The U.S. Department of Energy announced 278 research projects selected under the Genesis Mission at the inaugural Genesis Mission Summit in Washington, D.C., on Wednesday — the largest response to any DOE funding opportunity in the department's history — while the White House simultaneously committed more than five billion dollars in federal resources to expand the program across 15 or more agencies. The combined announcement, which drew senior officials from HHS, NASA, NSF, the Department of Transportation, the Department of the Interior, NIST, and the newly renamed Department of War, establishes the Genesis Mission as the largest coordinated deployment of artificial intelligence into American scientific infrastructure ever attempted.

The 278 selected projects span all 50 states and involve 342 participating institutions — 142 universities, 157 companies, 16 DOE and National Nuclear Security Administration national laboratories, 13 nonprofits, and 14 other institutions. In total, more than 5,000 applications arrived in response to the agency's March 2026 Request for Applications — a figure the DOE described as without precedent for any single funding call it has issued.

"America has no shortage of bold ideas or talented scientists, and the response to the Genesis Mission proves that," Secretary of Energy Chris Wright said at the Summit. "The 278 projects selected today represent the very best of our nation's scientific enterprise."

Genesis Mission traces its origin to Executive Order 14363, signed by President Trump on November 24, 2025, with the stated goal of doubling U.S. scientific productivity through AI-enabled workflows. What began as a DOE-led initiative at that point has since expanded into something considerably larger. More than 15 federal agencies will now contribute research awards, specialized scientific datasets, and research facilities to a set of mission-driven National Science and Technology Challenges covering healthcare, energy, infrastructure, manufacturing, and national security.

The total anticipated funding for the current RFA is $293.76 million, with $250 million in the current available tranche. That figure is separate from — and in addition to — more than $500 million in financial contributions reported Wednesday from Genesis Mission Consortium members in the form of compute access, AI model tokens, and software. The White House's $5 billion-plus commitment covers new funding commitments across 15 agencies going forward, and spans a much wider scope than the initial DOE RFA alone.

"America's greatest scientific achievements have been born of national mobilizations paired with the construction of new institutions," OSTP Director Michael Kratsios said at the Summit. "The Genesis Mission is the next chapter in that tradition, as we work together to harness AI for scientific discovery at a scale no one else on the planet can match."

The portfolio is deliberately broad. Of the 278 awards, 168 are led by universities, 87 by DOE and NNSA national laboratories, 19 by companies, and four by nonprofit organizations.

The single largest individual award goes to the Prometheus project at Idaho National Laboratory: a three-year, $60 million Phase II investment in nuclear energy that will apply AI to design, license, manufacture, construct, and operate nuclear reactors through human-in-the-loop workflows, while also integrating AI into nuclear fuel fabrication and legacy document management. The Prometheus team has raised more than $200 million in industry cost share and $30 million in industry capital alongside the federal award.

Other focus areas in the portfolio include critical mineral extraction, intelligent chip design, commercial fusion energy, autonomous laboratory systems, and grid optimization.

"Through the Genesis Mission, we are bringing together the nation's leading researchers, institutions, and technology partners to build the next generation of scientific capability," Under Secretary for Science Dr. Darío Gil said at the Summit.

Awardees do not merely receive funding — they gain access to the Genesis Mission Platform, a shared infrastructure the DOE has been building since late 2025. That platform sits within the broader American Science and Security Platform — the DOE's overarching infrastructure framework for the Genesis Mission — and has two formally named components: the American Science Cloud, the high-performance computing infrastructure program managed by Oak Ridge National Laboratory; and the Transformational AI Models Consortium, known as ModCon, the collection of AI models managed by Argonne National Laboratory.

The hardware underpinning those managed components is substantial. At Argonne National Laboratory in Illinois, Nvidia and Oracle deployed a cluster called Equinox — approximately 10,000 Nvidia Blackwell GPUs, targeting 2026 operations. At Oak Ridge National Laboratory in Tennessee, AMD and HPE built a comparably sized cluster called Lux, also targeting 2026 operations, running on the HPE ProLiant Compute XD685 platform with AMD Instinct MI355X GPUs, EPYC CPUs, and Pensando data processing units to accelerate data access.

HPE confirmed Wednesday that it has been selected for multiple Genesis Mission research and development projects. The company described Lux as the first dedicated AI system for science within the program, built in collaboration with AMD and Oak Ridge, and targeting critical lighthouse problems in fusion, fission, materials science, quantum computing, advanced manufacturing, and grid technology.

