Lab-Directed R&D at the National Labs

Overview

The Department of Energy’s (DOE) Laboratory Directed Research and Development funding model, or LDRD, offers a valuable test case of a decentralized funding mechanism that encourages high-risk, high-reward science. LDRD empowers its national laboratories to act as independent investors, allocating capital where emerging ideas show promise. These early investments often generate the technical validation and momentum needed to unlock follow-on funding from DOE headquarters. This model has repeatedly proven prescient: LDRD grants funded early-stage work that led to the Nobel Prize–winning discovery of the molecular mechanisms of CRISPR, for example. Today, LDRD grants back new technology platforms like autonomous laboratories. By delegating funding to the institutions closest to the research frontier, the LDRD model shows how decentralized funding systems can unlock high-risk, high-reward frontier science.

What are the Department of Energy’s national labs?

The Department of Energy’s (DOE) national labs, often called “the crown jewels” of US science, trace their origins to the Manhattan Project. Today, the labs conduct frontier research and operate world-class scientific user facilities. These facilities — including synchrotrons, molecular foundries, and the world’s largest supercomputers — are capital-intensive scientific assets that are available for use by the broader scientific community. In contrast to the National Science Foundation (NSF) and universities, which focus on curiosity-driven research and training, the national labs are more mission-driven. ¹ For example, Fermi Lab focuses on high-energy particle physics, while the Princeton Plasma Physics Laboratory advances plasma physics and nuclear fusion. 

Of the 17 national labs, 10 are under DOE’s Office of Science, 3 are under the National Nuclear Security Administration (NNSA), and the others are each under a different DOE office. The Office of Science is the largest federal funder of basic research in the physical sciences, including physics, materials science, computing, and chemistry. The Office of Science’s annual budget of around $8 billion is distributed in grants — not just to the national labs but also to universities and companies. ² NNSA’s mission is to enhance national security through nuclear science; its three national labs participate in both classified and unclassified research.

The Office of Science and NNSA national labs are Federally Funded Research and Development Centers (FFRDCs) and compete for federal awards to conduct scientific research. They are operated by the private sector under management and operating (M&O) contracts. M&O contracts allow for unique flexibility in moving large amounts of capital quickly, and M&O contractors, i.e. national lab employees, can be easily detailed to federal agencies to provide their expertise.

Like grants to universities, awards to national labs include both the direct project costs and indirect costs for overhead for the lab. A portion of this overhead feeds into a bucket known as Lab-Directed R&D (LDRD). LDRD funds are pooled together by each national lab and serve as an annual, internal R&D fund for national labs to allocate towards promising, internal, early-stage projects.

The LDRD model has been a first funder for some of the most important novel science in a generation, including research on CRISPR, which won the 2020 Nobel Prize. What attributes of this funding model have generated so much novel science?

Lab-Directed R&D as decentralized science funding

LDRD embodies a distinctive model of decentralized science funding. In contrast to LDRD, most federal research grants are administered by program managers in federal agencies. In some agencies, awards are selected based on how well they score under a peer review process — either under a general call or within specific topic or mission areas, while in others, like at DARPA, capital can move quickly in research directions defined by the program manager. ³ While this federal agency or program manager-centric approach allows science funding to draw from immense federal funding and align with congressionally determined national priority areas, it may be structurally disincentivized from risk-taking or, because of its national overview, miss earlier-stage ideas that are still developing in different institutions or schools.

By contrast, the LDRD model reflects a recognition that no single program officer can fully anticipate where breakthroughs will emerge. Rather than centralizing strategic direction with DOE program managers, LDRD empowers each national lab director to place early bets on promising, unproven ideas within their national lab ecosystem. Distributing decision-making to the institutions closer to the research frontier can enable a more responsive, grounded approach to scientific progress. 

