August 5, 2025
5 min read
Lawrence Livermore National Laboratory
LLNL researchers deploy AI agents on top supercomputers to automate and accelerate inertial confinement fusion target design and simulations.
LLNL Advances Fusion Target Design Using AI on Exascale Supercomputers
Lawrence Livermore National Laboratory (LLNL) researchers have reached a significant milestone by integrating artificial intelligence (AI) with fusion target design. They deployed AI agents on two of the world’s most powerful supercomputers to automate and accelerate inertial confinement fusion (ICF) experiments. Part of an AI framework called the Multi-Agent Design Assistant (MADA), LLNL scientists and collaborators combine Large Language Models (LLMs) with advanced simulation tools. This system interprets natural language prompts from human designers and generates full physics simulation decks for LLNL’s next-generation 3D multiphysics code, MARBL. MARBL enables the design and analysis of mission-critical, high-energy-density experiments, including ICF. In ICF experiments at LLNL’s National Ignition Facility (NIF), fusion energy is produced when 192 laser beams converge on a tiny target containing deuterium and tritium, triggering a fusion chain reaction. The MADA team tests their AI system on the exascale El Capitan supercomputer — the world’s fastest at 2.79 exaFLOPs peak — and its smaller sibling, Tuolumne. The framework includes an Inverse Design Agent (IDA) to engineer new ICF targets. Jon Belof, LLNL physicist and principal investigator, noted the project began in 2019 with the idea of combining AI and shockwave physics. As LLMs advanced, the concept of semi-autonomous AI systems collaborating with humans on ICF design became a natural evolution. The MADA team, including collaborators from the National Nuclear Security Administration (NNSA) Tri-Lab partners at Los Alamos (LANL) and Sandia National Laboratories (SNL), has transformed this concept into a sophisticated AI-driven design workflow. In a recent demonstration, an open-source LLM fine-tuned on MARBL documentation successfully interpreted a hand-drawn capsule diagram and a natural language request from a human designer. It then generated a complete simulation deck and ran thousands of simulations to explore variations in ICF capsule geometry, resulting in a novel target design. This AI-driven design paradigm arrives at a pivotal moment for fusion research. Following LLNL’s historic ignition achievement at NIF in December 2022, the lab aims to develop a robust ignition platform that could unlock new national security applications. Belof emphasized that tools like MADA drastically compress design cycles and explore vast design spaces. By pairing human insight with AI-driven exploration, LLNL hopes to navigate the complex physics of high-gain implosions faster and more efficiently.“In principle, AI agents offer a way for us to pursue not only 3-4 distinct ICF design concepts at once – but hundreds or possibly thousands,” Belof explained. “Rather than the human running ensembles of simulations, they will be able to run ensembles of ideas. This concept could be massively transformative in nature.”The core of MADA is its AI “agents” — autonomous software entities composed of two key components: an LLM that understands and responds to human language, and a specialized tooling interface that enables domain-specific tasks. For MADA, this tooling generates structured simulation input files and launches them on high-performance computing (HPC) systems.
“We are putting AI in the driver’s seat of a supercomputer, which is something that has never been done before,” Belof said.Supporting the Inverse Design Agent is the Job Management Agent (JMA), which manages the execution of large-scale simulation workflows across LLNL's supercomputers. It interacts with the Flux scheduler and workflow tools like Merlin. The JMA ensures jobs are queued properly, resources allocated, and simulation outputs efficiently harvested for analysis. Together, the IDA and JMA form a seamless loop between AI planning and HPC execution.
“The Job Management Agent brings AI and HPC together to coordinate agents that handle resource management and workflow optimization at massive scales, giving us a critical advantage as we push toward a robust ignition platform for fusion energy,” said JMA Team Lead Giselle Fernandez.This iterative workflow enables unprecedented interactivity between designers and simulations. Instead of manually coding and launching jobs — a process that could take days or weeks — researchers can now explore thousands of design variations in parallel by conversing with an AI agent.
“The agent can then take a capsule diagram and a plain-language prompt like, ‘Explore the effect of changing a certain part of the geometry and translate that into a valid simulation deck for MARBL,’” Belof explained. “It then runs that deck, collects results, and can even build a training dataset to power a surrogate model.”MADA leverages HPC to run massive ensembles — tens of thousands of ICF simulations in a single study — across LLNL’s Tuolumne supercomputer, the world’s 12th fastest. The simulation outputs train a machine learning model called PROFESSOR, which provides instant feedback for designers exploring new capsule geometries.
“Once trained, the PROFESSOR model generates implosion time histories — radius as a function of time — that change instantaneously when the human designer changes the input geometry,” Belof said. “It’s a powerful new tool to ICF designers that is made possible with AI/machine learning plus HPC.”By enabling natural language interaction, image interpretation, and rapid simulation-to-model pipelines, MADA demonstrates how AI can be embedded directly into high-stakes scientific workflows. This replaces slow, manual iteration with collaborative AI augmentation. The implications extend beyond ICF. As more exascale systems like El Capitan come online, MADA offers a blueprint for AI agents as digital collaborators in fields ranging from materials discovery to weapons certification.
“It’s really about enhancing human productivity through AI, in a transformative way,” Belof said. “And I think this project shows that we’re just beginning to tap what’s possible. AI tools have the potential for allowing us to best allocate resources and help understand tradeoffs that will be needed for the next generation of enhanced fusion facilities.”NNSA’s Advanced Simulation & Computing program funds this work. Other LLNL MADA team members include deputy principal investigator Charles Jekel, MARBL Project Lead Rob Rieben, and researchers Will Schill, Meir Shachar, and Dane Sterbentz. Nathan Brown of SNL and Ismael Djibrilla Boureima of LANL also contributed.
Source: LLNL Pushes Frontier of Fusion Target Design with AI (Lawrence Livermore National Laboratory, August 4, 2025)