AI Innovation & the Security Risks of Technological Interdependence

Sasha Allen, The University of California – Santa Barbara

---

Abstract

Artificial Intelligence technologies have boomed into society as a dual-use technology, serving both commercial innovation and national security interests. This paper examines how states balance the economic benefits of AI with the security risks it creates, emphasizing the strategic role of interdependence in global technology supply chains. Using frameworks of weaponized interdependence and technological nationalism, the study analyzes three cases: U.S. export restrictions on Nvidia chips, sanctions on Huawei’s semiconductor access, and Dutch controls on ASML lithography machines. These cases demonstrate how control over critical AI hardware and software can be leveraged as tools of statecraft, allowing great powers to shape adversaries’ technological capabilities while their own advancing economic interests. Because AI blurs the line between commercial and security applications, it creates new challenges for governance, supply chain management, and strategic competition. 

1. Introduction

Artificial Intelligence (AI) has emerged as one of the defining technologies of the 21st century, reshaping economies, societies, and international politics. Unlike traditional military technologies, AI is primarily developed in the private sector, often across borders, and serves a wide variety of uses from commercial software and cloud computing to surveillance, cyber operations, and autonomous military systems. This dual-use nature creates a dilemma: technologies that drive economic growth and innovation can simultaneously pose significant national security risks. The “dual-use dilemma” has become a central concern for policymakers, regulators, and scholars of international relations as states attempt to strike a balance between technological leadership and security imperatives.

The dual-use dilemma is especially prevalent in AI because of its centrality to both economic competitiveness and military power. Advanced AI chips, algorithms, and software frameworks are critical to commercial innovation, yet they can also enable sophisticated intelligence, cyber operations, or weapons systems. As a result, states are increasingly treating supply chains and technological interdependence as tools of statecraft; an extension to the traditional branches of economic, military, and diplomatic. Export restrictions, investment screening, and sanctions are not merely economic moves, they are strategic levers that influence adversaries’ access to critical technologies while protecting domestic capabilities. This intersection of economics and security highlights a broader trend: the dual-use dilemma and interdependencies in high-tech industries is no longer a neutral source of economic growth, but a source of both opportunities and vulnerabilities. ​

2. Theoretical Framework

This analysis draws upon core frameworks in international relations and political economy to explain how states manage the relationship between economic integration and national security in the context of artificial intelligence. AI highlights long-standing debates about whether interdependence promotes international cooperation or creates new vulnerabilities, while also reshaping how governments perceive and respond to technological threats. By integrating theories of security-economy tradeoffs, the politics of technological threats, strategic interdependence, and technological nationalism, this article highlights the political logic that drives state intervention in global AI supply chains. Together, the literature can explain why states increasingly treat advanced AI capabilities not as neutral commercial goods, but as strategic assets that require active management, control, and, in some cases, restriction.

Security-Economy Tradeoffs in International Relations

A central debate in international relations revolves around whether economic openness promotes peace through mutual gains or creates strategic vulnerabilities by exposing states to coercion, dependence, and supply-chain disruptions. Scholars note that while interdependence can raise the costs of conflict, it can also be weaponized, allowing powerful states to exploit asymmetries for political and security advantage. Classical neorealists like Kenneth Waltz and Hans Morgenthau argue that economic ties are always secondary to security concerns, with states ultimately prioritizing survival over commercial gains. Liberal theorists, in contrast, see complex interdependence as generating mutual benefits and limiting coercion, though they acknowledge that asymmetric dependencies can still be used as instruments of power. The tradeoff between security and economy is further complicated by Robert Gilpin’s political economy theory, which suggests that advanced industries provide states not just with economic rewards but also with military potential, encouraging intervention when technological change threatens their strategic position. This logic supports strategic trade theory, which explains why governments may use industrial policy or export controls to gain an edge in high-tech sectors that are critical to national power. In the context of AI, these classic debates take on heightened urgency. The technologies that enhance productivity and innovation also underpin intelligence capabilities, cyber operations, and autonomous systems. As a result, states confront a modern variant of the traditional dilemma: openness accelerates innovation, but it also increases vulnerability. Understanding this tradeoff provides the theoretical foundation for why governments increasingly intervene in AI supply chains and adopt restrictive economic policies even at the expense of short-term commercial gain.

