tools help researchers analyze millions of molecules efficiently, uncover hidden patterns, and rapidly identify promising candidates for next-generation battery materials.

Training Chemistry LLMs

To train its Chemistry LLM on 35 billion tokens sourced from 17 million scientific papers, SES AI utilized NVIDIA DGX Cloud and NVIDIA NeMo, a framework for building domain-specific generative models.

For enhanced accuracy, SES AI applied NeMo Curator features, including:

• Exact deduplication

• Number and word count filters

• Repeated line and non-alphanumeric filters

These optimizations resulted in a customized Chemistry LLM that surpasses its base model in molecular ranking and scientific reasoning—delivering faster, more precise insights for battery material discovery.

Breakthrough Results: Transforming Battery Technology

The collaboration between SES AI and NVIDIA has led to groundbreaking advancements:

• 17 candidate materials identified – Two already synthesized and tested at SES AI’s Electrolyte Foundry in Woburn, Massachusetts.

• 20% increase in battery cycle life – Early tests reveal significant improvements in battery longevity.

• 1,600x simulation acceleration & 2x energy efficiency – Achieved through NVIDIA H100 GPUs, ALCHEMI’s BGR AIMNet2 NIM, and Batched DFT Microservice.

• 10x faster data exploration – Enabled by NVIDIA cuML.

• 25% improvement in LLM accuracy – Chemistry LLM, trained on NVIDIA DGX Cloud and NeMo, outperforms the base model in molecular ranking and reasoning.

Innovations Driving the Future of Batteries

These advancements have fueled the development of next-generation battery technologies:

• Lithium-metal batteries – Offering superior performance and safety.

• High-silicon lithium-ion batteries – A newly identified solvent molecule addresses expansion challenges without emitting harmful gases at high temperatures.

Driving Sustainability Through Accelerated Innovation

Traditionally, discovering a breakthrough battery electrolyte required five years. Today, with SES AI and NVIDIA’s cutting-edge technologies, transformative advancements are emerging every six months. These innovations are shaping the future of EV batteries—enhancing performance, safety, and longevity while significantly reducing greenhouse gas emissions and promoting sustainable practices worldwide.

By accelerating materials discovery, SES AI and NVIDIA are not only propelling technological progress but also stimulating economic growth and reinforcing global efforts to combat climate change.