DOE's $159M Superintelligence Bet: 12 New Genesis Mission Projects for Fusion, Quantum, and Critical Minerals

The Department of Energy today announced $159 million in new awards to advance superintelligence for science (what DOE officially brands as Super Intelligence, or SI), capping year one of the Genesis Mission at 297 projects.

WASHINGTON, October 8, 2026. DOE unveiled 12 new Phase II Genesis Mission awards plus 6 new Phase I awards. The Genesis Mission, launched in November 2025, is DOE's flagship program to pair America's scientists with advanced AI, supercomputing, scientific data, and national lab infrastructure to accelerate discovery.

These projects represent a critical step in turning the promise of Super Intelligence for science into transformative scientific capability," said DOE Under Secretary for Science Dr. Darío Gil.

For founders, this is the clearest signal yet that DOE is buying superintelligence for science as infrastructure, not just research.

TL;DR for Founders

• $159M across 12 Phase II projects announced today, $6M-$15M each over 3 years
• 297 projects total in Genesis Mission year-one portfolio
• Startup can lead: One winner is Commonwealth Fusion Systems, plus NVIDIA is collaborating on multiple Phase II winners in quantum, fusion, accelerators and microelectronics
• Direct on-ramp for startups: Phase I is $500K-$750K for 9 months. Phase II is $6M-$15M. You can apply directly to either in FY26, and Phase I winners can compete for Phase II
• Plus $196M in talent funding today: $100M Graduate Fellowship for 100 SI-ready PhDs + $96M for 67 Early Career scientists

The 12 Superintelligence Bets, Decoded for Builders

Energy & Critical Minerals: immediate commercial pull

1. SPARC Digital Twin, Commonwealth Fusion Systems. AI-enabled digital twin to simulate and optimize a fusion demonstration device. The playbook for de-risking commercial fusion ops.

2. MAESTRO, UC Irvine. Multi-agent AI expert for subsurface reasoning to map, stimulate and manage deep geothermal reservoirs. Infrastructure AI for the geothermal boom.

3. AXIS, UIUC. AI-extended interfacial separations to design electrochemical processes for extracting domestic rare-earth elements. Onshoring play.

Chips & Compute: the superintelligence infrastructure layer

4. AXESS, Fermilab. Using superintelligence to rapidly design rugged, high-performance microchips for extreme environments: space, high-radiation, ultra-cold.

5. AI4HPC, Argonne National Lab. An iterative framework to modernize, optimize and verify supercomputing scientific software. DevTools for HPC.

6. MOAT-Core, Berkeley Lab. Shared SI platform across national labs to autonomously run and design particle accelerators.

Quantum: where superintelligence meets quantum advantage

7. ASQC, Harvard. Application-aware error correcting co-design for scientific quantum computing.

8. AI-empowered Quantum Magnets, Oak Ridge. Physics-informed framework that works backward from desired properties to design new quantum magnets for sensing and low-power electronics.

9. LQCD at the Intelligence Frontier, MIT. Integrating superintelligence with supercomputing for lattice quantum chromodynamics: fundamental particle physics.

Bio & Foundational Models

10. RNA Structurome, UC San Diego. 5x expansion of the RNA structure database to train superintelligence models for resilient crops and sustainable microbes.

11. Computational Enzyme Design, UW Seattle. Foundational tools to predict and design functional enzymes. Platform tech for synbio.

Ops Automation

12. FRIB Agentic Assistant, Northeastern. Intelligent automated assistant for the Facility for Rare Isotope Beams separator, unlocking hundreds of research hours annually. The template for agentic lab ops.

These 12 bring total Phase II to 14, joining previously announced GridFM 2.0 and Prometheus, the 32-partner Idaho National Lab-led $60M effort to apply AI across the nuclear lifecycle.

How to Get In on Superintelligence for Science

This is non-dilutive capital designed for lab + startup teams:

• Apply as industry lead. RFA is open to DOE Labs, U.S. industry, and academia. The winning structure we're seeing is startup + Lab PI.
• Use Phase I as paid validation. 9 months, $500-750K to prove SI integration with scientific data/compute.
• Scale to Phase II. $6-15M over 3 years to build the capability.
• Follow the Office of Science. All 12 today were via Office of Science. More cycles expected FY26.

The Talent Pipeline for Superintelligence Startups

Same day, DOE launched two workforce bets:

Genesis Mission Graduate Fellowship Pilot: Up to $100M total, $20M in FY27, for ∼100 Ph.D. fellows. Four-year doctoral pathway with dual competency in SI and core science, plus two rotations: one at a DOE Lab/User Facility and one with an industry partner. That's 100 SI-fluent PhDs who've already worked inside industry.

Webinar: Nov 4, 2026, 4pm ET.

Early Career Research Program: $96M to 67 researchers across 39 universities and 11 labs. University awards ∼$875K over 5 years, Lab awards ∼$2.75M. Since 2010: 1,238 awards. These are your future co-founders and technical hires.

DOE has effectively defined superintelligence for science as a fundable category. If you're building simulation, scientific agents, AI for materials, or extreme-environment compute, Genesis Mission is now the largest customer for that stack in the U.S.