Foreign Affairs
The Crumbling Foundations of American Strength
Knowledge Is Power—and the United States Is Losing It
Amy Zegart
September/October 2024
Published on August 20, 2024
When Russia’s invasion of Ukraine appeared imminent in early 2022, U.S. intelligence officials were so confident that Russian tanks would roll quickly to victory that staff evacuated the U.S. embassy in Kyiv. Based on traditional measures of power, the intelligence assessment made sense. In 2021, Russia ranked fifth in the world in defense spending, whereas Ukraine was a distant 36th, behind Thailand and Belgium. Yet more than two years later, Russia and Ukraine are still fighting their brutal war to a standstill.
Ukraine’s resilience is a telling indicator that power is not what it used to be. The country’s surprise showing is in no small part a result of its highly educated population and a technology innovation ecosystem that has produced vast quantities of drones and other homemade weapons on the fly. Ukraine has even managed to wage naval warfare without a navy, using homemade drones and other devices to destroy nearly two dozen Russian ships and deny Russia control of the Black Sea.
For centuries, a nation’s power stemmed from tangible resources that its government could see, measure, and generally control, such as populations that could be conscripted, territory that could be conquered, navies that could be deployed, and goods that could be released or restricted, such as oil. Spain in the sixteenth century had armies, colonies, and precious metals. The United Kingdom in the nineteenth century had the world’s strongest navy and the economic benefits that emerged from the Industrial Revolution. The United States and the Soviet Union in the twentieth century had massive nuclear arsenals.
Today, countries increasingly derive power from intangible resources—the knowledge and technologies such as AI that are super-charging economic growth, scientific discovery, and military potential. These assets are difficult for governments to control once they are “in the wild” because of their intangible nature and the ease with which they spread across sectors and countries. U.S. officials, for example, cannot insist that an adversary return an algorithm to the United States the way the George W. Bush administration demanded the return of a U.S. spy plane that crash-landed on Hainan Island after a Chinese pilot collided with it in 2001. Nor can they ask a Chinese bioengineer to give back the knowledge gained from postdoctoral research in the United States. Knowledge is the ultimate portable weapon.
The fact that these resources typically originate in the private sector and academia makes the job of government even more challenging. Foreign policy has always been a two-level game; U.S. officials have to wrangle both domestic actors and foreign adversaries. But more and more, the decisions of private companies are shaping geopolitical outcomes, and the interests of the U.S. private sector are not always aligned with national objectives. Meta, the parent company of Facebook, Instagram, and WhatsApp, is determining what constitutes truth for the three billion people worldwide that use its platforms. In the past year, American CEOs with vested Chinese business interests have met face-to-face with Chinese leader Xi Jinping about as often as Secretary of State Antony Blinken has. And when war erupted in Ukraine, the tycoon billionaire Elon Musk singlehandedly decided whether, where, and when the Ukrainian military could communicate using the Starlink satellite network he owns.
At the same time, many of the U.S. government’s capabilities are deteriorating. Its traditional foreign policy tools have withered: confirming presidential appointments has become so fraught that at least a quarter of key foreign policy positions sat vacant halfway through the first terms of the last three U.S. presidents. Thanks to spiraling federal debt, this year, for the first time ever, the U.S. will spend more on interest payments than on defense. Because Congress often cannot pass an annual budget, the Pentagon increasingly runs on stopgap budget measures that fund only existing programs, not new ones, preventing new research and development initiatives or weapons programs from getting off the ground. This broken system disproportionately hinders new, small, and innovative companies. As a result, big, expensive weapons systems persist while new, cheap solutions wither on the vine. If China were to design a budget process with the intent to stifle invention, send weapons costs through the roof, and weaken American defense, it would look like this. Meanwhile, and critically, the health of the United States’ K–12 education and research universities—the sources of the country’s long-term innovative potential—are in decline.
