The DOE Just Opened $293 Million for AI-Driven Science. Here Is How to Compete.

Twenty-four of the world's largest technology companies — Google, Microsoft, Amazon Web Services, OpenAI, Anthropic, Nvidia, xAI, IBM, AMD, Cerebras, and fourteen others — have signed memorandums of understanding with the Department of Energy to build what amounts to a national AI infrastructure for scientific discovery. Google DeepMind is providing accelerated access to frontier AI models across all 17 DOE national laboratories. Anthropic is embedding dedicated teams to build purpose-built AI agents tailored to lab workflows. AWS is co-developing tools with Idaho National Laboratory to compress research timelines that historically took years into months.

This is the Genesis Mission, and on March 18, DOE opened the application window for its largest competitive funding opportunity yet: $293 million for interdisciplinary teams that can use artificial intelligence to tackle more than 20 national science and technology challenges. Phase I applications are due April 28 — five weeks from now. As our news brief reported, DOE is hosting an informational webinar today, March 26, for prospective applicants.

The funding opportunity — DE-FOA-0003612 — is unusual in both its scale and its structure, and the teams that understand what DOE is actually buying will have a substantial advantage over those who treat it as a standard research grant.

What the Genesis Mission Actually Is

Launched by executive order in November 2025, the Genesis Mission is the Trump administration's signature science initiative — an attempt to position the United States as the undisputed global leader in AI-driven scientific research. The premise is that artificial intelligence has reached a capability threshold where it can fundamentally accelerate the pace of scientific discovery, but only if the computational infrastructure, the scientific talent, and the data systems are integrated at national scale.

DOE is the natural home for this effort. The department operates 17 national laboratories employing roughly 70,000 scientists and engineers. It runs the most powerful civilian supercomputers on Earth, including Frontier at Oak Ridge (the first exascale system) and the upcoming El Capitan at Lawrence Livermore. Its Office of Science funds approximately $8.1 billion in basic research annually. No other federal agency has the combination of computing power, multidisciplinary scientific talent, and physical infrastructure that DOE brings to the table.

The Genesis Mission adds a coordination layer that did not exist before. Rather than funding isolated AI research projects across scattered programs, DOE has identified 26 specific national challenges — from developing new nuclear reactor designs to discovering critical mineral substitutes to accelerating drug development — and is asking teams to propose AI-driven approaches that can produce measurable breakthroughs within defined timescales.

Under Secretary for Science Darío Gil, who joined DOE from IBM Research, framed the ambition bluntly: "With these investments we seek breakthrough ideas and novel collaborations leveraging the scientific prowess of our National Laboratories, the private sector, universities, and science philanthropies."

Inside the $293 Million Structure

The funding is split into two phases with fundamentally different purposes.

Phase I ($500,000–$750,000, 9 months) is designed as a feasibility and team-building stage. DOE wants to see whether your proposed AI approach can actually work on a specific challenge, and whether your interdisciplinary team can function as a unit. Think of Phase I as a paid audition. You are not expected to solve a national challenge in nine months. You are expected to demonstrate that your combination of AI methodology, domain expertise, and data access can produce preliminary results compelling enough to justify a multi-million-dollar Phase II investment.

Phase II ($6–$15 million, 3 years) is where the real work happens. Teams that succeed in Phase I become eligible for Phase II competitions in subsequent funding cycles. However — and this is critical — DOE has also opened Phase II to direct applications in this cycle. Teams that already have established AI-science collaborations and can present compelling letters of intent do not have to wait for a Phase I result to compete for the larger awards.

The dual-track structure is deliberate. DOE knows that some of the strongest AI-science teams in the country are already working together through existing collaborations at national labs. Forcing them through a nine-month Phase I when they could be producing results with Phase II funding would be wasteful. At the same time, new teams — particularly those bringing novel AI approaches from industry or emerging academic groups — need the Phase I pathway to prove themselves.

The 20+ Challenge Areas

The funding opportunity spans more than 20 national challenges, clustered around areas where DOE believes AI can produce transformational results:

Energy systems: Nuclear reactor design optimization, grid resilience modeling, fusion energy pathway analysis, advanced battery chemistry discovery, and carbon capture efficiency improvements. DOE's national labs already run some of the most sophisticated energy simulations in the world. The Genesis Mission asks whether AI can compress the design-test-iterate cycle from years to months.

Materials and manufacturing: Critical mineral alternatives, advanced alloy development, semiconductor manufacturing optimization, and additive manufacturing process control. The United States imports more than 50 percent of its supply for 31 of 35 minerals classified as critical. DOE wants AI systems that can predict and validate substitute materials faster than traditional experimental chemistry.

Biotechnology and health: Drug target identification, protein structure prediction beyond what current tools like AlphaFold can handle, synthetic biology pathway design, and pandemic preparedness modeling. This is where the Genesis Mission overlaps with ARPA-H's health-focused portfolio, creating opportunities for researchers who can position their work at the intersection.

Quantum information science: Error correction algorithms, quantum algorithm design for specific scientific problems, and hybrid classical-quantum computing workflows. DOE operates several of the nation's leading quantum computing testbeds and views AI as essential for making near-term quantum systems useful before full fault tolerance arrives.

National security and resilience: Threat detection, supply chain vulnerability analysis, climate impact modeling, and infrastructure resilience simulation.

Who Should Apply — and Who Should Not

The eligibility is broad: interdisciplinary teams from DOE national laboratories, U.S. industry, and academia. But "eligible" and "competitive" are different things.

The teams most likely to win Phase I awards share three characteristics. First, they include at least one member with deep domain expertise in the specific challenge area — not just AI expertise applied generically. DOE reviewers will immediately identify proposals that describe machine learning techniques in search of a problem. Second, they have access to relevant datasets or experimental facilities. AI models are only as good as their training data, and the most competitive proposals will identify specific datasets at national labs or partner institutions that can be used immediately. Third, they demonstrate a clear understanding of what "success" looks like at the nine-month mark — not a vague promise of future breakthroughs, but a concrete deliverable that DOE can evaluate.

Teams that are purely academic with no national lab connection face an uphill battle. The Genesis Mission's design reflects DOE's conviction that the national labs are the right platform for AI-driven science. Proposals that integrate lab resources, computing infrastructure, and existing experimental capabilities will be scored more favorably than those proposing to build parallel systems from scratch.

How This Connects to the Broader Federal AI Landscape

The Genesis Mission's $293 million does not exist in isolation. Congress preserved NSF's $8.75 billion budget for FY2026, including substantial AI research funding through the National AI Research Institutes program. ARPA-H continues issuing milestone-based contracts for health AI applications. The SBIR/STTR reauthorization creates new $30 million Strategic Breakthrough Awards that small businesses with AI capabilities can pursue.

For research teams with AI expertise, the strategic question is not whether federal AI funding exists — it does, in abundance — but which programs match your team's specific capabilities and stage of development. A university research group with a novel AI methodology and preliminary results should target Genesis Mission Phase I. A small company with a working AI product for a specific scientific application should look at SBIR Strategic Breakthrough Awards. A lab with an established AI-science program and multi-institutional partnerships should go straight for Phase II.

The April 28 deadline for Phase I applications and Phase II letters of intent is firm. The informational webinar happening today will clarify challenge-specific requirements that are not fully detailed in the published NOFO. If you are considering an application, the webinar recording will be essential viewing.

For research teams assembling proposals across multiple federal AI funding programs, Granted can help you track deadlines, match your capabilities to the right opportunities, and structure applications that speak directly to what each agency is looking for.