This report uses procurement records published by the Department of Defense (DOD) and the People’s Liberation Army (PLA) between April and November of 2020 to assess, and, where appropriate, compare what each military is buying when it comes to artificial intelligence (AI). Specifically, it analyzes the U.S. and Chinese militaries’ respective purchases of equipment, services, and research and development (R&D) activities in seven categories commonly identified as priority areas for military AI: intelligent and autonomous vehicles; intelligence, surveillance, and reconnaissance (ISR); predictive maintenance and logistics; information and electronic warfare; simulation and training; command and control (C2); and automatic target recognition (ATR). Our key findings include:
- Procurement data between April and November of 2020 indicates that U.S. and Chinese military forces are devoting comparable levels of attention to a similar suite of AI applications, particularly in regard to AI for intelligent and autonomous vehicles, as well as ISR tools. Neither military appears to have made significant investments in procuring AI for command and control, though this may be a result of our data and research approach.
- Despite the fact that both the United States and China have relatively concentrated defense-industrial bases, each country’s ecosystem of military AI suppliers (as reflected in the procurement records we reviewed), appears to be more distributed and composed of smaller vendors. The 300 U.S. military AI contracts in our dataset are distributed among 249 unique vendors. Notably, only 36 vendors were awarded multiple contracts and just eight won three or more contracts. Most of the remaining vendors were small, bespoke defense companies. The 119 Chinese military AI contracts in our dataset were distributed across 102 unique vendors. Furthermore, universities accounted for a much larger share of AI contracts in China (14 percent) than the United States (3 percent).
Our research methodology, data, and overall analysis all have limitations. The aforementioned list of seven application areas for military AI development is far from exhaustive, and in fact, a significant number of procurement records could not be neatly categorized into these buckets. The procurement records themselves are also limited, providing a snapshot of military purchasing patterns during an extraordinary period of time when the COVID-19 pandemic disrupted supply chains and business operations. The findings from this study may therefore not be generalizable to broader military AI procurement patterns. Also, any comparison made between U.S. and Chinese military AI purchases must account for their vastly different military bureaucracies, defense-industrial bases, and procurement systems, all of which inherently shape how each organization develops, obtains, and potentially uses emerging technologies such as AI going forward.
Even with these limitations, our study provides a unique perspective on military AI procurement as well as a rich set of concrete examples of U.S. and Chinese AI-enabled technologies and capabilities that are in relatively advanced stages of research and development or are already ready for deployment. The findings of this report, when combined with other types of data and analysis, could provide useful insights for U.S. defense planners as they think about technological competition with China and examine how AI is being integrated into core U.S. and Chinese military functions, doctrine, and operational concepts for future warfare.
This report was updated on September 5, 2023.