In today’s knowledge- and technology-driven world, U.S. policymakers need to think in new ways about what constitutes U.S. power, how to develop it, and how to deploy it. Future prosperity and security will depend less on preventing adversaries from acquiring U.S. technologies and more on strengthening the country’s educational and research capacity and mobilizing emerging technologies to serve the national interest.
INNOVATE AND ANTICIPATE
For decades, U.S. policymakers have employed hard- and soft-power tools to influence foreign adversaries and allies. To advance U.S. interests with hard power, they built military might and used it to protect friends and threaten or defeat enemies. With soft power, they shared U.S. values and attracted others to their cause. Both hard and soft power still matter, but because they do not determine a country’s success the way they once did, the United States must work to expand its knowledge power—advancing national interests by boosting the country’s capacity to generate transformational technology.
Knowledge power has two essential elements: the ability to innovate and the ability to anticipate. The first relates to a country’s capacity to produce and harness technological breakthroughs. The second has to do with intelligence. Part of this work fits into the traditional mission of U.S. spy agencies, which are tasked with discovering the intentions and capabilities of foreign adversaries to threaten U.S. interests. As the boundaries between domestic industry and foreign policy blur, however, intelligence agencies also need to help the government understand the implications of technologies developed at home.
Innovation and anticipation are not merely ingredients that strengthen the United States’ military and its powers of attraction. They may do both, but the primary function of knowledge power lies closer to home. Whereas traditional foreign policy tools aim outward—using threats, force, and values to affect the behavior of foreign actors—building and using knowledge power requires Washington to look inward. It involves marshaling ideas, talent, and technology to help the United States and its partners thrive no matter what China or any other adversary does.
Education and innovation are key to the United States’ ability to project power.
The components of knowledge power can be hard to see and quantify. But a good place to start is national educational proficiency levels. Overwhelming evidence shows that a well-educated workforce drives long-term economic growth. In 1960, East Asia nearly tied sub-Saharan Africa for the lowest GDP per capita in the world. Over the next 30 years, however, East Asia vaulted ahead, spurred in large measure by educational improvements.
The geographic concentration of technological talent is another useful indicator of knowledge power, suggesting which countries are poised to leap ahead in critical areas. There is a reason leading scientists and engineers congregate in labs and recruit superstar teams instead of isolating themselves in their offices, designing experiments alone and reading research papers online. Physical proximity matters; the world’s top minds working closely together is a recipe for technological breakthroughs.
Gauging a nation’s long-term power prospects also requires measuring the health of its research universities. Companies play an essential role in technological innovation, but the innovation supply chain really begins earlier, in campus labs and classrooms. Whereas companies must concentrate their resources on developing technologies with near-term commercial prospects, research universities do not face the same financial or temporal demands. Basic research, the lifeblood of universities, examines questions on the frontiers of knowledge that may take generations to answer and may never have any commercial application. But without it, many commercial breakthroughs would not have been possible, including radar, GPS, and the Internet.
More recently, what looked from the outside like the overnight success of mRNA-based COVID-19 vaccines was in fact the result of more than 50 years of basic research in universities. Before pharmaceutical companies advanced vaccine development, academic researchers had discovered that mRNA could activate and block protein cells, and they had figured out how to deliver it to human cells to provoke an immune response. Similarly, the cryptographic algorithms protecting data on the Internet today stemmed from decades of academic research in pure math. And many new advances in AI, from ChatGPT to image recognition, build on the pioneering work developed at the University of Toronto, the University of Montreal, Stanford University, and elsewhere.
BRAIN DRAIN
If education and innovation are key to the United States’ ability to project power, then the country’s prospects are on shaky ground. American K–12 education is in crisis. Students today are scoring worse on proficiency tests than they have in decades and falling behind their peers abroad. U.S. universities are struggling, too, as they face greater global competition for talent and chronic federal underinvestment in the basic research that is vital for long-term innovation.
