Why China is winning in tech—and what the US is overlooking

In January 2025, DeepSeek jolted the global race in artificial intelligence (AI) with a model that rivaled American systems and was built by a team entirely educated in China. Soon after, Manus drew fresh attention as another Chinese-founded AI breakout. These companies reflect a new reality: China’s frontier AI is being built by talent that never left the country.

The core of the U.S.-China technology competition is not chips or subsidies, but the cultural and structural conditions that produce leading technical talent. America’s advantage in this deeper competition is eroding.

The missing human infrastructure

The emergence of frontier AI innovation from domestically trained Chinese talent exposes a critical gap in how the United States understands the technological competition with China. Most policy analysis focuses on industrial subsidies, export controls, and semiconductor supply chains, but often overlooks the deeper human infrastructure: the cultural, educational, and social conditions that produce AI talent at scale and increasingly at the frontier.

To understand why, I recently asked students at Syracuse University to analyze media using sociological imagination. One student discussed a film featuring Wendy Wu, a Chinese American teenager, focusing on how she tried to fit into American teen culture by downplaying academics to be popular.

One discussion centered on how academic achievement is perceived among American teenagers. Several students said candidly that in many American schools—especially large public ones—studying hard and excelling, particularly in math and science, is often seen as uncool. A student offered an exception: in schools with large Asian American populations, being hardworking is the norm and socially valued.

That exception, however, carries a troubling corollary. Research shows that as Asian enrollment rises in affluent suburban school districts, white enrollment declines. The academic culture valued in those settings is precisely the culture white families are fleeing.

From culture to scale: China’s talent pipeline

This maps onto what is happening inside Chinese school systems today. Academic excellence, especially in mathematics and science, is not socially marginal or an individual pursuit. Instead, it is often a precondition for recognition and status and an important pathway to social mobility.

In the United States, by contrast, intellectualism—especially in science, technology, engineering, and mathematics (STEM)—can be socially ambivalent or even stigmatized in certain contexts. Ninety-three percent of Americans report experiencing some level of math anxiety, and many people comfortably self-identify as “not a math person,”  shaping the majors students choose and ultimately the careers they pursue.

These cultural norms translate directly into educational outcomes and workforce supply. China now produces STEM talent at a scale that is difficult for the United States to match. The National Science Board’s Science and Engineering Indicators 2024 confirms that China has now surpassed the United States as the top producer of STEM doctoral degrees globally. At the undergraduate level, China’s pipeline is even more striking: 12.22 million students are projected to graduate from Chinese universities in 2025 alone—a record high, and more than double the total degrees awarded at all levels of higher education in the United States.

A necessary caveat: quantity is not the same as quality

A common counterargument is that scale does not automatically translate into innovation. China’s education system, while highly effective at producing large numbers of graduates, has long been criticized for emphasizing standardized testing and rote learning. For example, China’s national college entrance exam is among the most high-stakes tests in the world, with students routinely describing a severe psychological toll.

These concerns are reinforced by labor market signals: youth unemployment in China has remained elevated in recent years, with official urban rates exceeding 20% in 2023 before methodological revisions, and analysts pointing to structural mismatches between graduate skills and labor market demand.

Global university rankings suggest a more nuanced picture. While U.S. institutions continue to dominate the top tier, China’s leading universities have rapidly climbed in global standings. Tsinghua University and Peking University are now consistently ranked among the top 20 globally, particularly strong in engineering and computer science.

In a system of such scale, even a small percentage of top-tier talent translates into a large absolute number of highly capable engineers and researchers. The emergence of DeepSeek and Manus—whose founders and core technical teams are largely domestically trained—suggests that China is not only producing talent at scale but is increasingly capable of cultivating high-end innovators within its own system.

The key question is not whether average quality varies, but whether the system produces enough top-tier talent to sustain technological leadership. Increasingly, the answer appears to be yes.

The US model: strengths and emerging constraints

The United States has historically relied on a hybrid model: strong domestic institutions combined with a steady inflow of high-skilled immigrants. For decades, this model worked extraordinarily well. Global AI talent data show that a substantial share of top AI researchers received their undergraduate education in China, with many later working in the United States. Temporary visa holders earned over half of U.S. doctoral degrees in key STEM fields, including 59% in computer science and 60% in engineering in 2021.

This model, however, faces mounting structural constraints—from tightening immigration pathways to declining enrollment for Chinese and international students. Visa challenges, heightened scrutiny of Chinese nationals in STEM fields, and the deterioration in U.S.-China relations have all made American universities less attractive. At the same time, the rapid rise of Chinese institutions like Tsinghua and Peking University in global rankings has reduced the premium that once made studying abroad feel necessary.

The more urgent and underexamined question is not only how to restore the external talent pipeline, but what is happening inside the domestic one. The U.S. education pipeline faces serious challenges. National assessments show that American students’ math performance has declined significantly, with scores falling to levels last seen two decades ago. Meanwhile, broader indicators suggest a shifting global balance in science and engineering output. The National Science Board notes that China has surpassed the United States on several aggregate measures of patents and research outputs.

A different policy conclusion

The United States now faces a dual challenge that cannot be resolved by immigration policy alone. Importing talent is a short-term workaround; it does not address the structural and cultural conditions that determine how many Americans pursue and sustain technical careers in the first place. Three interventions, largely absent from current policy discourse, deserve serious attention.

First, dismantle the tracking systems that close the gateway to STEM careers before students are old enough to choose. Differentiated math curricula do not begin in middle school—they begin as early as elementary school, where roughly four in 10 elementary school principals report grouping students by achievement level in math, setting trajectories that shape the rest of a child’s academic career. Research consistently shows that Black, Latino, and low-income students are disproportionately assigned to lower tracks—and separately, that Black and Hispanic children are 66% and 47% less likely than white students, respectively, to be referred to gifted programs, even after controlling for test scores. These compounding disadvantages cannot be undone by college remediation programs. Reform is not about lowering standards but about ensuring that the pipeline is not structurally narrowed at the very moment children are forming their sense of what is possible.

Second, confront the broader cultural aversion to academic competition that runs through American society at every income level. The white flight evidence is instructive here not only as a story about race, but also as a story about values: when high-achieving school cultures take hold, a significant share of American families opt out rather than adapt. Meanwhile, girls across all demographics continue to be socialized away from technical fields because of persistent social signals about who those fields are for, signals that take hold as early as first grade. These are not separate problems. They are symptoms of a society in which athletic competition confers social prestige and academic competition provokes social anxiety—a cultural asymmetry whose consequences are manifesting in the national talent pipeline.

Third, build the institutional infrastructure that makes academic achievement publicly visible and socially prestigious. Culture does not change through exhortation alone; it changes when institutions create new incentives and signals. This means putting nationally celebrated science and math competitions on par with athletic championships, school-wide recognition of academic achievement that rivals the attention given to sports, and federal investment in programs that make STEM excellence visible in communities where it has historically been invisible. President Barack Obama made the underlying aspiration plain at the White House Science Fair in 2016: “As a society, we have to celebrate outstanding work by young people in science at least as much as we do Super Bowl winners. Because superstar biologists and engineers and rocket scientists and robot-builders… they’re what’s going to transform our society.” He said this over a decade ago, but the institutional infrastructure to make it real has not been built.

Industrial policy, immigration reform, and export controls are necessary. But they address symptoms, not sources. The deeper competition is cultural and structural: whether American society decides that mathematical excellence is as admirable as athletic excellence, and whether American institutions ensure that the path to technical excellence is not narrowed at every turn by race, income, or zip code. Until both change, no amount of semiconductor subsidy will close the gap that matters most.