As of January 2026, the U.S. Department of Energy (DOE) has significantly ramped up investments in artificial intelligence (AI), including agentic systems—autonomous, goal-oriented AI agents capable of perception, reasoning, and action—to advance scientific discovery. This aligns with the Genesis Mission, launched in November 2025 via Executive Order 14363, which aims to integrate AI with DOE's datasets and facilities to create AI agents that automate workflows, test hypotheses, and accelerate breakthroughs in energy, science, and security.[23] While "agentic systems" is an emerging term often linked to large language model (LLM)-based autonomous agents, DOE funding emphasizes AI-driven autonomy in areas like predictive modeling, experimental automation, and data curation. Total AI investments exceed $320 million as of December 2025, with applications in geosciences (e.g., earth system modeling, critical minerals) and biosciences (e.g., bioenergy, genomics).[7] Funding is channeled through the Office of Science (SC), including programs like Basic Energy Sciences (BES), Biological and Environmental Research (BER), and Advanced Scientific Computing Research (ASCR), often via open calls or targeted announcements.[6] Below is a summary of relevant grants, focusing on those incorporating agentic or autonomous AI elements.
DOE supports geosciences through BER's Earth and Environmental Systems Sciences and BES's geosciences research, emphasizing AI for modeling complex earth systems, critical minerals extraction, and climate resilience. Agentic AI is integrated for autonomous experimentation and predictive maintenance in geothermal and mineral resources.
Over $6 million awarded across national labs, including Lawrence Berkeley National Lab's Energy Geosciences Division (EGD), for advancing lithium extraction from geothermal brines. This includes AI-driven autonomous systems for process optimization, economic modeling, and environmental sustainability assessments. Projects explore agentic frameworks to automate brine analysis and enhance geothermal efficiency, potentially using AI agents for real-time hypothesis testing.[5] Funding supports four projects, with EGD leading two, focusing on replicating Salton Sea lithium models in other regions.
Part of the $320 million package, includes $40 million for the American Science Cloud and foundational AI models at national labs. Specific to geosciences, awards fund AI agents for climate modeling, wildfire prediction (e.g., FireSat model), and autonomous workflows in earth system simulations. Self-improving AI models harness DOE data for embodied AI in laboratory environments, applicable to geosciences like seismic analysis and resource mapping.[20][23] Collaborations with 24 organizations (announced December 18, 2025) emphasize public-private partnerships for agentic AI in energy infrastructure, including geosciences.[24]
Up to $500 million available for ongoing proposals in environmental system science, including AI/ML for earth system predictability. Encourages agentic systems for multi-scale modeling of geochemical processes, with deadlines through September 30, 2026. Pre-applications are required; focuses on autonomous data integration for climate and geosciences.[14][26]
DOE's biosciences funding, primarily through BER's Biological Systems Science and the Bioenergy Technologies Office (BETO), leverages AI for genomics, bioenergy, and biotechnology. Agentic systems are highlighted in AI-centric reinventions, such as at the Joint Genome Institute (JGI), for autonomous biological discovery.
Up to $20.2 million for algal systems R&D, including AI for biofuels and bioproducts. Supports agentic AI for optimizing biomass conversion, with integrations like Aviary (LLM agents for biological tasks) to automate hypothesis generation and experimental design. Recent awards (e.g., $10 million for algal feedstock) emphasize autonomous systems for mixed algae processing.[0][11]
JGI is transforming into an AI-centric user facility, funding AI agents for genomic science. This includes $320 million Genesis-related investments for self-improving AI models in biosciences, such as protein sequence prediction and drug repurposing using LLMs. Agentic systems enable autonomous viral protein analysis and biosecurity guardrails.[27][29]
Similar to geosciences, this supports biosciences with emphasis on AI for bioenergy genomics and environmental biology. Proposals for agentic AI in plant feedstock genomics (joint with USDA) aim at autonomous optimization for biofuels, with total funding up to $500 million.[14][36]
$30 million for self-improving AI agents in biosciences, including foundation models like Med-Gemini for multimodal biological data. Focuses on agentic systems for drug discovery, genomic analysis, and bioengineering, harnessing DOE's datasets.[20][32]