In 2023, math and reading scores among American 13-year-olds were the lowest in decades, according to the National Assessment of Educational Progress. Half of U.S. students could not meet their state’s proficiency requirements. And scores on the ACT, the popular college admissions test, declined for the sixth year in a row, with 70 percent of high school seniors not meeting college readiness benchmarks in math and 43 percent not meeting college readiness benchmarks in anything. Notably, these trends began before the COVID-19 pandemic.
While students in the United States fall behind, students in other countries are surging ahead. According to the Program for International Student Assessment, which tests 15-year-olds worldwide, in 2022 the United States ranked 34th in average math proficiency, behind Slovenia and Vietnam. (Reading and science rankings were higher but barely cracked the top ten and top 20, respectively.) More than a third of U.S. students scored below the baseline math proficiency level, which means they cannot compare distances between two routes or convert prices into a different currency. At the top end, only seven percent of American teens scored at the highest level of math proficiency, compared with 12 percent of test takers in Canada and 23 percent in South Korea. Even pockets of excellence inside the United States don’t fare well internationally. Massachusetts was the top-scoring U.S. state in math in 2022 but would rank just 16th in the world if it were a country. Most U.S. states rank near the global median. And the lowest-scoring state, New Mexico, is on par with Kazakhstan.
Part of this story is the rise of the rest; the global population has become vastly more educated in the past several decades, redrawing the knowledge power map in the process. Since 1950, average years of schooling have risen dramatically and the number of college graduates worldwide has increased 30-fold. As the educational playing field levels, U.S. universities and companies increasingly rely on foreign talent to remain world-class. In 1980, 78 percent of doctorates in computer science and electrical engineering awarded by American universities went to U.S. citizens or permanent residents. In 2022, it was 32 percent. About one million international students now study in the United States each year. The largest share comes from China, at 27 percent.
The United States’ record of attracting talent from around the world is an enormous asset. Nearly 45 percent of all Fortune 500 companies in 2020, including Alphabet, SpaceX, and the chip giant NVIDIA, were founded by first- or second-generation immigrants. About 40 percent of Americans awarded Nobel Prizes in scientific fields since 2000 have been foreign-born. Yet here, too, the country is forfeiting its short-term advantage and creating long-term vulnerabilities. Outdated immigration policies have created a self-sabotaging talent system that educates exceptional foreign students and then requires many of them to leave the United States, taking everything they learned with them.
What’s more, this talent supply chain works only as long as foreign students want to study in the United States and their governments allow it. Foreign universities have improved substantially in recent years, offering more alternatives for the best and brightest. Already, polls show that the share of Chinese students who prefer to study in Asia or Europe instead of the United States is rising. If the Chinese government were ever to restrict the flow of top students to the United States, many university labs and companies would be in serious trouble.
The innovation advantage that U.S. universities have over their foreign counterparts is eroding, too. A decade ago, the United States produced by far the most highly cited scientific papers in the world. Today, China does. In 2022, for the first time, China’s contributions surpassed those of the United States in the closely watched Nature Index, which tracks 82 premier science journals.
The pull of the private sector is draining the sources of future innovation.
Funding trends are also headed in the wrong direction. Only the U.S. government can make the large, long-term, risky investments necessary for the basic research that universities conduct. Yet overall federal research funding as a share of GDP has declined since its peak of 1.9 percent in 1964 to just 0.7 percent in 2020. (By comparison, China spent 1.3 percent of GDP on research in 2017.) The 2022 CHIPS and Science Act was supposed to reverse this downward slide by investing billions of dollars in science and engineering research, but these provisions were later scrapped in budget negotiations.
