A NEW GOLDEN AGE

America stands at the cusp of a revolution in science, in which AI will accelerate discovery, multiply human cognitive capabilities, and unlock solutions to some of our greatest challenges. But “AI for science” will still find itself subject to the frictions and inefficiencies of human institutions. We can only fully harness AI and its associated productivity uplift by boldly reforming our scientific institutions, building national-scale infrastructure, and ensuring rigorous verification of the knowledge base from which AI will learn.

• Launch and Scale the Genesis Mission: Fully fund and expand the Genesis Mission as America’s flagship AI for science initiative, integrating supercomputers, AI models, scientific instruments, and datasets across national laboratories to double the productivity and impact of U.S. science within a decade. Direct it at cross-cutting problems where breakthroughs unlock entire branches of downstream discovery and where AI can transform the practice of science itself.

• Institutionalize Gold Standard Science: AI operating on a flawed knowledge base will only entrench bad science. Enforce reproducibility, transparency, data sharing, and falsifiability across all federally funded research through the Restoring Gold Standard Science Executive Order, creating a trusted foundation for AI-powered discovery.

• Build Verification Infrastructure at Scale: While the cost of generation has decreased exponentially, the cost of verification has not. Invest in AIenabled verification systems, open standards, and continuous replication mechanisms. Set standards to enable the development of machineauditable replication packages, and reward those who replicate or disprove influential scientific results.

• Accelerate Autonomous Experimentation: Closed-loop autonomous laboratories can collapse discovery timelines by orders of magnitude and enable science at a truly industrial scale. Focus investments in robotics and automated laboratories, leveraging industry demand and federal R&D to ensure our scientific equipment industrial base is built on the world’s best hardware and software and leads the charge in the coming scientific revolution.

• Experiment With AI-Native Scientific Institutions: Today’s funding structures, publication systems, and credit mechanisms were built for a world of human-paced discovery. Begin the transition to AI-native institutions, including through faster and more open forms of scientific publication, more granular credit attribution, and new market mechanisms that direct resources to problems where breakthroughs matter most.

SCIENTIFIC PROGRESS REMAINS ESSENTIAL

Eighty-one years ago, Vannevar Bush wrote that scientific progress would be an “essential key to our security as a nation, to our better health, to more jobs, to a higher standard of living, and to our cultural progress.”1 Bush had seen a glimpse of science’s promise for America in the triumphs of penicillin and radar in securing victory in World War II. His vision proved prescient through the eight decades that followed.

Since then, incredible breakthroughs have emerged from our nation’s laboratories: the transistor and the integrated circuit, the laser and the LED, the mapping of the human genome and the tools to edit it. The “new products, new industries, and more jobs”2 Bush envisioned have materialized as entire economic sectors, such as computing, biotechnology, aerospace, and telecommunications, which today employ tens of millions and generate trillions in wealth. Enabled by technologies unimaginable in Bush’s time, the energy revolution has made America the world’s largest oil and gas producer.

Now small computers in our pockets connect us instantly to family across the continent, unlock the world’s knowledge, and guide us through unfamiliar streets. Great advances in materials science have given us everything from nylon stockings to bulletproof vests, artificial joints, and fighter jets. We placed GPS satellites in orbit that guide our tractors to precision planting, our packages to on-time arrival, and our troops through hostile terrain. We established the field of modern biotechnology, invented MRI, and developed the lithium batteries that power our cordless world. And in mere decades after Bush’s letter, we walked on the Moon and sent scientific instruments to the edge of the solar system. Our scientists have achieved even more than this. We often forget that American farmers have tripled their output while using about a quarter of the labor used in the 1940s,3 or that the average American lives more than 10 years longer than when Bush penned his report.4 Within living memory, cancer has been transformed from a death sentence to a treatable condition for millions of Americans. The most common form of childhood leukemia has gone from universally fatal to curable in 90% of cases,5 and deaths from heart disease have fallen by more than half since their peak.6

AS THE SOURCE OF OUR TRIUMPHS

These triumphs happened, and happened here in America, only because of intentional choices made by our people and institutions. First, consistent with what Bush outlined in his essay, the government played a vital role in supporting the scientific enterprise to achieve national goals. These include conquering disease, creating jobs, and ensuring security. Against the backdrop of pre-War federal research funding, which was largely focused on agriculture, this proved to be a key insight.7 Not all valuable research attracts private capital, particularly research that promises only slow, diffuse returns. In the capital environment of the mid-20th century, no investor would have funded efforts to build particle accelerators or discover the fundamental insights that underlie the genomic revolution. As Bush argued, there are “areas of science in which the public interest is acute but which are likely to be cultivated inadequately if left without more support than will come from private sources.”8 Today, we benefit from that insight with an extensive set of federal organizations to advance scientific research, including NSF, DOE national laboratories, NIH, NASA, the National Institute of Standards & Technology (NIST), the Defense Advanced Research Projects Agency (DARPA), and other research arms of federal departments and agencies. Second, our government recognized that achieving those national purposes requires more coordination than any single institution can provide, and that our unique advantage, whether in defeating the Soviet Union or winning the technological race today, lies in our dynamic private sector. Describing the development of penicillin, Bush spoke of how the government launched a “coordinated attack on special problems,” supporting research and development among medical schools, universities, and the pharmaceutical industry, and helping ideas progress from early laboratory experimentation to large-scale production and use.9 This model became the foundation of the fruitful public-private partnerships that sent Americans to the Moon and built the internet. This dynamism between publicly funded science and private enterprise remains the engine of American innovation. Third, we stayed true to the call for science to remain dynamic. “The pioneer spirit is still vigorous within this nation,” Bush wrote.10 “Science offers a largely unexplored hinterland for the pioneer who has the tools for his task.”11 Each of our past triumphs required substantial courage and institutional transformation.

