Navigating AI in a Geopolitical Innovation Race

Political actors may adopt the arms race framing to preserve leadership or highlight capability gaps,¹ but this is more rhetoric than description. Such framing risks encouraging speed of action over safety and ethics.² Amidst rising geopolitical tensions and contestation of civil rights and democracy, AI policies should go beyond nationalistic visions of technological capabilities and reflect openness to cooperation as well as competition.³

While the concept of an innovation race is established in economics, it requires adaptation for the AI case. The framing of a geopolitical innovation race helps normalize debates around a technology whose capabilities are often dangerously exaggerated. Much of the hype relates to geopolitical and technical shifts, such as the rise of China as a major power,⁴ and advances in cloud computing, developer bases, and AI training data. These factors accelerated research and development (R&D), with over half of all existing AI patents filed between 2013 and 2019.⁵As global politics has become more rivalrous, AI has become a status symbol for competing states, with researchers and firms invoking nationalism to secure funding and political backing. While other technologies such as cloud computing have also been ‘geopoliticized’,⁶ the increasing involvement of states in centrally funding and regulating AI has meant that this research field in particular has been linked to security and power.

The authors of this chapter agree with previous criticism that the ‘arms race’ concept oversimplifies the global environment by depicting states as solely security-driven and the system as purely competitive.⁷ However, while the alternative notion of an ‘innovation race’ may be more accurate, it remains too narrowly economic, missing AI’s geopolitical dimension. Drawing on prior debates in innovation economics, critical geopolitics, and international relations, we identify four crucial characteristics of a geopolitical AI innovation race: payoff structure, actor networks, motivations, and social construction of technology.

By examining the arms race framing our study highlights the ambiguous relationship of innovation and warfare. While geopolitical struggles foster national R&D silos and tech war rhetoric,⁸ AI development also relies on collaborative innovation and instrumental cooperation. Building on our previous publication, Arms Race or Innovation Race? Geopolitical AI Development,⁹ we advance the concept of a ‘geopolitical innovation race’, defined by the above four characteristics. Our analysis¹⁰ of 34 policy documents from China, the U.S., and the EU shows how these dynamics play out in practice and co-constitute the geopolitical innovation race.

The Uncertain Pay-Off Structure of a Geopolitical Innovation Race

A geopolitical innovation race may lead to negative or zero-sum outcomes, resembling an arms race. In such scenarios, competition – especially military – is accelerated at the expense of ethically aligned cooperative development, while civilian AI applications are framed as tools for relative advantage.¹¹ Expertise and talent, patents, and research output are treated as key state capacities shaping geopolitical performance.¹² From the ambitious U.S. semiconductor production goals to China’s incentive policies for early investment, the perception of a ‘first mover’ advantage reinforces this logic.¹³

However, in practice, the ‘race’ for AI development also allows for absolute gains through cooperation. While international standards and institutions such as the Organisation for Economic Co-operation and Development (OECD) are often promoted as frameworks for cooperation and ‘fair competition’, collaboration is nonetheless framed strategically along ‘issue-based’ lines. For example, despite vocal support for cooperation and fairness, the EU has stated that when it comes to AI developments, it will “push back where (...) values are threatened”.¹⁴ Yet such a perspective overlooks that cooperation does not remove competition but rather enables fair exchange, regulation, and shared gains. This approach is – with certain exceptions – promoted by the EU, while the U.S. continues to stress leadership and define success in terms of struggle for relative advantages.

Collaborative Actor Networks within Technopoles

In an arms race, states act as coherent entities, but in an innovation race they frame themselves as governing collaborative networks. The EU stresses input from ‘all stakeholders’ and highlights AI as a “complex enabling ecosystem”.¹⁵ The U.S. depicts its ecosystem as empowered by free-market capitalism¹⁶ and “the envy of the world”.¹⁷ China emphasizes intrastate collaboration, linking ministries, provinces, and leading enterprises, while fostering AI clusters and ‘national champions’.¹⁸ Despite this emphasis on cooperation, states situate themselves in global competition, with collaboration being limited by intellectual property and patents, reflecting the assetization of AI.¹⁹

While the EU highlights thematic clusters of collaboration,²⁰ the U.S. and China establish geographically bounded hubs.²¹ Global AI talent acquisition and international cooperation suggest boundless networks, yet actors remain tied to local industrial bases and supply chains.²² Whereas the U.S. stresses removing barriers to innovation,²³ EU bordering processes emphasize regulation and liability across the AI life cycle,²⁴ calibrating territory through domestic reorganization and regulation of everyday practices of design and development.²⁵

Security Concerns Complementing Economic and Status Motivations

Unlike an arms race, state strategies in a geopolitical innovation race combine overlapping economic, status, and security concerns.²⁶ The U.S. highlights AI’s potential to enhance both security and economic growth,²⁷ presenting the two as intertwined. China similarly stresses global leadership and economic spillovers, with military-civilian fusion framed as a broad integration.²⁸

In the EU, knowledge is linked to security, while in the U.S. it is linked to scaling leadership and to swarm intelligence in China.²⁹ Knowledge thus becomes a geopolitical asset, its scale shaped by local orders and global hegemonic ideas.³⁰

Applying AI: Weaponizing Technology and Contested Understandings

If AI development were a true arms race, AI would have to be a weapon. Such framings appear occasionally, with Chinese documents describing AI weapons such as “killer bees”.³¹ U.S. texts also mention ‘AI-enhanced capabilities’,³² but AI is more often framed as a key technology with broad societal and economic implications. The EU stresses AI’s potential for social change, China for economic transformation, and the U.S. for growth. Across all three, AI is presented as an instrument of national interest rather than a ‘super-intelligence’ or a common good.³³

The transformative potential of AI is tied to a non-linear view of R&D, where disruptive technologies require experimentation and adaptation. The U.S. highlights ‘risk-taking’ and ‘learning by failing quickly’,³⁴ while China promotes top-down ‘rapid development’.³⁵ Each technopole shapes AI’s meaning and parameters of success.

Further, governmental texts discuss AI risks, from job losses and bias to the loss of human control. The EU frames AI as opaque and liability-prone, using this risk-oriented approach to promote its own ‘brand’ of AI.

Navigating AI Research and Development in a Geopolitical Innovation Race

All three technopoles seek to strengthen their position by aligning state, industry, and academia. Configurations vary, with China advancing a state-led model, the EU emphasizing regulation, and the U.S. promoting an ecosystem approach. Yet in their policy documents all de-emphasize transnational ties. The arms race metaphor itself nationalizes innovation while still allowing cooperation across networks, stabilizing technopoles as actors in global competition.

Today’s wars and ‘tech war’ narratives highlight how competition over general-purpose technology intersects with security.³⁶ Still, global supply chains and knowledge diffusion constrain inward orientation. By reconstructing how technopoles imagine the race, we can explain their behavior. Framing AI competition as a security issue is interwoven with economic and ideological aims. This mirrors wider race dynamics in quantum computing, semiconductors, and other emerging technologies.³⁷

Our findings suggest several lessons for policy and debate. First, the arms race metaphor should be avoided, as it oversimplifies AI development and risks fueling escalation. Instead, framing AI as a geopolitical innovation race captures both competition and cooperation. Second, rhetoric matters: reflecting a wider ‘return to geopolitics’, nationalistic ‘race talk’ aims not only to describe reality but also to create it, leaving little room for optimism (e.g., for arms control of autonomous weapons). Finally, strengthening cooperative frameworks through international standards and regulation can moderate rivalry and promote more responsible innovation.