Basic research has been particularly hard hit. Until 2014, the National Institutes of Health allocated the majority of its budget to basic university research about disease and human health. Now, it spends more on clinical trials and other applied research. The CHIPS and Science Act was supposed to double the budget of the National Science Foundation, the premier government sponsor of basic research in nonmedical science, technology, engineering, and math, this year. Instead, the agency’s budget was cut by eight percent.The NSF awards smaller, shorter grants than it did a decade ago, which forces scientists and engineers to spend more time chasing funds and less time conducting research. “We are fast approaching the point where standard NSF grants aren’t minimally viable,” said one senior administrator at a large research university, who spoke on the condition of anonymity to avoid jeopardizing relations with the NSF. “For some of our faculty, it’s just not worth it for them to apply.” Although the United States still funds more basic research than China does, China’s investment in research rose more than 200 percent between 2012 and 2021, compared with a 35 percent rise in U.S. investment. If current trends continue, China’s basic research spending will overtake U.S. spending within ten years.
The gravitational pull of the private sector is bolstering short-term innovation and economic benefits, but it is also draining the sources of future innovation. In AI, for example, the talent exodus from academia to industry is fueling extraordinary commercial advancements. It is also diverting talent and attention away from basic research on which future innovation depends and depleting the ranks of faculty who teach the next generation. The problem is acute at the very top. In one top-ranked U.S. computer science department, nearly a third of the senior AI faculty a decade ago have left academia. At another top-ranked department, an AI scholar, who spoke on the condition of anonymity, has estimated that half the AI faculty have gone part-time. Doctoral students and faculty at an AI lab at another leading university do not have the ability to discuss their research freely, which is vital for collaboration, because some are working at OpenAI and have signed nondisclosure agreements. Last year, more than 70 percent of newly minted AI Ph.D.’s in the United States went directly to industry, including a disproportionate share of the top students. As a U.S. government commission on AI put it, “Talent follows talent.”
A generation from now, policymakers will lament, “How could we not have seen this talent crisis coming?” But all they need to do is look.
A NEW POWER BASE
U.S. policymakers need a new playbook that will help them assess, enhance, and use the country’s knowledge power. The first step is developing intelligence capabilities to gauge where the United States is ahead in emerging technologies and where it is behind, and to determine which gaps matter and which do not. The Pentagon has legions of analysts comparing U.S. and foreign military capabilities, but no office in the U.S. government does the same for emerging technologies. This needs to change. The Office of the Director of National Intelligence has already begun building stronger relationships with companies and universities to gain insight into U.S. technological developments. These efforts must be institutionalized, with channels to share expertise faster and more frequently. To spur progress, Congress should hold annual technology net assessment hearings with intelligence officials and academic and industry leaders. And universities must step up by sharing the details and implications of their latest lab discoveries. For instance, my institution, Stanford University, launched a new initiative last year called the Stanford Emerging Technology Review to provide more accessible and regular information to policymakers about ten key emerging technologies—including AI, bioengineering, space technologies, materials science, and energy—from leading experts in those fields. It is now essential to broaden and deepen these efforts, building trusted expert networks and increasing information sharing between universities and the U.S. government, state and local officials, and international partners.
Washington also needs to invest in the national infrastructure necessary for technological innovation. In the 1950s, President Dwight Eisenhower developed the Interstate Highway System to bolster U.S. economic growth and to make it easier to evacuate civilians and move troops in the event of a Soviet attack. After the 1973 oil crisis, President Gerald Ford established the Strategic Petroleum Reserve, the largest stockpile of emergency crude oil in the world, so that a foreign oil embargo or other disruption would never again cripple the U.S. economy. The missing national security infrastructure today is computational power. Progress in nearly every field relies on artificial intelligence, which in turn requires advanced computational power to operate. For example, the computational power required to train the ChatGPT-3 AI model is so huge that the task would take 9,000 years on a typical laptop. Today, only large companies such as Amazon, Google, Meta, and Microsoft can afford to buy the massive clusters of advanced chips required for developing frontier AI models. Everyone else struggles to afford the bare minimum. This year, Princeton University announced that it would dip into its endowment to purchase 300 advanced NVIDIA chips to use for research (at a cost of at least $9 million), while Meta announced plans to have 350,000 of the same chips by year’s end, spending an estimated $10 billion.
