OpenAI и Минэнерго США объединяются: frontier-ИИ для ускорения научных открытий

OpenAI has announced a collaboration with the U.S. Department of Energy (DOE) and its national laboratories: the goal of the partnership is to apply frontier artificial intelligence models to accelerate scientific research and discovery.

What OpenAI announced

OpenAI is committing to advancing a new generation of American science by working directly with the U.S. Department of Energy and its network of national laboratories. The focus is on applying the company's most powerful models to real research problems — from fundamental physics to energy and materials science. The company describes the move as a contribution to "the next era of national science" — that's the name it gave its announcement.

The Department of Energy oversees one of the country's largest scientific infrastructures. According to the DOE, it manages a network of 17 national laboratories — among them Los Alamos, Oak Ridge, Argonne, and Lawrence Berkeley, where research is conducted in high-energy physics, nuclear security, climate, and computing.

Key facts of the announcement:

• Partners — OpenAI, the U.S. Department of Energy (DOE), and the national laboratories
• Tool — frontier AI models, the most advanced models of the latest generation
• Goal — accelerating scientific research and discovery
• Infrastructure — the DOE manages a network of 17 U.S. national laboratories

Why the national laboratories need AI

The national laboratories need AI to compress research cycles that today take years. Frontier models help sift through hypotheses, simulate experiments, and analyze massive datasets faster than research teams can do manually. In practice, this means faster screening of battery materials, the search for stable catalysts, or calculating scenarios for fusion reactors.

The DOE's national laboratories trace their history back to the 1940s — Los Alamos was founded in 1943 as part of the Manhattan Project. Today their scope is far broader: nuclear fusion, developing new materials and batteries, climate modeling, genomics, nuclear security.

The DOE's national laboratories are also among the world's largest computing centers. The Frontier supercomputer at Oak Ridge National Laboratory became the world's first exascale-class system in 2022 — over a quintillion operations per second. Combined with frontier AI models, such computing power creates the "big compute plus big models" pairing that the partnership is betting on.

Part of OpenAI's strategy in science

OpenAI positions accelerating scientific discovery as one of the main goals of its technology. The partnership with the federal Department of Energy moves this line of work from the realm of demonstrations into applied work with U.S. national research infrastructure.

OpenAI reaffirms its commitment to advancing American science by working with the U.S.

Department of Energy and the national laboratories to apply frontier AI to accelerate discovery, — as the company frames the essence of the initiative in its blog.

The bet isn't unfounded: AI has already rewritten entire areas of science. DeepMind's AlphaFold predicted the structure of more than 200 million proteins and in 2024 earned its developers the Nobel Prize in Chemistry. OpenAI and the DOE are counting on repeating such a breakthrough in physics, energy, and materials science. For the government, this is a way to connect the most powerful commercial models to national-level tasks; for OpenAI, it's access to the unique scientific data and computing power of the national laboratories.

What this means

Frontier AI is moving beyond consumer products and chatbots into the realm of big science: OpenAI and the U.S. Department of Energy are betting that frontier models will accelerate discoveries in energy, materials, and fundamental physics. Specific projects and timelines were not disclosed in the announcement — but the very fact of a direct partnership between a leading AI lab and the national scientific system marks a new stage in AI application. For the industry, it's a signal: governments are increasingly seeing frontier AI as a tool of scientific and technological sovereignty.