The transformation unfolding around artificial intelligence concerns not only technological innovation, but the reconfiguration of power relations that structure the global economy. When a technology becomes systemic, it ceases to be a set of tools and takes the form of an infrastructure. In this phase, AI operates as a material architecture of power, capable of redefining dependencies, hierarchies, and possibilities for action for states, enterprises, and institutions. Its impact exceeds the perimeter of efficiency or sectoral competitiveness and is situated on the geopolitical plane, where physical resources, supply chains, and strategic decisions determine access to the future.
Considering artificial intelligence as infrastructure implies recognizing its material density. AI rests on energy-intensive data centers, high-capacity transmission networks, advanced semiconductors, cooling systems, and stable electricity supplies. According to estimates by the International Energy Agency, the energy consumption of data centers could double by 2026, driven largely by the adoption of large-scale artificial intelligence models. This growth is not geographically neutral: the localization of infrastructure concentrates value and power in specific territories, while others remain excluded from direct benefits while bearing environmental and systemic effects.
The materiality of AI introduces a growing dependence on scarce and concentrated strategic resources. The production of advanced chips is dominated by a small number of global actors, with strong geographic concentration. According to data from the Semiconductor Industry Association, more than 70 percent of the production capacity of the most advanced semiconductors is located in East Asia, making the entire digital economy vulnerable to geopolitical and logistical shocks. Added to this is dependence on critical minerals, such as gallium and germanium, whose control has become the object of explicit trade restrictions and industrial policies. Artificial intelligence thus brings the digital economy back into a classical logic of power, in which control over resources determines hierarchies.
In this context, control of infrastructure precedes and conditions every application possibility. Those who possess computational capacity, energy access, and stable supply chains do not merely control a market, but the very definition of what is technically and economically feasible. Sovereignty is no longer exercised only through explicit political decisions, but through the configuration of conditions of possibility. In the corporate world as well, real power rarely coincides with the final decision-making act; it resides rather in the definition of standards, architectures, and available options.
The global rivalries emerging around AI therefore take the form of infrastructural competitions. Public investments in computational capacity, industrial reshoring policies, and technological alliances become instruments of geopolitical positioning. The United States, the European Union, and China are adopting explicit industrial strategies to reduce critical dependencies and secure control over AI supply chains. These dynamics do not necessarily produce open conflict, but redraw the global balance through apparently technical choices with systemic consequences. For enterprises, this means operating in an environment where technological decisions are inseparable from geopolitical considerations, even when they are not perceived as such.
The thought of Bruno Latour offers a useful interpretative key to understand this transformation. Latour showed how technology is not a domain separate from politics, but a set of assemblages that organize relations and produce power effects. Artificial intelligence, in this perspective, is not a neutral actor, but a device that structures what becomes visible, calculable, and governable. Treating it as material geopolitical infrastructure means recognizing that every technical choice incorporates a worldview and an asymmetric distribution of possibilities for action.
In the corporate context, this awareness redefines the role of leadership. Deciding where to locate data centers, which semiconductor suppliers to privilege, which standards to adopt, and which models to integrate is no longer a matter of cost optimization, but of strategic positioning within an unstable ecosystem. Global enterprises are required to manage risks that exceed the traditional business horizon, including geopolitical vulnerabilities, regulatory tensions, and environmental impacts. Individual cognitive limits make this complexity difficult to govern, but delegating it entirely to automated models would mean renouncing a critical understanding of long-term implications.
The infrastructural dimension of AI also produces new social and territorial asymmetries. Territories that host infrastructures bear high environmental costs, while economic benefits tend to concentrate elsewhere, in decision-making and financial nodes. Other territories, despite intensive use of AI-based applications, remain excluded from processes that define standards and technological trajectories. This separation between those who benefit from innovation and those who materially sustain its functioning mirrors known dynamics of global value chains, but with intensity amplified by the pervasiveness of artificial intelligence.
Interpreting AI as material geopolitical architecture therefore imposes a change of paradigm. It is not sufficient to question model capabilities or application uses; it is necessary to consider where infrastructures are located, who controls access to them, and under what conditions they are integrated into economic and institutional systems. The global rivalries taking shape are not the result of an uncontrolled technological race, but of deliberate choices that intertwine economic interests, visions of power, and long-term strategies.
In this framework, the stake is not technological supremacy in an abstract sense, but the capacity to govern infrastructures that shape the collective future. Enterprises and institutions that recognize the geopolitical materiality of artificial intelligence stop treating it as a simple factor of competitiveness and begin to consider it as a strategic environment to be inhabited with awareness. AI does not float above the digital world: it crosses it, excavates it, and redesigns it. Understanding it as infrastructure means accepting that the power of the future will not be played out only in code, but in the invisible foundations that make it possible.
Global AI Observatory