A national strategic computational reserve would provide free or low-cost advanced computing to researchers through competitive grants that lease time on existing cloud-based services or supercomputing systems at national labs. The reserve could also build and operate smaller-scale computing clusters of its own. This infrastructure would be accessible to researchers outside large tech companies and well-endowed research universities. It would facilitate cutting-edge AI research for public benefit, not just private profit. And it would help stem the flow of top computer scientists from academia to industry by offering them resources to do pioneering work while remaining in their university positions. Improvements are already underway. In January, the National Science Foundation launched a pilot program called the National AI Research Resource, awarding access to computational power, data, and other resources to 35 projects out of more than 150 proposals. A bipartisan group of legislators has introduced a bill to make the NAIRR permanent.
At an elementary school in Artesia, California, January 2024
Mike Blake / Reuters
Enhancing U.S. knowledge power is not just about developing new capabilities. Washington also needs to fix problems in the country’s immigration system and defense budgeting. Congress must pass immigration reforms to allow more of the world’s best and brightest students to stay and work in the United States after they graduate from American universities, provided measures are in place to protect U.S. intellectual property and guard against espionage risks. The secretary of defense should make reform of the Pentagon’s weapons acquisition process a top priority, putting real funding behind long-standing promises to embrace affordability and innovation and making clear to Congress and the American people that budget dysfunction makes the country less safe.
If U.S. research universities are to remain engines of future innovation, the federal government must also reverse years of chronic underinvestment in basic research. Some private-sector leaders are trying to fill gaps through philanthropic programs such as Schmidt Sciences’ AI2050, which is committing $125 million over five years to fund bold academic research in AI. But this is a drop in the bucket. Only the U.S. government—which spends $125 million on a single F-35 fighter jet—can invest on the scale that is necessary. A bipartisan group of lawmakers, led by Senators Martin Heinrich, Mike Rounds, Chuck Schumer, and Todd Young, has renewed calls to fulfill the original promise of the CHIPS and Science Act by ramping up current government funding for nondefense AI research and development tenfold, to $32 billion. Yet the road between this proposal and passing a bill is long; the idea has been floating around Congress since 2021. That’s a lifetime ago in AI development. Given the pace and stakes of technological change, it is not enough for funding to increase. It also needs to be delivered faster.
Finally, the United States needs to fix K–12 education. Warnings that educational decline threatens the country’s future prosperity, security, and global leadership are nothing new, but education reform has not been treated as the urgent national security priority that it is. Today, in most of the country’s 13,500 public school districts, teacher compensation is based on years of experience and graduate education, which means that physics and physical education teachers receive the same pay. So do the best and worst teachers. Some cities are already piloting better approaches. In Dallas, Houston, and Washington, D.C., education officials have been experimenting with incentive funds to evaluate teachers and reward the most effective ones. In some places, districts can receive even more discretionary funding if they deploy the best teachers to the worst schools. These practices are already producing promising results, and they should be studied and scaled.
None of these changes will be easy, but without them, the United States’ knowledge capacity will continue to erode and U.S. power will grow weaker in the years ahead. Washington has been clinging to the idea that restrictions on China’s access to U.S. technology through export controls and outbound investment limits can preserve the country’s technological advantage. But simply thwarting China will do nothing to spur the long-term innovation the United States needs to ensure its future security and prosperity. Now more than ever, Washington must understand that knowledge is power—and that it must be cultivated at home.
AMY ZEGART is the Morris Arnold and Nona Jean Cox Senior Fellow at the Hoover Institution, a Senior Fellow at Stanford University’s Human-Centered AI Institute, and the author of Spies, Lies, and Algorithms: The History and Future of American Intelligence.
More by Amy Zegart
Topics & Regions: United States China Economics Politics & Society Education Science & Technology Security Defense & Military Intelligence Strategy & Conflict U.S. Foreign Policy Artificial Intelligence Elon Musk
Ukrayna Devlet Başkanı Zelenskiy ve beraberindeki heyet,