Takeaways

In analyzing these requests for proposal (RFPs), the authors find that the People’s Liberation Army (PLA) is pursuing AI-enabled capabilities across all domains. The applications include decision support systems (AI-DSS), sensor enhancement tools, data fusion algorithms, and much more.

The RFPs reflect China’s desire to generate, augment, and fuse increasing quantities of data to speed military decision-making and improve the precision and efficacy of the PLA’s operations. Specifically, the authors found requests for AI-DSS that can leverage open-source data for strategic decision-making. They also came across requests for AI-DSS to support tactical decisions, such as for targeting. While many militaries are investing in AI-DSS, these systems are of particular importance to the PLA, which views them as a means of compensating for perceived weaknesses in its officer corps.

More specifically, the authors found an emphasis on AI applications that would counter perceived U.S. military advantages. This emphasis was especially apparent in numerous requests for technologies to detect U.S. naval assets on and under the sea, as well as technologies that could counteract U.S. space-based systems.

Outside of the maritime and space domains, the PLA’s RFPs reveal it aims to acquire increasingly sophisticated surveillance and cognitive domain capabilities. RFPs for facial and gait recognition systems, digital surveillance tools capable of recovering deleted data, and technologies for generating and detecting deepfakes point to ongoing efforts to better secure military installations and develop AI-enabled psychological warfare and cognitive targeting tools.

These documents, which are published by the PLA, are strikingly explicit in their requests for sensitive capabilities, including those related to operations in the space, cyber, and cognitive domains. The level of specificity is notable and likely reflects an effort to access advanced capabilities from nontraditional, commercial vendors outside China’s traditional defense industrial base. Moreover, most of the RFPs involve relatively small budgets and short acquisition timelines—often three to six months—suggesting an emphasis on experimentation, prototyping, and rapid iteration.

The PLA’s AI acquisition efforts have several implications for U.S. policymakers and defense planners. First, Washington should respond to China’s interest in AI-enabled sensing and surveillance, especially in the maritime and space domains, with investments in counter sensing, deception, and resilience against China’s increasingly capable ISR systems.

Second, the United States should prepare to counter emerging Chinese AI-DSS while pursuing dialogue with China on the responsible use of these systems to reduce the chances of miscalculation and escalation. While it is unlikely that Washington and Beijing will make binding commitments, continued discussions could help establish technical minimum standards and norms around the use of these systems.

Third, the PLA’s interest in AI systems that ingest and process vast volumes of open-source data for strategic decision-making demands a U.S. rethink of approaches to military signaling, deterrence, and crisis management. Increasing reliance on AI to interpret global events could complicate previous approaches to escalation control.

Fourth, China’s military appears ready to use AI to greatly expand its surveillance and information-manipulation capabilities, including more systematic use of deepfakes. Washington should counter with stronger public awareness efforts and public–private collaboration, as well as develop technical standards and tools to detect and blunt these tactics.

Fifth, China’s embrace of prototyping and rapid acquisition timelines should further motivate U.S. leaders to support defense acquisition reform, fund a diverse research portfolio, and set the conditions for rapid experimentation and responsible AI adoption in the operating forces.

Sixth, access to advanced AI hardware developed by U.S. companies enables the PLA’s modernization. Evidence that the PLA is seeking advanced U.S.-designed semiconductors, as well as leveraging large language models trained on U.S. GPUs, suggests that relaxing export controls will facilitate China’s development and use of AI-enabled C5ISRT technologies.

Finally, the breadth and diversity of the PLA’s AI wish list reinforce the importance of conducting sustained open-source monitoring to track capability development, detect shifts in priorities, and identify early signs of operational deployment. Such analysis helps to rightsize current challenges to U.S. national security.