NBER study reveals Chinese universities lead in critical tech patents

Every time China delivers a technological surprise, the American narrative follows a familiar script: a single firm or project leapfrogged expectations, and the US was caught napping. DeepSeek’s low-cost AI model that matched top American systems. A hypersonic missile test that a senior US general called “very close” to a “Sputnik moment.” Each episode gets treated as a one-off shock.

A new working paper from the National Bureau of Economic Research suggests that framing is not just incomplete. It’s dangerously wrong.

The university engine behind China’s patent boom

NBER Working Paper 35551, titled “Chinese Sputnik Moments?,” examines approximately 14 million domestic Chinese patents. The researchers zeroed in on technologies classified as critical by the US Department of Defense, covering everything from artificial intelligence to advanced materials and hypersonics.

The headline finding: Chinese universities are contributing more to critical technology patents than state-owned enterprises or government research labs. Fewer than 10% of these patents involve inventors with US experience or training.

Quantity without a quality trade-off

Quality metrics applied to these critical technology patents show no relative decline compared to US patents, even as the volume has climbed sharply.

The AI patent numbers are particularly striking. Since around 2020, China has surpassed the US in annual AI patent grants. By 2023, Chinese institutions were issuing nearly four times the volume, with a cumulative total exceeding 183,000 grants.

A companion NBER paper (No. 35022) found that Chinese AI patenting is more geographically distributed across the country than its American counterpart. In the US, AI innovation clusters around a small number of large private firms, mostly in a few coastal metros. In China, universities and state-owned enterprises spread the activity across a broader base of cities and institutions.

What DeepSeek actually revealed

When DeepSeek released an AI model that competed with America’s best at a fraction of the development cost, Western observers focused on the company itself. DeepSeek’s engineering talent was drawn from the same university system that the patent data highlights. The company didn’t emerge from a vacuum or from stolen trade secrets. It emerged from a domestic educational infrastructure that has been producing critical technology researchers at scale for years.

Implications for the global technology landscape

Export controls on advanced chips, which the US has escalated repeatedly since 2022, were designed to slow China’s progress at the hardware level. The patent data suggests the bottleneck may not be where Washington thinks it is. If China’s innovation engine runs on human capital cultivated domestically, restricting semiconductor access creates friction but doesn’t address the underlying dynamic.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.