RAND Wargame: AI-Driven 'Autonomous Mass' Could Shift U.S.-China Conflict Dynamics

The integration of artificial intelligence into military operations may fundamentally alter the nature of great-power competition, according to a new study by the RAND Corporation. Through a series of operational-level simulations, researchers found that autonomous systems could enable new forms of limited conflict and shift the advantages of protracted warfare.

To reach these conclusions, RAND conducted an unclassified wargame titled 'Camp(ai)gn,' which analyzed a hypothetical 2035 conflict between the U.S. and China over Taiwan. The study consisted of ten iterations played by more than 20 researchers and affiliates. The resulting paper outlines six initial observations regarding how AI and autonomous systems influence campaign planning, escalation dynamics, and the role of robotic mass in a high-end fight.

The Role of Autonomous Mass

One of the most significant findings from the simulations is the critical importance of 'autonomous mass.' The study suggests that in a protracted conflict, the ability to deploy large numbers of autonomous systems could become a decisive factor—a dynamic that researchers note could potentially favor China. This shift suggests that the industrial capacity to produce robotic systems at scale may be as important as the sophistication of the AI itself.

Furthermore, the research indicates that autonomous systems may facilitate new types of limited, low-intensity conflicts. These engagements would be fought primarily with robotic systems, potentially creating escalation dynamics and end states that differ from traditional expectations of great-power warfare.

Operational Trade-offs and Resilience

For military leadership, the adoption of AI presents a paradox of capability. According to the RAND findings, AI-enabled systems expand a commander's decision space and reshape operational trade-offs. However, this expanded capability simultaneously increases the overall decision-making burden on the commander as the speed and volume of AI-driven options grow.

The study also challenged common strategic assumptions regarding the vulnerability of AI infrastructure. The researchers found that attacking an adversary's data centers is unlikely to serve as a decisive operational blow. This suggests that AI-enabled forces possess a higher level of resilience than previously assumed, meaning that targeting the 'brain' of the AI may not be a critical vulnerability.

Strategic Implications

These findings have immediate implications for force structure planning and defense industrial capacity. As AI integration accelerates, the U.S. must weigh the value of high-end precision against the necessity of autonomous mass to avoid being overwhelmed in a long-term engagement. The realization that data centers are not a 'silver bullet' target further suggests that victory in a future conflict will depend more on the operational application of autonomous systems than on the destruction of centralized computing hubs.

Moving forward, the RAND study serves as a baseline for further hypotheses about the 2030s. Military planners will likely need to examine how allies can integrate into these new roles—ranging from self-defense to coalition production—as the operational reality of AI-driven warfare continues to evolve.