Partnerships

Argonne partners with leading technology companies to advance AI research that accelerates innovation, strengthens U.S. competitiveness and drives scientific discovery.

National Security

American Battery Technology Company: This partnership is advancing domestic battery material manufacturing and battery material recycling. AI is helping researchers investigate membrane performance and durability in electrodialysis processes used in lithium extraction and refining as well as battery recycling.

Dow Chemical USA: Argonne applied computational fluid dynamics simulations and an Argonne-developed machine learning algorithm called ActivO to optimize mixing equipment design while requiring significantly fewer simulations than traditional optimizers.

Sentient Science: Argonne and software company Sentient Science are making the discovery and design of new materials — a notoriously time-consuming and costly process — significantly easier through the development of an artificial intelligence-based computer modeling framework, known as BLAST (Bridging Length/Time Scales Atomistic Simulation Toolkit).

Energy

ComEd: Argonne and ComEd are using AI for smart meter anomaly detection. Millions of smart meters record voltage and usage signals, contributing to a shared AI model. In turn, AI detects unusual patterns earlier, resulting in early detection of equipment faults and unusual events.

DOE Office of Nuclear Energy: Through a partnership with the Office of Nuclear Energy, Argonne is creating the Regulatory Context Protocol (RCP), which is a framework to streamline the licensing process for advanced nuclear reactors. It automates applicant-regulator communication using AI agents that represent both the nuclear facility and the regulator, reducing delays in regulatory workflows, improving information quality, and ensuring compliance with Nuclear Regulatory Commission standards.

Kairos Power: Using machine learning, Argonne and Kairos Power have developed algorithms to analyze sensor data, helping to automate nuclear reactor monitoring.

U.S. Nuclear Regulatory Commission: Argonne and the U.S. Nuclear Regulatory Commission are partnering to develop and evaluate AI applications for the nuclear industry. Some potential uses for AI are predictive maintenance, modeling, and optimization.

Science

AI companies (AWS, Anthropic, Google, Microsoft, OpenAI): Argonne is working with leading AI companies to explore using their cutting-edge AI capabilities to directly benefit scientific research. Partners have provided early access to advanced models, and Argonne scientists in collaboration with other DOE labs have tested these models on novel problems, sharing their feedback to improve model development.

ALCF AI Testbed vendors (Cerebras, Graphcore, Groq, SambaNova): The ALCF AI Testbed is Argonne’s facility for exploring next-generation AI accelerator hardware.

Google: In addition to AI model testing, Argonne is testing Google’s Tensor Processing Unit as a potential platform for the next generation of AI computing at Argonne. Argonne and Google also have a project to examine the energy impact of the rapid adoption of AI technologies and determine if AI adoption will lead to energy savings through efficiency gains or if AI-driven industrial development will result in increased energy use due to overall expansion of production and services.

Intel and Hewlett Packard Enterprise: The Aurora exascale supercomputer, purpose built for AI, was built in partnership with Intel and HPE.

NVIDIA: NVIDIA GPUs are in several current and future ALCF systems. The Polaris computer is GPU-accelerated, using NVIDIA A100 Tensor Core GPUs. Argonne’s ThetaGPU is a GPU-augmented version of its Theta system, using NVIDIA DGX A100 nodes.