AI-for-Science: A New Kind of Manhattan Project

I’m Ylli Bajraktari, CEO of the Special Competitive Studies Project. In this week’s edition of our newsletter, SCSP’s Nyah Stewart and P.J. Maykish discuss the need for a national AI-for-Science platform. This newsletter is adapted from SCSP’s answer to a Department of Energy open request for information regarding AI for science last year.

---

The Case for a National AI-for-Science Platform

Imagine a scientific revolution where breakthroughs occur not in decades, but in days. Artificial intelligence has already proven it can accelerate discovery in targeted areas such as materials science and drug development. But today, those successes are scattered—driven by private firms and research groups working in isolation.

To transform these fragmented efforts into a coherent national capability, the United States must establish a national program that builds a comprehensive AI-for-Science platform that would link our national laboratories, supercomputers, datasets, and private-sector ingenuity into a unified engine for innovation. This AI-for-Science capability would not serve a single entity, but instead be a tool for the entire U.S. research ecosystem, and, at a fundamental level, completely transform the way science is done.

Note: The key differences of the AI-for-Science platform compared to current AI capabilities are its connection to experimental facilities. This connection allows the software stack to utilize data generated by physical experiments, rather than relying solely on training data, and move past simulation-based approaches. As a result, the software stack can make observations and generate insights about real-world phenomena, not just simulated environments. Additionally, the connections to experimental facilities give the AI models the ability to directly interface with and control physical scientific equipment, like DNA synthesis and protein construction, helping to break down the barrier between the cyber and physical domains.

The potential returns are extraordinary. The biomedical sector alone demonstrates the scale of opportunity: NIH-funded R&D generates nearly $95 billion in annual economic activity. A cross-domain AI-for-Science platform could multiply that impact, advancing not only our prosperity but also our strategic power.

The Global Stakes are High.

The competition has already begun. Earlier this year, the Chinese Academy of Sciences unveiled ScienceOne, a state-backed AI model designed to speed discovery and consolidate global scientific influence. It is a warning sign and a call to action.

Without decisive U.S. leadership, China’s command-driven innovation model could define the frontier of advanced research. America’s open, democratic innovation ecosystem—our universities, labs, and private sector—remains our greatest strength, but it requires national coordination and investment equal to the scale of the challenge.

This is our generation’s “Manhattan Project for the AI Era.”

America’s Strategic Edge is Our Scientific Enterprise.

No other nation possesses the infrastructure, partnerships, and scientific culture that the United States does. To lead this new era, we must leverage one of our unique strengths: the Department of Energy.

DOE is positioned to spearhead this historic nationwide effort to create an AI-for-Science platform due to its:

• Unmatched Infrastructure: The DOE’s network of 17 national laboratories is the world’s most concentrated ecosystem of data, instruments, and multidisciplinary expertise—a foundation no competitor can replicate.

• Leading-Edge Resources: DOE operates some of the planet’s most powerful supercomputers and is already planning co-located AI data centers that merge public datasets with private innovation, multiplying discovery power.

• A Proven Record of Big Science: From splitting the atom to decoding the genome, the Department of Energy has executed nation-altering projects that transformed geopolitics and expanded human capability.

Innovation Power Is National Power.

Artificial intelligence is redefining not only how we innovate, but who leads. As SCSP has written, innovation power will be the decisive metric of national strength in the years ahead. It combines speed, scale, and strategic intent—the ability to translate discovery into economic vitality, military advantage, and geopolitical influence.

This Capability Will Shape the Next American Century.

For the United States, this means acting now. Congress should fully fund, and the government build, a whole-of-nation effort—anchored in public-private collaboration, led by the DOE, and unified by a single mission. This is the AI-for-Science platform that will define the American century and the next giant leap for mankind.