The Genesis Mission: Can the United States’ Bet on AI Revitalize U.S. Science

Critical Questions by Navin Girishankar and Chris Borges

Published December 4, 2025

On November 24, 2025, President Trump signed an executive order launching the Genesis Mission with an audacious goal: to double U.S. scientific productivity in 10 years using AI. The administration frames it as this generation's Manhattan Project-a national mobilization to accelerate scientific discovery, strengthen national security, and restore U.S. technological leadership against the People's Republic of China (PRC). The initiative brings together elements of the administration's AI Action Plan, which includes a goal of revitalizing the U.S. scientific enterprise, and the president's March 2025 letter to his science advisor, Michael Kratsios, which urges a renewal of U.S. capacity for breakthrough innovation.

Q1: What is the Genesis Mission?

A1: The Genesis Mission is a federal initiative to accelerate scientific discovery using AI. It centers on building the American Science and Security Platform-a research engine of sorts that offers integrated IT infrastructure comprising high-performance computing resources, AI modeling frameworks, scientific foundation models, AI-powered tools, and the world's largest collection of scientific datasets built over decades of federal investments. The platform will draw on resources from across the Department of Energy (DOE) and the national labs system to train scientific foundation models and create novel AI systems to test new hypotheses, design experiments, analyze results, and run autonomous research workflows at a speed and scale far beyond what human researchers can achieve alone.

Complementing this technical effort, the executive order directs the Assistant to the President for Science and Technology and the DOE to identify a set of national science and technology challenges across priority domains, in collaboration with the National Science and Technology Council (NSTC). This list will guide how agencies utilize the platform and direct AI-powered research toward areas of national importance. While the administration's stated focused is on grand challenges related to AI, quantum, and chips, it remains to be seen how future priorities for grand challenges will be established-for instance, will they draw from existing lists established in law, such as the NSTC's Critical and Emerging Technologies List, the National Science Foundation (NSF)'s Key Technology Focus Areas, and the Department of Defense's Critical Technology Areas.

While other administration actions have sent mixed signals about the direction of its science strategy and commitment to the scientific community, the Genesis Mission presents a clear statement of its strategic intent. It is a high-conviction bet that AI can serve as an accelerant for curiosity-driven and use-inspired research-one that will ensure American leadership for decades to come. But Genesis is more than that. It is also a big bet on the United States' institutional capacity to coordinate, implement, and sustain a national science mission across the world's most complex and sophisticated science and technology ecosystems. The U.S. system-distributed across federal agencies, universities, national labs, private firms, and state governments-is resilient by design but notoriously difficult to align. Past attempts at coordination have struggled or failed outright. Without addressing this ecosystem challenge, the Genesis Mission risks becoming another well-intentioned initiative that fails to truly transform how the U.S. scientific community-not just the national labs-works.

Q2: Why is the administration launching this mission now?

A2: The Genesis Mission reflects the administration's desire to "accelerate research and development," "revitalize America's science and technology enterprise," and pursue "new paradigms for the research enterprise," as articulated in President Trump's letter to Michael Kratsios. The Genesis Mission's high level of ambition is warranted for two reasons.

First, the PRC is pacing-in some areas outpacing-the United States in scientific publications, advanced research facilities, talent pipelines, and AI-enabled discovery platforms; and advancing in critical domains like battery chemistry, nuclear energy, and quantum computing. China is not going to slow down its scientific efforts: early indications suggest that China's upcoming five-year plan will continue to emphasize scientific advancement, and Beijing continues to invest heavily in STEM education, large-scale research facilities, and AI-driven research platforms.

Second, U.S. research productivity has stalled-as part of a global trend. Research inputs, such as the number of researchers and expenditures, are rising faster than outputs, such as new breakthroughs and productivity growth. The reason for this decline is unclear. A common theory is that scientific discoveries become harder as fields advance-that the "low hanging fruit" has all been picked-and require considerably more sophisticated and expensive tools to achieve. While Galileo could study gravity using household objects, frontier physical sciences research today requires complex instruments such as billion-dollar gravitation wave detectors. Other observers argue that researcher quality has diminished or that artificial boundaries between scientific disciplines-constructs of academic organization rather than natural divisions in knowledge-channel scientists into narrower problems and constrain broader advances. Regardless of the cause, the administration has consistently embraced the view that scientific productivity has stalled and that AI can reverse the trend.

Q3: How can AI accelerate scientific discovery?