They compelled us to invent new models to drive scientific progress: dedicated science funding agencies, innovative partnerships that enabled the widespread commercialization of modern electronics, and reforms like the deregulation of space that opened the door to today’s vibrant era of commercial spaceflight. This willingness to venture into unknown territory, to challenge established methods, and to create new institutions when old ones prove inadequate, built the scientific supremacy that undergirds our vibrant economy and national security today.

THE LANDSCAPE IS CHANGING

The principles that government must support basic research, that this research drives national prosperity, and that America’s advantage lies in the dynamism of our institutions, remain as sound today as when Bush first articulated them. But principles are not processes. Bush would be the first to recognize that the landscape in which fundamental research is conducted has completely transformed since he wrote Science: The Endless Frontier. In 1950, a dozen engineers in basic laboratories drove progress in semiconductors. Today, the semiconductor industry invests more than $100 billion in capital and R&D each year and employs hundreds of thousands.12 They regularly solve physics and materials problems at the edge of possibility and build fabrication plants filled with robots that manipulate silicon atom by atom. In 1950, scientists mailed typewritten manuscripts to journal editors, who sent copies to reviewers from their personal networks at top universities. Today, researchers post papers online within hours of completion. Thousands read and debate the merits of the work immediately on social media and in discussion channels. Code gets replicated on the internet months before the paper appears in print. In 1950, mathematicians worked alone with chalkboards and stacks of papers from the library. Today, they look up theorems instantly online. Computers enable experimental mathematics that would have been impossible with pencil and paper. Software languages modularize massive proofs, letting dozens of mathematicians collaborate on a single problem simultaneously from coast to coast. 4 Chapter I – Introduction NEW FRONTIERS AND NEW APPROACHES The institutions we build determine what problems get solved, which approaches get tried, what risks get taken, and whose talent contributes to discovery. When these institutions align with the nature of the scientific frontier and with our national needs, science advances; when they are misaligned, abundant resources and brilliant researchers go to waste. The misalignment shows up as diminishing returns to R&D investments, a decline in the pursuit of breakthrough ideas, and a slowdown in the benefits that technological progress delivers to the American people. Institutional design matters because individual researchers follow the signals their institutions send. Consider the incentives of a talented researcher working at the frontier of quantum information science. As a Ph.D. student in a university lab, this researcher must publish regularly to graduate, craft narratives that satisfy journal reviewers and his dissertation committee, and build the personal connections that lead to academic jobs. His professor’s grant funding limits what equipment he can afford. He designs experiments around the apparatus in his lab more than the questions most worth asking. With two years until graduation, he actively looks for results that advance his dissertation’s narrative. When unexpected results appear, he sometimes chooses to pursue them, but remains cognizant of potential risks to his professional progress and his lab’s future funding. As a startup founder raising venture capital, the same researcher faces different pressures. He pitches a bold vision of scalable quantum computing to investors. He can hire engineers and build quickly with tens of millions of dollars in seed funding. But he must also deliver revenue within five years and sustain a clear narrative as funding rounds continue. The technical approach he outlined to investors may not be the best path forward, but changing course risks losing investor confidence. Market pressure imposes scientific constraints that grant reviewers might not. Both paths advance science and technology, but both channel talent toward different problems in different ways. The discoveries that get made depend not only on the questions that are scientifically salient, but on the fit between those questions and the incentives researchers must navigate. It therefore falls to the public officials who steward federal funding, as the architects of the national scientific enterprise, to understand the constraints our researchers face, to create the right incentive structures wherever possible, and to drive R&D in whatever gaps remain. Only then can we unleash American scientists and give them ever greater freedom to explore.

Universities remain essential for training scientists and pursuing fundamental questions. Venture capital mobilizes private resources toward high-impact technologies. Federal agencies fund research that markets alone will not support. Each serves an essential purpose, but the scientific frontier is constantly shifting, requiring vigilant self-improvement to ensure these institutions remain suited to the answering the most important questions today.

