For nearly eight decades, the U.S. innovation ecosystem has been underpinned by a deliberately decentralized model of federal research support. Rooted in the vision laid out by Vannevar Bush in Science, The Endless Frontier, the model rests on a simple but profound premise: sustained public investment in basic research fuels private sector dynamism and long-term national competitiveness. Bush, writing in the aftermath of World War II, saw that wartime collaboration between academia and industry had borne rapid technological fruit — and wisely foresaw that the country would benefit from facilitating scientific inquiry not just during wartime, but also on an ongoing basis.

Key institutions including the National Science Foundation, National Institutes of Health, and a wide network of research (R1) universities have operationalized this vision. Rather than making top-down strategic bets, the U.S. innovation ecosystem spreads relatively modest sums across a wide array of scientific inquiry, trusting that some portion will yield breakthrough insights. These public investments are not about immediate returns; they are about creating a reservoir of knowledge and capability from which the private sector can draw and often profit.

The Trump administration, in a matter of months, has cut research funds, lowered the overhead rate applicable to federally funded research, and attempted to claw back research grants already approved. Narrowing this research portfolio in the name of efficiency will hurt the venture economy, and ultimately, American competitiveness.

Research and Development Dollars as Venture Portfolio

American public research funding functions as an initial venture portfolio, upstream from the more lauded venture capital ecosystem. The government places a vast number of small bets across disciplines, generating a pipeline of discovery that, over time, feeds downstream commercialization.

As of 2022, the United States spent an estimated $885.6 billion on research and development, of which $159.8 billion was funded by the federal government. U.S. gross domestic product that year was $26 trillion; total federal outlays were $6.3 trillion. Thus, about 2.5 percent of federal outlays went to research. The chart below (drawn from National Science Foundation data), shows spending on research and development in the United States from 1953 to 2022.

Consider the case of the Advanced Research Projects Agency Network of the late 1960s. This network begat the Internet, which then needed a browser. Federally funded research conducted at the University of Illinois yielded Mosaic, which later became Netscape, the first major Internet browser. Or take the early work on lipid nanoparticles, funded by the National Institutes of Health and other agencies for decades. That research became essential to the deployment of the first mRNA COVID-19 vaccines in 2020. In terms of impact, NASA estimates that every dollar spent on research and development has a 7X multiplier effect as it travels through the economy.

This research innovation ecosystem is an essential enabler of the downstream venture capital ecosystem that has transformed the U.S. economy. Like the venture economy, the research innovation ecosystem is characterized by a power law distribution of returns. Some research strands yield today’s mRNA or CRISPR. Others never traverse the valley of death — the stage where commercial interest is too uncertain and technical risk too high to attract private investment. . The very heterogeneity of the portfolio is the point. The successes in the short tail emerge because of the breadth of exploration that has characterized the U.S. innovation ecosystem for 80 years.

There are two valleys of death: one technological and one commercial. The research ecosystem addresses the former; the venture capital ecosystem focuses on the latter. The research ecosystem de-risks technological seeds; meanwhile, venture capitalists invest in teams, technologies, and markets. And venture capital itself thrives on a power law: a few big fund-returning winners make up for a long tail of middling or failed investments. The genius of the U.S. innovation ecosystem is that it tolerates this uncertainty, understanding that asymmetric returns are a feature, not a bug.

From Discovery to Deployment: A Division of Innovation Labor

To fully appreciate the strength of the U.S. innovation ecosystem, it helps to understand how ideas move from the laboratory to the marketplace. One widely used framework for tracing this process is the Technology Readiness Level (TRL) scale, a nine-point system originally developed by NASA.

At the earliest stages (TRLs 1–3), research is often theoretical or experimental. Universities and research institutes, supported by agencies including the National Science Foundation, National Institute of Health, and Department of Energy Office of Science, play a lead role here. These institutions generate foundational knowledge without needing to predict practical application. For example, the discoveries behind today’s therapeutics and gene editing emerged from curiosity-driven work long before their relevance to global public health or biotechnology became clear.

As technologies advance to TRLs 4–6, the challenge becomes one of validation and integration — moving from the lab bench to functional prototypes. This is the domain of national laboratories, University Affiliated Research Centers, and Federally Funded Research and Development Centers like the RAND Corporation. These institutions conduct applied research and system integration, helping promising technologies navigate the valley of death. It was in this middle stage that Lawrence Berkeley and Oak Ridge National Laboratories helped transition early battery chemistries into viable storage technologies, and where MIT Lincoln Laboratory refined radar and sensor systems that became essential to missile defense. Today, innovations in nuclear fusion are following the same trajectory.