Technological Threat Politics

The politics of technological threat construction explains how certain technologies become framed as security priorities rather than ordinary economic goods. Securitization theory posits that issues become security threats not solely because of their objective characteristics, but because political actors successfully frame them as existential threats through speech acts, thereby justifying extraordinary measures beyond normal politics. When government agencies, defense establishments, or political leaders successfully portray a technology as existentially significant, they can legitimize extraordinary measures, such as export controls, sanctions, or restrictions on foreign investment, that would otherwise face political resistance.

AI is particularly vulnerable to this dynamic because its applications span intelligence, surveillance, cyber operations, autonomous systems, and advanced analytics. In a rapidly evolving technological environment, uncertainty encourages what Ben Buchanan calls “security imaginaries,” in which policymakers extrapolate worst-case scenarios from commercially developed technologies. For example, advances in machine learning developed for data analytics or consumer platforms are imagined as enabling autonomous weapons. As a result, AI chips, cloud compute, data architectures, and machine-learning models, originally designed for civilian innovation, are reframed as potential enablers of adversary capabilities. Once AI is securitized, domestic political authority shifts. Security agencies gain influence relative to economic ministries, enabling the state to intervene more aggressively in private-sector decision-making and global supply chains. Measures that would typically be viewed as distortions of the market are reinterpreted as necessary steps to prevent strategic vulnerability. At the international level, technological interdependence that once facilitated innovation begins to be perceived as a channel for espionage, coercion, or dependency, reinforcing the state’s impulse to restrict flows rather than deepen them. In this way, technological threat politics help explain why states treat AI as a security priority even when its economic benefits are immense. It shows how perceptions and narratives shape the governance of AI and motivate states to adopt increasingly interventionist policies toward global technology flows.

Strategic Interdependence

Strategic interdependence refers to the ways in which economic and technological links between states create both mutual benefits and asymmetric vulnerabilities. Unlike traditional views of interdependence as primarily cooperative, scholars have long noted that the distribution of dependence matters more than the existence of ties themselves. Hirschman’s classic analysis of trade politics showed that states can leverage asymmetric dependence as a form of political influence, while Keohane and Nye’s theory of complex interdependence emphasizes that unequal reliance on specific technologies, inputs, or networks can become a source of strategic power. These insights provide a foundation for understanding why global supply chains, especially in high-technology sectors like AI, produce both prosperity and geopolitical risk.

In the AI ecosystem, interdependence is deeply structural: advanced chips, fabrication equipment, cloud infrastructure, and data pipelines are all distributed across jurisdictions, firms, and regulatory systems. Yet this distribution is uneven. Certain states possess irreplaceable capabilities, while others depend heavily on access to these nodes. This concentration generates what Farrell and Newman described as “asymmetric network power,” in which states positioned at critical choke points can influence, constrain, or monitor the technological activities of others. Interdependence thus becomes a strategic resource, not a passive condition.

As geopolitical competition intensifies, states increasingly interpret technological interdependence through the lens of vulnerability rather than efficiency. Dependencies that previously facilitated innovation and global integration are now seen as potential liabilities, channels through which rivals might gain military advantages, exert economic coercion, or shape technological trajectories. This shift motivates efforts to control supply chains, secure domestic capacity, and limit adversaries’ access to strategic technologies, even when doing so reduces economic efficiency. The logic is clear: in a world where technological advantage confers both economic and strategic power, managing interdependence becomes central to national security strategy. Consequently, states increasingly pursue policies aimed at reducing asymmetric vulnerabilities while exploiting areas where they hold structural advantage.