A3: The use of AI in scientific research is not new. The very first AI program-the Logic Theorist, created in 1956-was designed to prove mathematical theorems and solve programming problems, and scientists have incorporated computational tools into their work ever since. What has changed is the sophistication of AI systems, which can help researchers generate hypotheses, design experiments, and analyze complex datasets far more quickly than traditional methods allow. These capabilities enable AI to act as a scientific assistant-screening possibilities, narrowing the search space, and identifying the most promising avenues for investigation.

In numerous fields, AI tools are already yielding promising results. Biological models such as DeepMind's AlphaFold solved scientific problems once thought to require years of laboratory work, predicting the shapes of nearly every known protein with remarkable accuracy and opening new pathways for drug discovery. In meteorology, researchers are using AI systems to generate synthetic storms and identify new precursors to tornadoes, as the volume of real-world tornadoes is too low to generate sufficient data for analysis. And in materials science, researchers are experimenting with AI-guided platforms and robotic "self-driving" labs to generate and test new materials autonomously, rapidly iterating through thousands of possibilities to identify promising chemistries. In one example, a robot conducted 688 experiments in an eight-day period, which is more than a typical human PhD student would run in four years.

Thus, the launch of Genesis is well-timed. Scientists are already adopting new scientific discovery tools at scale, with demand expected to grow exponentially. A massive GenAI infrastructure build-out is underway that can provide the much-needed compute power American scientists will need. And the scientific community will need to shape the principles, standards, and governance frameworks that govern how AI is used across the U.S. research ecosystem.

Q4: What are the implementation challenges that can be anticipated?

A4: With an ambitious federal initiative of this kind, it is reasonable to expect-and important to anticipate and manage-implementation hurdles related to technical capacity, coordination, and operations. Navigating these three challenges, described below, will be all the more important given the initiative's aggressive timelines, including a 270-day deadline to demonstrate an initial operating capability.

First, developing scientific foundation models and leveraging state-of-the-art computational tools requires specialized expertise-skills that are scarce across the federal workforce and command lucrative compensation packages in industry. Federal hiring restrictions, political uncertainty, and constrained salaries make it especially challenging for the DOE and national labs to recruit the talent needed for Genesis. The mission's provisions on cultivating a talent pipeline-through fellowships, internships, and apprenticeships across national laboratories and federal research facilities-are a constructive step toward building long-term capacity. But in the near term, the administration will have to depend heavily on the workforce it already has. DOE and the national labs have a long tradition of integrating AI into scientific workflows, yet budget cuts and staff departures have strained its current workforce, complicating efforts to train, upskill, and retain the people needed to carry Genesis forward.

Second, data governance issues should not be underestimated. In practice, federal agencies struggle to share even basic datasets due to statutory limits, security requirements, and differing IT architectures. The American Science and Security Platform depends on aggregating and increasing the accessibility of federal scientific datasets, but the DOE's own laboratories operate on different networks. Additionally, external users of DOE facilities often store their data elsewhere: university-generated research data typically resides on university systems, and many large-scale scientific projects depend on international partners and cross-border data flows. Integrating these sources--along with privately held industry data--requires not just technical interoperability but legal agreements, security protocols, and trust, none of which can be mandated by executive order. Bringing assets into a secure, interoperable platform may require technical standardization, new data-use agreements, and stronger stewardship practices than currently exist across much of the federal data system.

A third is related to resourcing the Genesis Mission: the executive order envisages operational efficiencies to redeploy existing resources-not new funding. That means that progress will depend on the DOE's ability to reorganize and optimize current resources-supercomputing infrastructure, laboratory capabilities, ongoing research programs, and ESnet, the DOE's current high-capacity inter-lab data network-to fulfill its new directives. Change management challenges with winners and losers can be expected: research programs that lose funding to stand up the platform, labs whose operations get disrupted, and scientists whose computational access gets reallocated. These trade-offs will determine whether the research community embraces Genesis or resists it.

Given these constraints within the federal system, the plan under Genesis to partner extensively with the private sector, universities, and international partners is welcome. In fact, the executive order aims to leverage a number of institutional innovations, including those developed under earlier administrations-for instance, new standardized frameworks for cooperative research, data use, and model sharing-and orders agencies to launch coordinated funding opportunities, prize competitions, fellowships, and user-facility partnerships to encourage private-sector participation. Announced private-sector partners include Microsoft, OpenAI, Google, Nvidia, AMD, AWS, and Anthropic, although their precise role in the mission remains to be defined.