Looking further ahead, a 100,000-GPU cluster called Solstice — powered by Nvidia Blackwell GPUs, hosted on Oracle Cloud Infrastructure, and drawing 200 megawatts of continuous power — is planned for Argonne in 2027, which would make it the largest science-oriented AI cluster in the world at that time. A longer-horizon system called Discovery, built by HPE on the Cray GX5000 platform with AMD Venice EPYC CPUs and AMD Instinct MI430X GPUs, is slated for delivery at Oak Ridge in 2028.

The computational architecture of the Genesis Mission represents a structural first in American federal science. Prior DOE high-performance computing programs — including Frontier, Aurora, and Perlmutter — ran on DOE-designed software stacks and were managed primarily by the national laboratories themselves. The Genesis Mission clusters at Argonne and Oak Ridge run on Oracle Cloud Infrastructure software stacks. Jack Dongarra, a long-standing HPC expert, noted the changed arrangement directly: Oracle holds the majority of compute cycles on the new machines, providing a portion of those cycles back to DOE.

That arrangement is not a criticism of the program — the commercial partnership model is what enabled the DOE to bring Lux and Equinox online at what Darío Gil described as an "unprecedented" speed relative to any prior government supercomputing deployment. But it is the question that will define long-term access equity: which researchers, at which institutions, get how many cycles on a platform whose compute allocation is managed in part by a private company, and under what terms. The answer will matter considerably more when Solstice — at 100,000 GPUs, ten times larger — comes online in 2027.

The most immediately concrete demonstration of how the Genesis Mission Platform changes scientific workflows involves not the large-scale GPU clusters but an already-running project using publicly available foundation models from Meta.

SYNAPS-I — SYnergistic Neutron And Photon Science–Intelligence — is a multi-lab initiative led by Berkeley Lab senior scientist Alexander Hexemer, partnering with SLAC National Accelerator Laboratory, Argonne, Brookhaven National Laboratory, and Oak Ridge. The project's goal is to transform data analysis at X-ray and neutron science facilities from a months-long bottleneck into a near-real-time discovery engine. At its core are two open-source foundation models released by Meta: Segment Anything Model 3 (SAM 3), an interactive segmentation model, and DINOv3, a self-supervised vision model that learns visual patterns from raw images without requiring human labeling.

A grapevine drought-resilience demonstration at Berkeley Lab's Advanced Light Source showed what that means in practice. The SYNAPS-I pipeline applied SAM 3 and DINOv3 to micro-CT scans of vine stems to automatically reconstruct three-dimensional volumes and identify xylem vessels — the microscopic tubes responsible for water transport — tracking how they degrade under drought stress. What previously required approximately one month of expert annotation per time step now takes roughly 15 minutes. The same pipeline, once trained and validated, is designed to be deployed across all seven DOE Basic Energy Sciences user facilities participating in SYNAPS-I, covering X-ray and neutron instruments at the Advanced Light Source, the Advanced Photon Source, Brookhaven's National Synchrotron Light Source II, the Linac Coherent Light Source, the Stanford Synchrotron Radiation Lightsource, and Oak Ridge's Spallation Neutron Source.

The structural significance of the SYNAPS-I architecture goes beyond speed. The DOE's X-ray and neutron facilities collectively serve thousands of researchers annually who upload samples, collect data, and then face months-long queues for expert analysis. By embedding an AI segmentation pipeline into the facility's data infrastructure — rather than requiring each researcher to build their own analysis workflow — SYNAPS-I shifts the bottleneck from human annotation capacity to data-acquisition rate. That is the transition the Genesis Mission Platform is designed to produce at scale across all 278 funded projects.

Alongside the DOE awards, the White House unveiled mission-driven National Science and Technology Challenges spanning five broad categories — and in each case, the Genesis Mission Platform's infrastructure is the enabling layer.

In health, HHS will pair the nation's longitudinal health cohorts with DOE supercomputing to find root causes of chronic disease, unlock cures for pediatric cancer, and accelerate drug discovery. The Department of Veterans Affairs will connect electronic health records and genomic data from the Million Veteran Program — already comprising one million sequenced veteran genomes, stored at Argonne — with DOE supercomputing to detect disease risk earlier.

NASA and DOE will apply physics-informed and agentic AI to analyze more than 150 petabytes of data from telescopes, orbiters, landers, and satellites. The Department of Transportation and DOE will create digital twins and foundation models to accelerate infrastructure design, structural simulation, and predictive maintenance. The Department of War and NNSA will pair validated modeling and simulation with agentic workflows to compress the design, testing, and certification of conventional and nuclear weapon components.

Supporters of the program frame it as a national mobilization on par with the scientific investments that produced the internet and GPS. Critics have raised concerns that deserve a clear-eyed accounting alongside the ambition.