The system that emerged from Bush’s vision served the last American Century. But every generation of Americans must show the courage to reinvent our institutions when the frontier demands it. We established land-grant universities when agriculture needed scientific foundations. We created DARPA when the pursuit of breakthrough military technologies required an agency willing to fund high-risk ideas that traditional funders would reject. We developed the venture capital model when a gap emerged between the long time horizons of emerging technology companies and the capacity of traditional capital markets. The questions demanding answers, the tools required to answer them, the scale of coordination needed, and the timelines involved all shift as knowledge advances. As our predecessors did, we must continue to craft and refine the machinery of science, allowing each component to work to full advantage and freeing our innovators from pressures that keep them from the greatest goals. A new American Century will require new engines of scientific discovery.

GROWING PRIVATE SECTOR R&D

One particularly visible shift in the scientific machine is who funds and performs research. When Bush penned his report in the middle of the 20th century, the Federal Government stood as the dominant patron of American science, marshaling the nation’s research capacity for victory in war. The landscape today would astonish him. Private industry has become by far the largest source of R&D funding in the United States, with its share roughly doubling from the 1950s to today, even as federal funding has grown by leaps and bounds. American companies now deploy around $700 billion annually, more than triple the spending of government and higher education.13 While this investment has historically been dominated by late-stage product development, strikingly, the share devoted to basic research, which Bush thought markets could not sustain alone, has also grown rapidly, particularly over the past two decades (Figure 1). Consider two transformative inventions of recent memory, the transistor and the transformer architecture that underpins modern machine learning. Both came from corporate laboratories that employ thousands of researchers who often produce collaborative, well-cited papers on problems of deep intellectual interest.

Modern industrial powerhouses fund state-of-the-art experiments and pay salaries tens or hundreds of times more than the academy, drawing top talent from across the country. Researchers at American companies have earned Nobel Prizes for work on electron tunneling in semiconductors, surface chemistry, polymer science, and lasers, a testament to both the rigor of their research and the fundamental nature of their work.

This is not a sign that the academy has become less important, but rather that the scientific world has expanded. In certain domains, the scale of private investment dwarfs anything federal agencies can match. Universities face real limits in scaling up engineering efforts, with rare exceptions for government-sponsored big science projects like space probes and particle accelerators;

yet Ph.D. students and professors can now raise hundreds of millions of dollars to found companies that pursue fundamental breakthroughs. Small startups increasingly perform basic research themselves, giving our most talented scientists new paths for ambitious work. These firms blur the distinction between basic and applied science, combining research and development to accelerate both.

Figure 1: Private sector basic R&D has grown rapidly over the past two decades. It now rivals higher education among performers (left) and the Federal Government among funders (right) of basic research in the United States.14 BASIC ANNUAL R&D SPEND BY PERFORMER

Higher Education Business Federal Nonprofits 1957 1963 1969 1975 1981 1987 1993 1999 2005 2011 2017 2023 1957 1963 1969 1975 1981 1987 1993 1999 2005 2011 2017 2023 60 50 40 30 20 10 0

R&D Spend (Billions, 2017 Dollars) BASIC ANNUAL R&D SPEND BY SOURCE 7

THE LINEAR MODEL NO LONGER HOLDS

In his 1945 report, Vannevar Bush presented a progression from basic research through applied research to development, later termed the “linear model.”15 This framework laid out a clear role for each part of the research and development pipeline. Universities would pursue fundamental understanding without the pressure of practical application. Industry would turn discoveries into products.

At the time, the separation was natural and productive. Modern discovery, however, is increasingly shaped by continuous iteration between fundamental and applied work. Engineering challenges routinely expose unanswered scientific questions, and breakthroughs in basic understanding, in turn, open new engineering pathways. The relationship is recursive rather than unidirectional. The tools required to push the frontier, whether advanced fabrication equipment or specialized engineering teams, are often found outside traditional academic settings. Many of our most productive researchers now move fluidly between sectors, carrying ideas and techniques with them. Technology and science have become deeply interdependent, even, as we will discuss in Chapter V, in the purest fields of reason like mathematics.

This non-linearity is also captured by the framework Donald Stokes articulated half a century after Bush. “Pasteur’s quadrant,” as he termed it, now defines a growing share of the scientific frontier.16 Stokes argued that research can seek fundamental understanding while being motivated by considerations of use, observing that some of the most consequential scientific advances all arose precisely from this combination. Examples of such research include Pasteur’s investigations into why wine spoiled, the development of the transistor at Bell Labs, and Shannon’s work on information theory. Many of the transformative discoveries of our own era, from the computational study of protein folding that earned the 2024 Nobel Prize in Chemistry to the superconducting quantum devices that earned the 2025 Nobel Prize in Physics, emerged from efforts that were at once theoretically ambitious and deeply connected to practical problems. Our national laboratories and federally funded research centers have long been engines for this use-inspired research, and new centers of modern science reflect the same pattern. A greater share of scientists are leaving academia for industry, not because they have abandoned curiosity-driven inquiry, but because the tools, resources, and career opportunities required to pursue certain fundamental questions increasingly lie outside university walls. The task ahead is to enable scientists to move fluidly between problems of different shapes and to give them the freedom and resources to pursue discovery at today’s frontier.