3. Case Analysis

The following section applies the theoretical framework to three empirical cases that illustrate how states leverage positions within globally integrated AI supply chains to advance security objectives. Each case represents a different strategic node in the AI ecosystem – chip design, semiconductor manufacturing, and advanced lithography equipment – where asymmetric interdependence grants states the ability to shape technological access and constrain rivals’ capabilities. By examining U.S. export controls on Nvidia’s AI chips, restrictions on Huawei’s semiconductor supply, and Dutch limits on ASML’s lithography equipment, this analysis demonstrates how dual-use technologies transform commercial relationships into instruments of statecraft. Taken together, these cases show how states operationalize technological nationalism and weaponized interdependence in practice, illuminating the concrete mechanisms through which economic tools are deployed to produce strategic outcomes in the emerging AI order.

U.S. AI Chip Export Bans

The United States’ export controls on advanced AI chips represent one of the clearest examples of how states use control over strategic technologies to manage the security risks of interdependence. Beginning in 2022, the Bureau of Industry and Security (BIS) introduced a series of measures restricting the sale of high-performance GPUs, most notably Nvidia’s A100 and H100 chips, to China. These chips are foundational to training and deploying large-scale machine-learning models, giving the U.S. and its allies substantial structural leverage over global AI development. The rationale behind the controls, as the Biden administration articulated, was to prevent U.S.-origin technologies from “fueling the military modernization” of a strategic competitor. 

In practice, the restrictions highlight the logic of weaponized interdependence: rather than attempting to regulate specific uses of AI, the U.S. targeted the hardware that made advanced AI possible in the first place. By limiting the export of these chips to a state seen as a threat to U.S. supremacy, policymakers aimed to slow China’s progress in areas such as military AI applications, intelligence analysis, and autonomous systems. This approach reflects a broader shift in U.S. strategy, moving from reactive technological safeguards to proactive restriction of adversaries.

The implementation of the controls also reveals the dynamic nature of power in global supply chains. When Nvidia introduced modified chips designed to comply with the initial restrictions, U.S. regulators expanded the rules to include these workarounds, signaling a willingness to adapt policy as firms adjust. While these measures effectively constrained China’s access to the most advanced AI accelerators, they also carried economic and strategic costs. Nvidia publicly acknowledged that the controls significantly reduced its sales in China and warned that the restrictions could incentivize accelerated domestic alternatives within China’s semiconductor ecosystem.

More broadly, the U.S. chip export bans illustrate the political logic behind securitizing AI supply chains. Policymakers accepted the economic tradeoffs of lost revenue, reduced market share, and potential long-term erosion of U.S. industry dominance in order to mitigate perceived strategic vulnerabilities. The case shows how dual-use technologies push states toward restrictive policies even when those policies disrupt globally integrated industries. Ultimately, the export bans demonstrate how states operationalize strategic interdependence by leveraging critical controls, reshaping incentives for firms, and using economic tools to influence the technological capabilities of rivals.

Huawei Semiconductor Restrictions

U.S. restrictions on Huawei’s semiconductor access represent a deeper and more targeted application of specific strategy, aimed not just at a sector but at a single company positioned at the center of China’s AI and telecommunications ecosystem. Beginning with the 2019 placement of Huawei on the Entity List (which required U.S. firms to obtain licenses before exporting certain technology to the company), the United States progressively tightened controls to prevent the company from acquiring advanced chips or the tools needed to produce them. Unlike the broader GPU export bans, this case demonstrates how states can weaponize supply-chain dependencies to degrade a firm’s long-term innovative capacity, not merely constrain short-term access to specific technologies.

Huawei’s reliance on foreign semiconductor manufacturing, particularly Taiwan’s TSMC, and on U.S.-origin design tools such as EDA software made the company uniquely vulnerable. U.S. policymakers strategically leveraged these dependencies by extending export controls extraterritorially: any firm using U.S. technology, even indirectly, required a license to supply Huawei with chips. This move reflected a more expansive interpretation of U.S. jurisdiction, illustrating how control over upstream nodes enables states to reach deep into global supply chains. The strategic intent behind the restrictions was twofold. First, the United States aimed to curtail Huawei’s ability to dominate global 5G infrastructure, a leadership position with profound intelligence and security implications. Second, by limiting the company’s access to advanced semiconductors, the U.S. sought to slow China’s progress in AI-enabled telecommunications, sensing, and military communications systems. The policy thus aligns with a broader logic of tech nationalism: safeguarding domestic and allied networks while constraining a competitor’s technological trajectory.