Asad Ramzanali, a former Chief of Staff and Deputy Director for Strategy at the White House Office of Science and Technology Policy during the Biden administration and now director of AI and technology policy at the Vanderbilt Policy Accelerator, welcomed the investment but flagged the policy context in which it arrives. The administration proposing $293 million in Genesis Mission grants is the same administration whose FY2027 budget requests a 55% cut to NSF, reducing it from approximately $8.8 billion to $4 billion, according to TechPolicy.Press. The concern, as Ramzanali put it, is that the Genesis Mission is a targeted acceleration sitting alongside a broader reduction of the foundational science funding base it depends on.

Beth Simone Noveck, writing for RebootDemocracy in December 2025, raised a structural governance concern: the Genesis Mission consolidates unprecedented research power in the DOE and White House, but creates no mechanism for public participation, no Federal Advisory Committee oversight, and no process for assessing social impacts. The Information Technology and Innovation Foundation, in analysis published this month, added a focus concern: "An overly broad Genesis Mission that fails to deliver will lower overall long-term support for AI at DOE." The foundation's recommendation is quantifiable metrics and congressional oversight report language.

It is worth noting that the Genesis Mission's own design incorporates some of this caution: the "human-in-the-loop" workflows specified in the Prometheus nuclear energy award, for example, are explicit engineering requirements, not aspirational language.

For the technology industry, the Genesis Mission is a live production deployment of frontier AI at a scale no private-sector organization has attempted in a research context. The program does not merely fund AI research — it deploys AI into production scientific workflows, embedding foundation models from industry partners directly into national laboratory infrastructure used by hundreds of institutions simultaneously.

By structuring the Genesis Mission Platform as shared, multi-tenant infrastructure managed at the national laboratory level — Oak Ridge running the American Science Cloud, Argonne running the Transformational AI Models Consortium — the DOE created a public equivalent of an enterprise AI platform. One that can be iterated on, benchmarked against real scientific throughput, and expanded as each generation of funded projects demonstrates what actually moves the needle. The SYNAPS-I example — 15 minutes where a month once passed — is the kind of concrete, measurable result the program needs to produce consistently, at scale, across all 278 projects, to justify the infrastructure investment and the governance questions that come with it.

With more than 340 institutions now in the portfolio, a $293.76 million grants program backed by a $5 billion-plus federal commitment, over $500 million in private consortium contributions, and a stated ambition to double the country's scientific productivity within a decade, Wednesday's Summit marks the start of a deployment phase — not a planning phase. The next test is whether 15-minute results scale to 15 petabytes.

The Genesis Mission is a White House-led, DOE-executed national initiative to embed AI into American scientific research workflows at scale. It was launched by Executive Order 14363 in November 2025 and is directed by DOE Under Secretary for Science Darío Gil. The program unites the DOE's 17 national laboratories with 142 universities, 157 companies, and other partner institutions. The broader national effort is coordinated by OSTP Director Michael Kratsios and now spans more than 15 federal agencies.

Awardees gain access to a shared infrastructure with two formally managed components: the American Science Cloud — the high-performance computing infrastructure program run by Oak Ridge National Laboratory — and the Transformational AI Models Consortium (ModCon), the AI model library run by Argonne National Laboratory. That infrastructure includes AI agent frameworks, advanced models contributed by industry partners (including Meta, Google, OpenAI, and others), and compute access to national lab GPU clusters including Lux at Oak Ridge and Equinox at Argonne.

This is the most structurally novel aspect of the program. Unlike prior DOE supercomputers such as Frontier or Aurora, the Genesis Mission clusters at Argonne and Oak Ridge run on Oracle Cloud Infrastructure software stacks. HPC expert Jack Dongarra has noted publicly that Oracle holds the majority of compute cycles on these machines, providing a portion back to DOE. What this means for independent researchers — particularly those at smaller universities without existing Oracle relationships — in terms of access allocation and priority is not yet publicly specified, and will become increasingly consequential as the larger 100,000-GPU Solstice system comes online in 2027.

Concretely: faster drug discovery pipelines (HHS challenge), earlier disease detection for veterans (VA challenge), longer-lasting and cheaper infrastructure (DOT/DOE challenge), improved grid reliability (DOE grid challenge), and new materials for energy and manufacturing (DOE materials challenge). The SYNAPS-I project's reduction of a one-month scientific analysis bottleneck to 15 minutes is a small-scale example of the kind of compounding effect the program is designed to produce across hundreds of scientific domains simultaneously. Whether it delivers on that ambition depends on whether the 278 selected projects can demonstrate measurable results against the program's 270-day initial demonstration timeline, which the original Executive Order set at approximately August 21, 2026.