Summary of Key Points
The U.S. Department of Energy’s “Genesis Program” is a national scientific research initiative aimed at using artificial intelligence (AI) to solve complex scientific problems. The initial round of funding allocated $250 million to 278 interdisciplinary projects, with the potential to expand this amount to $5 billion in the future. The competition for these grants is exceptionally fierce—out of 5,000 applications, only 278 were approved, representing a success rate of 5.6%. The projects focus on the application of AI in areas such as nuclear reactors, battery recycling, and underground ecosystems. However, this has raised concerns among scientists regarding the potential diversion of funding and the high risks associated with the short project timelines.
I. Genesis Program: A National Initiative to Use AI to Tackle Major Challenges
Launched in November 2025, the Genesis Program is led by the Department of Energy with the goal of using AI to overcome 26 “major challenges of this century,” including the design of nuclear reactors and the treatment of soil pollution. An initial investment of $250 million has been made, and the White House has indicated a potential increase to over $5 billion, although the source of additional funding has not been specified. Participating teams must consist of at least two types of entities: universities/research institutions, Department of Energy national laboratories, and private companies—combining the expertise of academia, government research facilities, and industry to tackle these significant issues.
II. How Fierce is the Competition?
This year, the Department of Energy solicited proposals in the spring, giving scientists only six weeks to form teams across disciplines. More than 5,000 applications were received, but only 278 were approved, resulting in a success rate of less than 6%. Most of the projects in the first round will receive funding for the initial phase (ranging from $500,000 to $750,000) and have nine months to complete their work. Some teams applied directly for the second phase, with results expected in the fall.
III. What Are the Selected Projects Doing?
Several examples illustrate the type of problems that AI is being used to solve:
1. Research on Underground Microorganisms: The Olson team from the University of Maine used AI to simulate how underground microorganisms affect water quality, mineral movement, and pollutant distribution. Traditional physical models are inadequate for this purpose due to the dynamic nature of microbial activity. However, the Department of Energy has access to extensive coastal data, which, when combined with AI algorithms, can improve modeling tools for groundwater management and energy infrastructure planning.
2. Battery Metal Recycling: The Ilshad team from a Stanford laboratory used AI to identify chemical substances (ligands) that can efficiently extract metals such as cobalt, nickel, and manganese. AI can also independently search through literature, formulate hypotheses, and design experiments, significantly accelerating the research process.
3. Autonomous Nuclear Reactor Design: The “Prometheus” project, the only one to receive funding for the second phase (amounting to $60 million), involves 32 partners from five national laboratories, four universities, and twenty companies. The goal is to use AI to completely design, build, and operate a nuclear reactor, with companies contributing an additional $200 million to reduce the typical research and development time from fifteen years to half.
IV. Scientists’ Concerns
1. Funding Allocation: Scientists are concerned that funding for less popular fields (such as basic theoretical research) may be impacted due to the diversion of resources to the Genesis Program. Questions have arisen about whether the funds come from machine learning budgets or basic experimental funding.
2. Short Project Timelines and High Failure Rates: The first phase lasts only nine months, with an estimated success rate of 10%. If a team hires graduate students or postdoctoral researchers and does not advance to the second phase, what will become of them?
3. Lack of Clarity in Guidelines: Many teams are unsure about the project start dates and the evaluation criteria after six months, making it difficult to plan ahead.
4. Incompatibility with Basic Research: The director of the Oak Ridge National Laboratory argues that this “fast-track” approach is not suitable for fundamental research, which requires long-term commitment. There is concern that other institutions may adopt similar models, potentially disrupting the scientific research ecosystem.
V. Future Directions: More Funding, but Intensified Competition
The head of the Department of Energy has stated that additional funding will be provided to encourage more applications. However, the competition for the second phase will be even more intense (only one project out of ten will be selected). While the White House aims to increase the total funding to $5 billion, the source of these funds remains unclear. This could affect the program’s long-term sustainability.
In summary, the Genesis Program represents a significant effort by the United States to leverage AI in future technological advancements. It highlights both the potential benefits of AI and the practical challenges associated with resource allocation and risk management in scientific research. The outcomes of these projects may have a profound impact on energy, environmental, and technological development, but the ongoing debates surrounding them are also worth paying attention to.