The consequences for Huawei have been mixed but significant. On one hand, the company’s smartphone and advanced chip divisions suffered substantial setbacks, and its access to cutting-edge fabrication capacity was effectively severed. On the other hand, the restrictions accelerated China’s push toward semiconductor self-reliance, prompting state-led investment in domestic chip design and manufacturing. Huawei’s announcement of new 7nm-class chips produced with Chinese tools, despite their lower efficiency and yield, signals both the partial success of U.S. restrictions and the long-term limits of using economic bottlenecks as a sustained strategy. Ultimately, this case demonstrates how coercive power operates not simply through denial but through shaping the technological ecosystem in which firms compete. The Huawei restrictions show that weaponized interdependence can impose real costs on targeted firms, yet it also carries the risk of galvanizing domestic innovation, reducing future leverage, and fragmenting global technology supply chains.

Dutch ASML Restrictions

The Dutch government’s restrictions on ASML’s exports of advanced lithography equipment illustrate how middle powers can become pivotal actors in great-power technological competition. Unlike the U.S. cases, which involve direct control over proprietary chips or software, Dutch leverage stems from ASML’s near-monopoly over extreme ultraviolet (EUV) lithography systems, the machinery required to produce the world’s most advanced semiconductors. Because no other firm can manufacture EUV tools at comparable levels of precision, states seeking to shape global chip production must work through the Netherlands, giving a relatively small country outsized influence in the semiconductor ecosystem.

Beginning in 2019, the Dutch government, under sustained U.S. diplomatic pressure, declined to issue export licenses for ASML’s EUV systems to Chinese customers. In 2023, the scope of controls widened to include certain deep ultraviolet (DUV) models as well, effectively restricting China’s access not only to 5-nm-and-below production but also to higher-node tools needed for large-scale AI chip manufacturing. This case illustrates an important dynamic in weaponized interdependence: the United States does not need to control every node directly but can instead mobilize allied states whose firms sit at critical nodes. For the Netherlands, the policy reflects a complex balancing act between economic and security interests. ASML is Europe’s most valuable tech company, and China constitutes a major share of its customer base. Yet the Dutch government ultimately aligned with U.S. assessments that advanced lithography tools have dual-use potential, enabling not only commercial innovation but also AI-enabled military systems, surveillance architectures, and strategic computing capabilities. The decision reveals how middle powers internalize the security concerns of their larger allies, even when doing so imposes substantial economic costs.

Strategically, the ASML restrictions aim to slow China’s ability to produce cutting-edge chips domestically, thereby limiting its progress in advanced AI systems and reducing the military value of indigenous computing platforms. The controls do not prevent China from producing older-generation chips, but they deny the country the tools required to compete at the technology frontier. In this sense, the ASML case differs from the Huawei and Nvidia examples: rather than targeting a specific firm or the export of a discrete product, it seeks to constrain China’s entire industrial capacity at the most foundational stage of chip fabrication. At the same time, the case highlights the limits of vulnerable power. China has responded by accelerating investment in domestic lithography technologies and reorienting procurement strategies toward older tools available from global suppliers. Although replicating EUV is a multi-decade challenge, China’s progress in DUV and in developing indigenous resists, optics, and control systems suggests that sustained pressure may gradually erode the leverage the Netherlands (and, by extension, the United States) currently enjoys.

Overall, the ASML restrictions demonstrate how global semiconductor governance is evolving from market-driven interdependence to strategically managed dependence. The Netherlands plays a critical role as gatekeeper of the world’s most advanced manufacturing technology, illustrating how even small states can become central to the geopolitical contest over AI capabilities.

4. Discussion & Policy Implications

The three cases illustrate a broader transformation in how states approach technological interdependence: what was once viewed primarily as a pathway to efficient global specialization is now understood as a terrain of strategic competition. This shift has several implications for both international security and the governance of AI supply chains.

