Lawrence Livermore National Laboratory (LLNL) leaders, scientists and engineers joined national voices at the Special Competitive Studies Project's (SCSP) AI+ Expo May 7-9 in Washington, D.C., highlighting how AI is reshaping science, security and energy innovation.
The public Expo brought together government, industry, academic and Department of Energy (DOE) national laboratories for three days of sessions, demonstrations and exhibits focused on AI, national security and U.S. technological competitiveness. It was the third such event hosted by the SCSP, a nonprofit, nonpartisan organization focused on U.S. competitiveness in AI, emerging technology and national security.
For LLNL, the conference offered a national stage to show the Lab's efforts in AI for science and how DOE's Genesis Mission is moving from concept to capability, with Lab leadership and researchers participating in panels, technical talks and live demonstrations spanning AI-enabled design, molecular discovery, bioresilience, high-performance computing (HPC) and fusion energy.
LLNL's presence was especially visible during a busy stretch of programming on May 7, with Lab representatives contributing to sessions across the Expo. LLNL Director Kim Budil joined a public-private panel on AI for science, where she pointed to LLNL capabilities and AI focus areas, including the National Ignition Facility (NIF), the exascale supercomputer El Capitan, high-fidelity modeling and simulation and advanced materials and manufacturing.
Budil said the Lab is using AI and machine learning to speed cycles of learning, improve simulations and rethink how advanced manufacturing can support national security missions. Bringing AI into manufacturing, Budil added, could change not only how components are built, but how technologies are designed in the first place.
Budil also discussed how the Genesis Mission is designed to bring the national laboratories' scientific workforces, large-scale experimental facilities, advanced computing capabilities and mission focus together in a new AI-enabled framework.
"The goal for Genesis is to revolutionize how we execute all those missions and incredibly enhance the productivity of our researchers," Budil said. "Bringing together all this intellectual horsepower, with these incredible experimental and computational capabilities, is our opportunity to transform the way science is done in America."
Across the venue, Brian Spears, technical director for the Genesis Mission, joined a panel on Genesis and framed the effort as a national-scale push to harness AI for science, technology and security. Spears said Genesis could use AI to double the impact of U.S. scientific research and development while delivering what he called "innovation overmatch" for national security superiority. He said Genesis is intended to connect AI with the DOE complex's distinctive strengths, from HPC and precision experimentation to high-consequence production.
"We sit inside a computing revolution," Spears said. "These AI technologies are transformative. The Genesis Mission is AI to uplift the entirety of the U.S. ecosystem - public and private."
Meanwhile, in a conversation on bioconvergence, biosecurity and bioresilience, LLNL Bioresilience Incubator Director Shankar Sundaram highlighted LLNL's pioneering role in connecting frontier science and national security needs in biodefense. He emphasized that AI-enabled progress in building national bioresilience depends on bringing together biological data, predictive models and compute at scale.
"National labs have the ability to bring scientific and technological depth along with a national security mission and mindset," Sundaram said. "Data, especially functional data, is a strategic asset."
Over the past decade, LLNL has developed and demonstrated AI and computation-driven methods that are reshaping preparedness - improving early detection and accelerating the development of countermeasures against emerging biological threats, while working with the private sector to help broaden their application in the health sector, Sundaram explained.
The Lab's presence reflected a larger DOE message that opened the Expo on May 7. In a fireside chat, Secretary of Energy Chris Wright framed the Genesis Mission as a way to connect AI tools, national laboratory capabilities, scientific data and private-sector partners to accelerate progress on major science and energy challenges.
"Take AI tools, take our national labs, take the data sets, take awesome partners, and rapidly increase our ability to innovate things that take years to test, diagnose and figure out how they work," Wright said. "How can we do that in months so we can massively accelerate the pace of scientific discovery? That's the Genesis Mission."
DOE Under Secretary for Science Dario Gil also emphasized the speed and scale of the effort during his fireside chat on May 9. Gil said Genesis' recent call for proposals has drawn thousands of submissions from more than 800 institutions, adding that the response reflected a broad national push to organize universities, laboratories and industry around AI-enabled science.
"Ultimately, we seek to double the productivity and impact of America's trillion-dollar-a-year R&D engine within a decade," Gil said.
LLNL's participation extended beyond the main-stage conversations. LLNL computer scientist Brian Van Essen delivered a technical talk at the DOE booth on FLASK Copilot, an AI-enabled tool for molecular and materials discovery. The project stems from a Laboratory Directed Research & Development Strategic Initiative led by Van Essen (Foundation-Learning Artificial Intelligence for Synthesis Knowledge) and is designed to help researchers identify new molecules and optimize molecular properties.
In practice, the tool is intended to help scientists move more quickly from a desired material or molecular property to candidate molecules and possible synthesis pathways, reducing the time spent navigating disconnected tools and computational environments.
Van Essen said FLASK Copilot is designed to accelerate discovery by connecting commercial frontier AI models, including OpenAI's ChatGPT, with LLNL-developed chemistry tools, HPC resources, custom models and human domain expertise. The goal, he said, is to reduce the friction between scientific ideas and computational workflows.
"AI by itself cannot solve our national science and security problems," Van Essen said. "We need to couple it with our traditional modeling and simulation tools, our domain experts and our unique experimental resources."
That coupling is only useful if scientists can actually use the tools in the environments where they work, Van Essen added, explaining that the broader promise of FLASK Copilot and similar agentic systems is that they can extend team science into AI-enabled workflows.
"These agentic, multidisciplinary workflows can bring the team together and make everybody better outside of their domain," Van Essen said.
At the DOE booth, Van Essen also demonstrated the Multi-Agent Design Assistant (MADA), an LLNL-developed AI framework that combines large language models with simulation tools to help interpret natural language prompts from human designers and generate physics simulation inputs. MADA has been used in fusion target design work to generate simulation decks for MARBL, the Lab's next-generation 3D multiphysics code, and to explore variations in inertial confinement fusion capsule geometry using LLNL supercomputers like El Capitan and Tuolumne.
In another booth demo on May 9, Derek Mariscal and Mackenzie Nelson from LLNL's Livermore Institute for Fusion Technology discussed AI-driven tools to support the design and eventual operation of fusion power plants. The demo, which Gil visited ahead of his fireside chat, connected to LLNL research on high-repetition-rate laser systems - the kind that would be needed for future inertial fusion energy power plants - where lasers ignite targets 10 times or more per second and require AI-enabled prediction, control and operations.
"This tool represents a joint effort from national lab and academic partners to pool our expertise and apply frontier AI models to advancing the national goal of commercial fusion energy, in concert with the Genesis Mission," said Mariscal.
Throughout the three-day event, LLNL's presence reflected a central theme of the Expo: AI is moving from a standalone technology to connective tissue linking data, simulation, experimentation, manufacturing and mission execution.
For LLNL, that shift does not mean replacing scientists, but giving them new ways to move faster through complex design spaces, ask better questions and connect national laboratory capabilities to urgent challenges in security, energy and discovery science. Realizing that future, Budil said, also will require new ways of collaborating with industry and other external institutions.
"We're learning how to work in a very different way with the private sector," Budil said. "They're peer organizations and partners in a way we've not experienced before."