As part of the Genesis implementation, and in the spirit of institutional innovation, the DOE should consider creatively deploying existing federal mechanisms in a manner consistent with the public-private partnership approach envisaged under the executive order. These include the use of cooperative research and development agreements, other transaction authorities, and advisory bodies like Federal Advisory Committees. The DOE's Foundation for Energy Security and Innovation-a CHIPS-era 501C3 model-could provide an additional platform for mobilizing industry collaboration and philanthropic support. Together, these mechanisms give the Genesis Mission a broad toolkit, but its ultimate value depends on how well it responds to the strategic motivations driving the initiative.

Q5: How should the Genesis Mission be evaluated?

A5: Genesis's implementation should be evaluated against its dual objectives: research productivity as well as institutional innovation. The challenge is not simply increasing research output at national labs-there will be a learning curve, but all indications are that AI will accelerate discovery. The deeper test is institutional: whether Genesis strengthens the United States' capacity to coordinate, implement, and sustain focus on grand challenges across its federated research ecosystem. The U.S. system-distributed across federal agencies, universities, national labs, private firms, and regional hubs-derives strength from institutional diversity and competition among ideas. But large national missions succeed only when they align with and elevate the productivity of this distributed ecosystem.

Evaluating Genesis's success requires criteria that go beyond technical milestones. When the CSIS Economic Security and Technology Department was invited to share its perspectives on institutional reform at the National Science Board in December 2024, we identified five tests for renovating the national science and technology ecosystem:

• Does it set goals at the right level? National challenges must be ambitious but guided by scientific communities closest to the frontier. Top-down mandates disconnected from researcher priorities will fail. For Genesis, who decides which "national science and technology challenges" get platform access? The NSF funds curiosity-driven research. The National Institutes of Health focuses on health outcomes. The DOE serves energy and national security missions. These actors have overlapping but distinct goals, and no clear mechanism exists for prioritizing among them.
• Does it enable cross-sector collaboration to ensure viability? Are the institutional innovations Genesis aims to deploy-cooperative agreements, public-private partnerships, regional coordination structures-aligning incentives across actors with different timelines, risk tolerances, and success metrics? Are they bankable-that is, do they "stack" public, private, philanthropic, and regional resources to achieve scale?
• Does it integrate national security without stifling openness? Foundation models trained on sensitive federal data could leak dual-use insights. Is Genesis balancing research security and openness-a trade-off the U.S. research system has never managed well? The executive order acknowledges this tension but provides no governance framework.
• Does it create mechanisms for evaluation? Given the magnitude of this effort, are there sufficient eyes looking at the Genesis Mission both in terms of research output but also institutional outcomes, such as improved data sharing, broader participation by universities and small firms, reduced time-to-insight in key domains, and effective use of partnership mechanisms? Resolving many questions-including how to rigorously define and track research output-will require broad engagement from scientific and policy stakeholders.
• Does it support learning and course correction? Can evaluations be undertaken without stifling learning and course corrections? Models will underperform. Does Genesis have built-in mechanisms for shifting resources and updating priorities-something federal bureaucracies struggle with? The flexibility to fail well may matter more than getting the initial design right.

Q6: How can Genesis be understood in a historical context?

A6: When Vannevar Bush wrote Science, The Endless Frontier, in 1945, he faced a similar challenge: coordinating public and private actors to build the United States' postwar scientific enterprise. Bush relied heavily on universities, industry labs, and decentralized funding-a federated model, not centralized command. But he had two advantages Genesis lacks: wartime urgency that dissolved institutional resistance, and an opportunity to build new institutions rather than reform entrenched ones. The United States has not successfully coordinated its research ecosystem around a peacetime national mission in decades. The challenge today is whether American pluralism-creativity, diversity, decentralized experimentation-can achieve at scale what China's centralized system does through command.

Genesis is an institutional experiment: a test of whether the United States can upgrade its federated research ecosystem around a national mission with constrained resources and without emergency powers. The exemplars that inspired Genesis, such as the Manhattan Project and Apollo missions, required the development of new technical capabilities, institutional structures, and educational infrastructures on timescales measured in years or decades, not months. As White House officials acknowledge, that means broad national ownership and bipartisan congressional support for funding systemic improvements that Genesis and similar efforts deliver.

If the Genesis Mission succeeds, it will prove that U.S. institutions can evolve as fast as U.S. technology. If it fails, it will reveal that invoking the Manhattan Project is not a substitute for the hard work of institutional reform. Bush's generation built the institutions that made the United States the global science leader. This generation's challenge is adapting them for an age of AI-powered discovery and strategic competition.

Navin Girishankar is president of the Economic Security and Technology Department at the Center for Strategic and International Studies (CSIS) in Washington, D.C. Chris Borges is a senior program manager and associate fellow with the Economics Program and Scholl Chair in International Business at CSIS.

Programs & Projects