The U.S. government is betting big on frontier AI in US science — and OpenAI is now at the center of that bet. Through a formal collaboration with the U.S. Department of Energy‘s Genesis Mission, the company is channeling tens of millions of dollars worth of AI resources into the nation’s network of National Laboratories and universities, with a mandate that is as ambitious as it is specific: double the productivity and impact of American scientific research within a decade.
Summary
Key takeaways
- OpenAI is partnering with the U.S. Department of Energy’s Genesis Mission to bring frontier AI capabilities into National Laboratories and universities.
- OpenAI will provide $4 million in Codex access to approximately 2,000 Genesis researchers and commit $3 million in API support to two large-scale scientific campaigns.
- Researchers can receive up to $10 million in API usage credits for $2.5 million spent, dramatically extending their research capacity.
- Specialized tools include GPT-Rosalind for bioscience projects and advanced AI cyber capabilities for national-lab cybersecurity researchers.
- OpenAI has already deployed reasoning models on Venado, Los Alamos National Laboratory’s supercomputer, as a shared resource across National Nuclear Security Administration labs.
OpenAI and the Genesis Mission: What This Partnership Actually Does
The Genesis Mission, run by the Department of Energy, brings together DOE’s 17 National Laboratories, universities, and industry partners with a single directive: integrate AI into the full machinery of American science. That means federal scientific data, advanced computing, experimental facilities, and expert teams — all operating under the assumption that AI can compress what normally takes decades into something far shorter.
OpenAI’s role in this effort is neither advisory nor symbolic. The company is putting real resources on the table. For starters, it will provide approximately $4 million in Codex access to around 2,000 Genesis researchers spread across National Laboratories and universities. That access is designed to bring advanced coding and reasoning capabilities into the daily workflows of working scientists — not as a pilot program, but as standard infrastructure.
On top of that, OpenAI is committing $3 million in API support to two large-scale scientific campaigns that will test frontier models against some of the hardest open problems in science. And for researchers who want to scale further, the company is offering a compelling incentive: up to $10 million in API usage credits for every $2.5 million spent, a leverage ratio that meaningfully extends the reach of any research budget.
Specialized AI Tools for Bioscience and Cybersecurity
Not every researcher in the Genesis network will receive the same tools. OpenAI is tiering access based on specific needs — and for some, that means tools built for their domain.
Selected national-laboratory researchers working on eligible biology projects will gain access to GPT-Rosalind, a specialized model designed for bioscience applications. The rollout follows joint work between OpenAI and Los Alamos National Laboratory to develop evaluations for how multimodal AI systems can be used safely in realistic laboratory environments — work that is as much about understanding the risks of capable AI in consequential research as it is about unlocking its benefits.
Separately, national-lab cybersecurity researchers will receive expanded trusted access to advanced AI cyber capabilities. The goal is to strengthen defensive research and protect critical systems — a mandate that carries particular weight given the role these laboratories play in national security, including through the National Nuclear Security Administration.
OpenAI is also giving trusted national-laboratory leaders early access to selected models and features before broader rollout, allowing them to build the workflows, evaluations, and technical infrastructure needed to deploy AI effectively — rather than scrambling to adapt after the fact.
Two Scientific Campaigns That Could Define What AI Can Do for Research
High-temperature superconductors
The first large-scale campaign targets one of materials science’s most persistent open problems: high-temperature superconductors. The campaign would combine frontier AI models with simulation tools, materials science expertise, and experimental data to pursue superconducting materials capable of operating at higher temperatures and practical pressures. The potential applications stretch across energy transmission, transportation, medicine, scientific instrumentation, and national security — sectors where superconducting breakthroughs have long been anticipated but repeatedly delayed.
Mapping what AI can already reach
The second campaign takes a different approach. Rather than chasing a specific scientific outcome, the so-called Atlas of the Machine-Accessible Frontier would systematically examine where AI can already support meaningful scientific advances using existing knowledge, data, and computation. The intent is to help researchers and policymakers distinguish problems that are now tractable from those that still require new physical evidence — a kind of honest reckoning with the current capabilities of AI in science.
Together, these campaigns are designed to establish repeatable methods for applying AI to difficult national scientific challenges, which may matter as much in the long run as any individual breakthrough they produce.
AI Already Running on America’s Supercomputers
The partnership is not starting from zero. OpenAI has already deployed advanced reasoning models on Venado, the supercomputer at Los Alamos National Laboratory. Venado serves as a shared resource across National Nuclear Security Administration laboratories, meaning the deployment has implications well beyond a single facility.
Earlier groundwork also included an AI Jam Session with nine National Laboratories, where more than 1,000 scientists tested frontier models against domain-specific research problems and provided structured feedback. That feedback loop — scientists pushing models, models being refined — is part of what OpenAI describes as turning AI from a tool into something closer to national strategic scientific infrastructure.
That framing is worth taking seriously. Infrastructure implies permanence, shared access, and public benefit. It also implies a level of dependency that shapes policy choices for years. The fact that OpenAI is positioning its models as infrastructure — not software-as-a-service — signals how deeply embedded this collaboration is meant to become.
Why This Moment in US Science Policy Matters
The Genesis Mission’s stated goal — doubling American scientific productivity within a decade — is the kind of target that is easy to announce and difficult to measure. But the resources being mobilized around it are concrete, and the breadth of the partnership is significant. DOE’s 17 National Laboratories collectively represent an extraordinary concentration of scientific infrastructure, from nuclear physics to climate modeling to materials research. If frontier AI tools genuinely accelerate work at that scale, the downstream effects on American economic and technological competitiveness would be substantial.
At the same time, the collaboration raises questions that the partnership itself does not fully resolve. The integration of a single company’s AI models into the core workflows of federally funded research creates dependencies that will be difficult to unwind. How research outputs, data, and model evaluations are governed under this arrangement — and who ultimately owns the insights generated — are questions that will matter more as the collaboration deepens. The current framing emphasizes access and capability; the harder institutional questions tend to emerge later.
FAQ
What is the goal of the Genesis Mission?
The Genesis Mission, run by the U.S. Department of Energy, aims to double the productivity and impact of American scientific research within a decade by integrating AI with federal scientific data, advanced computing facilities, and expert research teams across DOE’s 17 National Laboratories, universities, and industry.
How is OpenAI supporting researchers through the Genesis Mission?
OpenAI is providing $4 million in Codex access to approximately 2,000 Genesis researchers at National Laboratories and universities, committing $3 million in API support for two large-scale scientific campaigns, offering up to $10 million in API usage credits for $2.5 million spent, and granting selected researchers access to specialized tools including GPT-Rosalind for bioscience applications.
What are the focus areas of the large-scale scientific campaigns supported by OpenAI?
The two initial campaigns focus on breakthroughs in high-temperature superconductors — combining frontier AI, simulation, and experimental data to pursue superconducting materials at higher temperatures and practical pressures — and on building an Atlas of the Machine-Accessible Frontier, which maps where AI can already support meaningful scientific advances using existing knowledge and computation.
How is OpenAI deploying AI models in national laboratories?
OpenAI has deployed advanced reasoning models on Venado, the supercomputer at Los Alamos National Laboratory, as a shared resource across National Nuclear Security Administration laboratories. The company has also conducted an AI Jam Session with nine National Laboratories, where more than 1,000 scientists tested frontier models on domain-specific problems.
Article produced with the assistance of artificial intelligence and reviewed by the editorial team.
