The US Department of Energy (DOE) has awarded funding to 12 Phase II projects under its "Genesis Mission" initiative.

Genesis Mission, first launched in November 2025, aims to create and train AI agents and scientific foundation models to speed up scientific progress. The winning projects use DOE supercomputers, cloud-based AI resources, and federal computing resources to support their research.

Between them, the DOE is providing $159 million in funding. This brings the total number of Phase II projects selected by Genesis Mission to 14, with six Phase I projects expected to follow.

“These projects represent a critical step in turning the promise of AI for science into transformative scientific capability,” said DOE under secretary for science Dr. Darío Gil. “The Phase II teams have already demonstrated what is possible when AI is integrated with scientific expertise, data, and computing, and these awards give them the opportunity to take that work to the next level. By bringing together our National Labs, universities, and industry partners, we are building on extraordinary work already underway and accelerating the pace of discovery.”

The 12 projects are as follows:

• Accelerating the path to commercial fusion energy via an AI-enabled digital twin platform for SPARC: Led by Commonwealth Fusion Systems, this project will build a digital twin to simulate and optimize operations of a fusion demonstration device.
• Decoding the RNA Structurome to Secure AI Advantage for the Bioeconomy: Led by the University of California San Diego, the project will expand the RNA structure database five-fold for training AI models to unlock the ability to engineer resilient crops and sustainable microbes.
• Lattice Quantum Chromodynamics (LQCD) at the Intelligence Frontier: Led by MIT, this project will integrate artificial intelligence and supercomputing calculations of fundamental particle physics to dramatically accelerate discoveries about the basic building blocks of our universe.
• Multi-agent AI Expert for Subsurface Reasoning and Optimization (MAESTRO): Led by the University of California Irvine, MAESTRO will establish an AI platform to accurately map, stimulate, and manage deep underground reservoirs to unleash vast resources of reliable geothermal energy.
• AI-extended Interfacial Separations (AXIS): A Multiscale Framework for Redox-Active Ligand Design: Led by the University of Illinois Urbana-Champaign, AXIS will harness AI to design more efficient electrochemical processes for extracting critical rare-earth elements from domestic sources.
• Overcoming barriers in computational enzyme design through improved sequence-structure ensemble modeling: Led by the University of Washington Seattle, this project will build foundational AI tools to predict and design functional enzymes.
• Accelerating Extreme Environment Specs-to-Silicon (AXESS): Led by Fermilab, AXESS will use AI to rapidly design rugged, high-performance microchips capable of operating in extreme conditions such as space, high radiation, and ultra-cold environments.
• The Multi-office Accelerator Team Core (MOAT-Core): Led by Lawrence Berkeley National Laboratory, MOAT-Core will develop a shared AI platform across National Laboratories to autonomously run and design particle accelerators faster and more efficiently.
• Application-Aware Error Correcting Codesign for Scientific Quantum Computing (ASQC): Led by Harvard University, ASQC combines AI with advanced quantum hardware to tackle quantum computing error correction.
• An Iterative Framework of AI-Assisted Scientific Development and Optimization (AI4HPC): Led by Argonne National Laboratory, AI4HPC will create an AI framework to modernize, optimize, and verify supercomputing scientific software.
• AI-empowered Design of Functional Quantum Magnets: Led by Oak Ridge National Laboratory, this project will develop a physics-informed AI framework to design new quantum magnets by working backward from desired properties, with applications in quantum sensing, low-power electronics, and quantum technologies.
• Agentic Assistance for Advanced Rare Isotope Separator Operations at the Facility for Rare Isotope Beams (FRIB): Led by Northeastern University, this project will develop an intelligent, automated assistant to safely streamline the operation and tuning of the FRIB particle separator, unlocking hundreds of hours of research time annually.

The Oak Ridge National Laboratory is providing its supercomputing capacity to six of the new projects.

“These selections align with ORNL priorities in [Artificial] Intelligence, quantum, fusion energy, critical materials and user facilities, underscoring ORNL’s focus on transforming discovery into national impact,” said ORNL director Stephen Streiffer. “The goal is straightforward — faster, more reliable science that delivers practical results where they matter most.”

Outside of the traditional supercomputing laboratories, cloud providers are also joining up to offer AI compute capacity.

Amazon Web Services (AWS) has pledged up to $50 million in credits over the next three years to support the Genesis Mission, while Crusoe Cloud has also committed to offering $50 million in AI compute credits to the initiative.

"AI-driven computation is fast becoming the foundational engine of modern scientific discovery," said Crusoe co-founder and CEO Chase Lochmiller. "By committing $50 million in AI compute and inference credits to the Genesis Mission, Crusoe is putting powerful AI infrastructure directly into the hands of federal researchers tackling the nation's biggest challenges in energy and health. We're proud to make Crusoe Cloud more accessible to more innovators."

Microsoft joined the Genesis Mission in July of this year, offering $40 million in Azure compute and AI credits, and $20 million in solution engineering enablement services. Google similarly offered $40 million in cloud credits that month.