For years, Chinese leader Xi Jinping has sought to enlist civilian enterprises to help modernize the People’s Liberation Army—a concept known as “military-civil fusion.” In artificial intelligence, that campaign appears to be bearing fruit.

According to research from Georgetown University’s Center for Security and Emerging Technology (CSET), more than 85% of winners of AI-related defense contracts in China are not traditional military-affiliated entities but civilian universities and private tech firms—many relatively new and not under U.S. sanctions.

The data indicates that, in recent years, the Chinese military has moved beyond its traditional network of state-owned defense contractors and affiliated research institutes, tapping hundreds of suppliers—including private companies and civilian universities—in an effort to integrate AI into its operations and weapons systems.

A prominent example is Shanghai Jiao Tong University (SJTU), which since 2023 has won numerous defense contracts for projects including adaptive weapon systems, underwater drones, and AI for tracking fast-moving targets.

Other key contributors include iFlytek Digital, a voice-AI specialist, and Sichuan Tengden Sci-Tech Innovation, a drone manufacturer.

This integration of civilian expertise gives China a potential strategic edge in military AI innovation compared with the United States, where collaboration between the private sector and defense agencies is less common. China’s approach raises fundamental questions for U.S. policymakers—whether to expand sanctions or find ways to cooperate while maintaining a technological competitive edge.

A Chinese Advantage and American Constraints

Both the U.S. and Chinese militaries have tried to leverage the knowledge and innovative energy of universities and the private sector, but the data suggests China has done so more systematically. Security analysts say this could give Beijing an advantage in the difficult task of integrating AI into national defense—complicating U.S. efforts to prevent China from developing technologies that might yield a military edge.

“In this dataset alone, the sheer ambition of what they are trying to do is striking,” said Cole McFall, a senior research analyst at CSET who helped compile the data. “The breadth of these technologies shows the limits of our ability to slow or constrain Chinese military modernization.”

The PLA’s Enthusiasm for AI

The People’s Liberation Army showcased its enthusiasm for AI during a military parade in Beijing on Wednesday, displaying information-warfare and unmanned combat units as Russian leader Vladimir Putin and North Korean leader Kim Jong Un watched from the stands. The display featured attack drones, unmanned boats, and robotic dogs—the kinds of assets military researchers expect to be pivotal in future AI-enabled conflicts.

To advance military-civil fusion, China allows public bidding on a portion of its defense contracts, including for sensitive systems that the United States and many other militaries keep classified.

Private Firms and Civilian Universities

By extracting vendor information from nearly 3,000 AI-related contract-award notices published by the Chinese military in 2023 and 2024, CSET identified more than 300 companies that won multiple bids. Private firms, civilian universities, and other entities not traditionally considered part of China’s defense industry accounted for over 85% of these multiple-bid winners and took the majority of contracts—the vast majority not subject to U.S. sanctions.

Many private firms are relatively young, most founded after 2010. That includes the top bid-winner, iFlytek Digital, a subsidiary spun out of AI speech-recognition firm iFlytek, which the United States blacklisted in 2019 for its role in state surveillance of religious minorities. Most of iFlytek Digital’s roughly 20 contracts—which it now pursues as a separate entity not on the U.S. blacklist—involved data processing and analysis.

Although the PLA still relies heavily on state-owned defense giants to build drones, contract data shows private firms are making inroads. Among them is Sichuan Tengden Sci-Tech Innovation, maker of the TB-001 “Twin-Tailed Scorpion”, a heavy attack drone spotted in recent years flying near Okinawa and Taiwan. Founded in 2016, the company won seven contracts, mostly for drone leasing and flight tests.

CSET’s dataset did not include secret procurements for AI systems the PLA may wish to conceal, and many notices did not specify end-users. The researchers also relied on companies’ self-declared affiliations, meaning some ties to larger state contractors may be undisclosed.

Even so, contracts like those awarded to Shanghai Jiao Tong University underscore the PLA’s aggressiveness in seeking outside expertise to compensate for shortcomings, said Alex Joske, an Australia-based China analyst who was not involved in CSET’s research but has studied how Chinese universities work with the military. “In an area like AI, the military is not at the cutting edge,” Joske said. “It has made significant progress in reducing barriers for civilian universities to contribute to defense needs.”

Maritime “Kill Webs”

One clear example is a maritime kill-web project developed and tested by researchers at SJTU’s School of Electronic Information and Electrical Engineering, working with engineers from a national laboratory and a state defense institute in Wuhan. In a paper, the researchers said the concept draws on a U.S. theory of decision-centric warfare, which emphasizes making swift, effective choices to keep adversaries off balance.

They describe using adaptive algorithms to build a model of a naval battlespace—drawing on radar, sonar, and other sources—and then coordinating assets such as missile systems and drones to eliminate targets, with continuous adjustments as conditions change. In simulations of an enemy missile attack, the system generated kill webs comprising dozens of drones in an average of 2.26 seconds in a high-intensity scenario with 31 enemy targets. Commanders could select among options or configure the system to execute what it deemed the most suitable automatically.

A week after winning a contract to build the system for the PLA, Shanghai Jiao Tong won a second contract to develop a supporting database. This marked its seventh publicly reported defense contract for AI-related systems since early 2023; the university would go on to work on seven more by the end of 2024.

Alongside the kill-web project, the university was tasked with helping the military track fast-moving targets using multi-layer AI models, rapidly generate designs for underwater drones, and make drone swarms more sensitive to changes in radio frequencies.

According to Sam Bresnick, a CSET researcher, the PLA’s use of suppliers like SJTU poses hard choices for Washington. The U.S. will need to determine the investments required to sustain competitiveness—and whether to impose sanctions on a much broader set of Chinese entities or preserve cooperation where possible