OpenAI and U.S. Department of Energy Expand AI Collaboration

OpenAI and the U.S. Department of Energy (DOE) have signed a memorandum of understanding (MOU) to explore collaborations on AI and advanced computing. This partnership aims to accelerate scientific discovery by integrating frontier AI models into real research environments, specifically supporting DOE initiatives such as the Genesis Mission.

Framework for AI-Driven Scientific Discovery

The MOU establishes a formal framework for information sharing and coordination between OpenAI and the DOE. This agreement creates a structured path for the two parties to develop specific follow-on agreements as individual projects take shape, ensuring that future work is clearly scoped and governed.

Key objectives of this collaboration include:

  • Advancing Basic and Applied Research: Strengthening U.S. leadership in AI and advanced computing.
  • Technical Expertise Exchange: Coordinating activities to apply AI to high-impact areas, such as fusion energy, leveraging the DOE's world-leading facilities, data, and modeling tools.
  • Integration with the Genesis Mission: Supporting the government-industry effort to apply advanced computing to accelerate scientific progress.

Integration into National Laboratory Workflows

This partnership builds upon existing deployments of OpenAI models within the DOE's national laboratory system. OpenAI's strategy focuses on working side-by-side with domain experts to understand where AI provides value and how it can be safely integrated into scientific workflows.

Deployment on Supercomputing Infrastructure

OpenAI has deployed advanced reasoning models on the Venado supercomputer at Los Alamos National Laboratory. This deployment serves as a shared resource for researchers across the National Nuclear Security Administration (NNSA) laboratories, including Lawrence Livermore National Laboratory and Sandia National Laboratories, to address complex scientific and national research challenges.

Large-Scale Scientist Feedback

To inform system development, OpenAI convened the "1,000 Scientist AI Jam Session." This event involved over 1,000 scientists across nine labs who used frontier AI models to test domain-specific problems and provide structured feedback on model responses.

Bioscience Safety Evaluations

In partnership with Los Alamos National Laboratory, OpenAI has developed evaluations to study the safe use of multimodal AI systems in laboratory settings. These evaluations move beyond text-based assessments to measure how models influence outcomes in high-consequence domains, emphasizing risk reduction and expert oversight.

The OpenAI for Science Strategy

This collaboration is a core component of "OpenAI for Science," an initiative designed to accelerate the transition from insight to validated results. The strategy is based on the synergy between two critical components:

  1. Scientific Tools: The necessity of simulation engines, analysis pipelines, domain databases, and high-performance computing for precision and throughput.
  2. Frontier Reasoning: The ability of scaling models to support conceptual work, such as connecting ideas across disparate fields, navigating extensive literature, proposing mechanisms, and stress-testing hypotheses.

Strategic Policy Recommendations

Concurrent with the MOU, OpenAI submitted recommendations to the White House Office of Science and Technology Policy (OSTP). This filing advocates for the United States to strengthen its science and technology leadership through AI, identifying 2026 as a "Year of Science" and emphasizing that access to compute, frontier models, and real research environments is essential for accelerating discovery.

Sources