Finally, TRLs 7–9 mark the transition to commercialization. This is where technologies are demonstrated in real-world settings, manufactured at scale, and brought to market. In this final stage, the private sector takes the lead. Think of SpaceX, which scaled up reusable rocket technology after foundational propulsion work conducted at NASA, or Moderna, which commercialized mRNA platforms first explored through National Institutes of Health-funded research.

This division of labor across the innovation lifecycle is a quiet but powerful engine. Universities generate possibility. Labs reduce risk. Markets deliver scale. Each plays a distinct role, and each depends on the health of the others. Strategic efforts to bypass or consolidate this structure — to shortcut curiosity in favor of targeted returns — risk breaking the feedback loops that have made the U.S. innovation model so resilient and productive.

The Double Power Law

We describe this phenomenon as an underappreciated “double power law” of American innovation. First, federal research funding, distributed widely and without rigid priorities, produces breakthrough scientific knowledge with highly unequal payoffs. Second, this knowledge is captured by a downstream venture capital system that also operates on a power law distribution of returns.

While the power law nature of venture capital is well-documented (and even provides the title of a book that Jon uses in class: The Power Law: Venture Capital and the Making of the New Future), the fact that a double power law is at work is less understood. Just as failure is rewarded in the venture economy if it leads to lessons that de-risk the entrepreneur’s next venture, so too must the breadth of inquiry in the research ecosystem be understood as essential. It is a feature, not a bug.

Together, this “double power law” creates an innovation engine that has made the United States a global leader across multiple sectors, not just one. The outcomes are lumpy, nonlinear, and difficult to plan — but they are unrivaled in their overall yield. It is the form of industrial policy that other countries cite with admiration as an unmitigated American policy success.

The Perils of Conflating Breadth of Research with Placing Strategic Bets

The current administration, however, has sought to disrupt this model. We have seen this movie in planned economies before and it has not gone well. In addition to creating a disincentive to pursue federally funded research, this also runs contrary to power law logic.

It is the prerogative of administrations to place bets. The Apollo program and Operation Warp Speed are both examples of the power of targeted federal support. The moonshots themselves deliver fruit, and so, too, do their by-products. But such bets should be placed separately from the broad research portfolio described above. Further, programs like DARPA’s benefit from program managers having the autonomy to dive deeply into solving one particular technological problem, and find the right answer, not a preferred answer.

By emphasizing “strategic sectors,” governments might leverage industrial policy based on short-term, non-scientific judgment about which technologies deserved support. Rather than betting on science, we begin betting on bureaucratic foresight.

This shift also undermines a basic understanding that has long governed American innovation. Universities and researchers have accepted uncertainty and competition (not to mention lower pay) because the system has been transparent and open to heterogeneity of research inquiry. Consequently, startups, investors, and society at large have benefited from a steady stream of publicly funded science. Further, since Bayh-Dole, universities and entrepreneurs have benefited from being able to participate in the commercialization of research originally funded by the federal government.

Strategic bets, combined with culling of the research portfolio, risk concentrating resources, narrowing participation, and ultimately bringing a wholly unnecessary end to this 80-year run of success.

Moving Forward

The innovation ecosystem that underpins U.S. national security is the product of decades of trust, investment, and experimentation. Preserving it requires resisting the temptation of control — recognizing that scientific discovery cannot be forced into strategic lanes without degrading its potential.

To compete with geopolitical challengers like China, the United States should double down on the very model that built its technological and human capital advantage: decentralized funding, open inquiry, and public-private complementarity.

Defunding the research innovation ecosystem in the name of efficiency is penny-wise and pound-foolish. Further, it will shrink the pool of technology available to be brought to market by the downstream venture economy. Would venture capitalists move upstream, to fill in gaps in the research funding ecosystem? Not likely — they have very different incentives, driven by the need to deliver consistent returns to limited partners. They are not in the business of funding research.

Any administration would do well to remember this. America’s edge does not lie in predicting the future. It lies in preparing for it by supporting the unpredictable and attracting the brightest in the world to work on inventing the future.

Jon Metzler is a continuing lecturer at the Haas School of Business at the University of California, Berkeley, where he teaches on strategy for the networked economy, and innovation and entrepreneurship. Jon is a faculty mentor at Berkeley SkyDeck, the accelerator for Berkeley-affiliated startups. He is also a faculty fellow and a member of the faculty advisory board at the Berkeley Risk and Security Lab.

Andrew W. Reddie is an associate research professor at the University of California, Berkeley’s Goldman School of Public Policy, where he teaches at the intersection of technology and international security and is the founder of the Berkeley Risk and Security Lab.

Image: NASA’s Scientific Visualization Studio via Wikimedia Commons