First, the cases demonstrate that choke point control has become a central instrument of great-power statecraft. Whether through U.S. control of advanced GPUs, its ability to influence foreign semiconductor production, or Dutch dominance in lithography tools, states increasingly view upstream technologies as points where they can exert asymmetric influence. This strategy allows governments to slow rivals’ access to foundational AI capabilities without directly engaging in military confrontation. Yet it also creates long-term risks: if firms and states anticipate politically motivated export controls, they have strong incentives to diversify supply chains, invest in indigenous alternatives, and reduce exposure to foreign technology. Over time, this dynamic may weaken the very leverage that export controls aim to preserve.

Second, the cases highlight the growing tensions between national security objectives and private-sector incentives. Firms like Nvidia and ASML operate in highly competitive and capital-intensive industries where global sales sustain R&D and economies of scale. When governments impose restrictions, they compel firms to absorb revenue losses and adjust business strategies in ways that may weaken their market position. While states justify these costs as necessary for national security, the cumulative effect may be a fragmented global market where commercial viability and technological progress become increasingly subordinate to geopolitical priorities. Policymakers must therefore grapple with the economic consequences of securitizing AI, not only for targeted adversaries but for domestic innovation ecosystems as well.

Third, these cases reveal the limits of weaponized interdependence as a long-term strategy. Restrictive measures may yield short-term advantages, but they also accelerate technological decoupling, encourage substitution, and push targeted states to develop independent supply chains. China’s efforts to build domestic semiconductor capacity, spurred directly by U.S. and allied controls, illustrate how coercive leverage can diminish over time. This dynamic raises questions about the sustainability of these strategies and suggests that states must balance short-term denial with efforts to maintain long-term technological leadership through investment, alliances, and innovation policy.

Finally, the analysis underscores the absence of stable international governance mechanisms for dual-use AI technologies. Unlike nuclear or missile systems, AI supply chains remain governed by ad hoc, unilateral, or minilateral arrangements rather than robust multilateral regimes. As states increasingly treat commercial technologies as strategic assets, the risk of escalation, retaliatory controls, and supply chain shocks grows. This environment suggests a need for renewed international coordination, whether through updated export control regimes, transparency measures, or cooperative frameworks focused on safety and standards, to prevent competitive restrictions from spiraling into full-scale technological fragmentation.

Together, these policy implications point to a central conclusion: managing the security risks of AI requires navigating a delicate balance between leveraging interdependence for strategic advantage and sustaining the openness that fuels innovation. States will continue to use these industries as tools of influence, but the long-term stability and performance of the global AI ecosystem will depend on whether they can reconcile national security priorities with the deeply interconnected nature of technological development.

5. Conclusion

The rapid rise of artificial intelligence highlights a central tension in today’s globalized world: the same connections that drive technological innovation also create vulnerabilities that states feel compelled to manage, restrict, or even exploit. This paper has shown that AI’s dual-use nature blurs the line between commercial and security interests, forcing governments to treat supply chains, firms, and technical standards as tools of statecraft. Looking at Nvidia’s GPU export restrictions, Huawei’s semiconductor constraints, and Dutch controls on ASML lithography equipment, it becomes clear that controlling critical points is now a defining feature of strategic competition.

Across these cases, the logic of weaponized interdependence is evident. States leverage their position within global networks not only to advance their own capabilities but also to limit the technological progress of rivals. AI magnifies the stakes of this dynamic: advanced chips, fabrication tools, and compute infrastructure are simultaneously engines of economic growth and enablers of military power. As a result, governments increasingly accept the economic costs of intervention in pursuit of long-term strategic advantage. This reflects a profound shift from an era of market-driven integration to one defined by security-driven management of technological flows.

Ultimately, the emerging politics of AI supply chains point toward a future in which technological leadership and national security are increasingly inseparable. The challenge for policymakers is not merely to restrict adversaries but to govern interdependence strategically, to preserve the benefits of openness while mitigating its vulnerabilities. As AI continues to reshape international security, the balance states strike between innovation, control, and cooperation will play a decisive role in determining whether the global technology order becomes more fragmented and adversarial, or more stable and collectively governed.