Arms race

Lore & Background

From 1897 to 1914, a naval arms race between the United Kingdom and Germany took place. British concern about rapid increase in German naval power resulted in a costly building competition of Dreadnought-class ships. This tense arms race lasted until 1914, the outbreak of the First World War. After the war, a new arms race developed among the victorious Allies in the early 1920s, which was temporarily ended by the Washington Naval Treaty of 1922. In addition to the British and Germans, contemporaneous but smaller naval arms races also broke out between Russia and the Ottoman Empire; the Ottomans and Greece; France and Italy; the United States and Japan in the 1930s; and Brazil, Argentina, and Chile.

The nuclear arms race occurred during the Cold War, a period of geopolitical tension between the Soviet Union and the United States that began in the late 1940s. The nuclear arms race was not a main cause of the Cold War's start but rather escalated after it had already begun, serving as both a consequence and a driver of the conflict.

Reader's Guide

Arms races represent a central concept in international relations, with scholarly debate over whether they correlate with war. Charles Glaser argues that numerous cases of arms races were suboptimal, as they entailed a waste of resources, damaged political relations, increased the probability of war, and hindered states in accomplishing their goals. However, arms races can be optimal for security-seeking states in situations when the offense-defense balance favors offense, when a declining state faces a rising adversary, and when advances in technology make existing weapons obsolete for the power that had an advantage in the existing weaponry. The term also extends beyond military contexts: an evolutionary arms race is a system where two populations are evolving in order to continuously one-up members of the other population, related to the Red Queen's Hypothesis. In technology, there are close analogues to the arms races between parasites and hosts, such as the arms race between writers of computer viruses and antivirus software, or spammers against Internet service providers and E-mail software writers. More generically, the term is used to describe any competition where there is no absolute goal, only the relative goal of staying ahead of the other competitors in rank or knowledge. An arms race may also imply futility as the competitors spend a great deal of time and money, yet with neither side gaining an advantage over the other.

Did You Know?

• The existing scholarly literature is divided as to whether arms races correlate with war.
• The pre–First World War naval arms race between the United Kingdom and Germany lasted from 1897 to 1914.
• The nuclear arms race during the Cold War was one of the main causes that began the Cold War.
• An artificial intelligence arms race is sometimes placed in the context of an AI Cold War between the US and China.

The AI Cold War and Geopolitical Stakes

The AI arms race is frequently framed as a new Cold War, particularly between the United States and China. Since the mid-2010s, analysts have identified intensifying geopolitical and military tensions as the engine driving a competition for superior artificial intelligence and its military applications. The stakes are articulated in sweeping terms: Russian President Vladimir Putin declared that whoever leads in AI will "rule the world," while researchers such as Leopold Aschenbrenner and Adrian Pecotic have cautioned that the race toward artificial general intelligence could fundamentally restructure the balance of global power. The scope of this contest extends well beyond any single weapon platform, encompassing surveillance capabilities, autonomous weapons, decision-making architectures, and cyber operations. The underlying logic mirrors earlier arms races over nuclear and conventional technologies: each side pursues a strategic or tactical edge, and the perceived urgency of being first generates pressure that accelerates development across every front simultaneously.

Defining "Autonomous": A Terminology Problem

One of the most persistent obstacles in discussing the AI arms race is the absence of a shared definition for what qualifies as a lethal autonomous weapon system. The word "autonomous" shifts meaning depending on which scholar, nation, or organization is doing the talking, and no universally accepted definition has been agreed upon among countries. In popular discourse, LAWS are often branded "slaughterbots" or "killer robots," evoking machines that independently identify and eliminate human targets with zero human involvement. In practice, however, most systems still retain a "human in the loop," meaning a person remains embedded in the decision chain. This gap between public perception and technical reality complicates policy debates and international negotiations. More broadly, any contest for superior AI capability can be loosely labeled an "arms race," and the advantages gained in military applications frequently spill into economic sectors, echoing the dual-use patterns observed in every major arms competition throughout history.

Safety Risks and the Black Box Problem

The competitive pressure inherent in an arms race creates a dangerous incentive structure: development teams are pushed to cut corners on safety in order to claim a first-mover advantage. When one group appears close to a breakthrough, rivals feel compelled to ignore precautions and deploy systems that are not fully ready, raising the probability of critical failures and unintended consequences. Some observers argue that even using the word "race" amplifies this dynamic. A deeper technical concern is the so-called "black box problem," where AI systems—often built on complex deep learning models—operate in ways that remain opaque to their own users and to outside observers. The Holy See has raised alarms on multiple fronts. Pope Francis warned that removing direct human involvement from military operations dulls the perception of devastation and erodes accountability, while also asserting that AI-based weapons carry the potential to threaten the survival of entire regions or humanity itself. Pope Leo and other ethicists have called retaining human control over autonomous weapons a non-negotiable requirement.

National Investments and Military Programs

The competition is visible in concrete spending and research output. In the United States, former Secretary of Defense Chuck Hagel outlined a "Third Offset Strategy" in 2014, arguing that rapid AI advances would define the next generation of warfare. Department of Defense investment in AI, big data, and cloud computing grew from $5.6 billion in 2011 to $7.4 billion by 2016, while private-sector investment reportedly reached around $70 billion per year by 2018. The 2019 Interim Report of the National Security Commission on Artificial Intelligence confirmed AI as critical to maintaining U.S. technological and military superiority. Specific programs illustrate the ambition: the Sea Hunter autonomous warship is designed to operate at sea for extended periods without a single crew member and to navigate itself in and out of port. On the other side, China surged in AI research during the 2010s, publishing more AI papers than the entire European Union by 2016, and 23 percent of presenters at the 2017 AAAI conference were Chinese. Eric Schmidt predicted China would lead in AI by 2025.

Gallery