LDRD also allows science funding to move more quickly to early-stage ideas. National labs have far more context on what their own staff are working on than DOE federal program managers may have when reviewing applications from multiple national labs. Further, LDRD funding for internal pilots at national labs generates data that de-risks novel technologies, making them more appealing for downstream funding. Labs that make smart bets often secure follow-on investment from DOE headquarters, creating a feedback loop that rewards sound institutional judgment. This model of decentralized funding, which empowers national labs to act as capital allocators, placing early bets on emerging research, is analogous in many ways to the venture capital model of backing high-risk ideas and rewarding effective allocators.LDRD is funded at each national lab through a small portion of the indirect costs from awarded grants. By statute, a lab’s total annual LDRD spending cannot exceed 6% of its combined operating and capital equipment budget, which is determined through congressional appropriations and is separate from the competitive grants the labs win each year. LDRD is a significant discretionary research funding mechanism in the federal science funding system: In FY 2023, DOE laboratories collectively invested roughly $870 million across more than 2,400 projects through the LDRD program. ^{4 5} While modest in size at the project level, these funds provide enough runway to test new ideas, build early prototypes, or assemble the evidence needed to compete for larger DOE or external grants. ⁶ In effect, LDRD turns a portion of overhead into discretionary R&D capital, allowing labs to act as venture investors inside the federal science ecosystem.

LDRD has a strong track record of supporting breakthrough science ahead of federal agency funding. For instance, in 2008, Lawrence Berkeley National Laboratory awarded an LDRD grant to a team of scientists to investigate an “RNA-based adaptive immune system” known as CRISPR. ⁷ Four years later, one of the principal investigators from that LDRD grant, Jennifer Doudna, would publish a paper on the structure of CRISPR and the Cas9 protein that would later win her the Nobel Prize in 2020.⁸

Today, LDRD continues to provide early funding for research and infrastructure ahead of federal agencies. For instance, there has been significant interest recently in autonomous experimentation platforms: combining robotics and AI to automate scientific experiments and accelerate the discovery pipeline from new materials like high-performance lithium-ion battery electrolytes ⁹ to new drugs to fight disease. ¹⁰ However, while several new federal programs in self-driving labs have recently been announced, ¹¹ multiple DOE national labs recognized the promise of this area several years ago and allocated LDRD funding ahead of the curve. ¹² Much of the development of self-driving labs in the U.S. today owes its early development to DOE LDRD. These kinds of internal bets, made before external consensus formed at the agency level, illustrate precisely the type of institutional foresight LDRD is designed to cultivate.

Where else can the LDRD model be applied?

Universities can draw valuable lessons from LDRD. University indirect costs similarly fund important research activities, such as research infrastructure, graduate fellowships, and new faculty hires, but those funds are typically pooled with broader administrative overhead. This commingling makes indirect costs harder to defend politically and obscures their strategic value. LDRD, by contrast, isolates a defined portion of indirect costs into a transparent internal venture mechanism: labs must competitively solicit proposals, justify their alignment with institutional and national priorities, and report outcomes annually to DOE and Congress. More clearly articulating the capital flow of indirect costs — through explicit models like LDRD or the recently proposed Financial Accountability in Research (FAIR) model — or more carefully ring-fencing the funding within the department or school of science or engineering can help policymakers understand how these funds support early-stage innovation.

LDRD is one embodiment of a broader principle that deserves to be expanded: giving decision-making authority to those closest to discovery. It formalizes local autonomy within a mission-driven framework — letting directors and scientists make early bets, within guardrails of accountability. It also provides a mechanism for channeling overhead from competitively won grants into discretionary funds that reward researchers’ success with future flexibility, and ensures that indirect costs cross-subsidize only work endorsed by proven winners. Congress could expand this mechanism by increasing the 6% LDRD cap rate and encouraging more national labs to utilize LDRD funding up to the available threshold. 

DOE and other agencies could also explore LDRD-like funding mechanisms by, for example, making awards to university departments or non-profit research centers that have demonstrated continued excellence in a specific topic area, similar to the National Science Foundation’s Materials Research Science and Engineering Centers. Proposals like “X-Labs” could also create semi-independent research organizations with their own leadership teams, long-term core institutional support, and flexibility to allocate resources dynamically across projects. Both approaches shift discretion from federal program managers toward local scientific entrepreneurs, testing whether decentralization can accelerate breakthroughs.

The evidence from LDRD suggests that decentralizing a share of research funding can improve responsiveness to new ideas without sacrificing oversight. Evaluating and scaling this model across other mission agencies and research institutions would help the federal government test how distributed decision‑making can accelerate scientific progress.