Artificial intelligence is not an autonomous technology, but the advanced expression of a highly concentrated material base. At the root of its expansion lies an apparently neutral yet strategically decisive element: the semiconductor. In the contemporary economy, chips are not merely an industrial input, but a geopolitical threshold that delineates who can develop, scale, and govern artificial intelligence systems and who remains structurally dependent on them. Global technological competition is increasingly organized around this critical material, transforming the semiconductor chain into an infrastructure of power.
Productive concentration constitutes the first factor of systemic instability. The design of advanced chips is dominated by a small number of firms, while high end manufacturing capacity is essentially concentrated in very few facilities. According to data from the Semiconductor Industry Association, more than 90 percent of semiconductors below 5 nanometers are produced by a single industrial actor located in a geopolitically sensitive area. This concentration is not contingent, but structural: it requires investments exceeding 20 billion dollars per facility, hyper specialized skills, and a learning curve that makes the entry of new operators extremely costly. In economic terms, this is an oligopoly; in institutional terms, a redistribution of technological power on a global scale.
From this structure derives an intrinsic vulnerability of supply chains. The semiconductor value chain is fragmented into highly interdependent stages: design in the United States and Europe, manufacturing in East Asia, assembly and testing in lower labor cost countries, and just in time global distribution. The chip crisis between 2020 and 2022 clearly showed how a localized disruption can generate systemic effects, slowing entire industrial sectors from automotive to consumer electronics. Artificial intelligence amplifies this fragility, as it requires increasingly specific, energy intensive, and difficult to substitute chips. Extreme supply chain efficiency thus translates into a reduction of overall resilience.
In this context, the semiconductor becomes a geopolitical lever. Control over access to advanced chips makes it possible to orient the technological development of entire economies. Export restrictions introduced in recent years on specific categories of semiconductors and production equipment have not aimed at blocking individual applications, but at structurally slowing advanced computing capabilities. Power is thus exercised not through visible decisions, but through the delimitation of the horizon of possibilities. Limiting access to hardware means indirectly influencing the trajectory of innovation, industrial competitiveness, and ultimately the decision making capacity of states and enterprises.
The dimension of national security emerges as a direct consequence of this configuration. When civilian infrastructures, industrial systems, and strategic applications depend on a continuous supply of advanced semiconductors, the distinction between economy and security dissolves. Productive continuity becomes a strategic interest. For global enterprises, security no longer concerns only data or network protection, but stable access to physical components without which the entire operating system comes to a halt. Risk management acquires a structural dimension that exceeds traditional planning and requires an integrated reading of industrial, political, and geographical factors.
The thought of Carl Schmitt allows this dynamic to be interpreted in terms of sovereign decision. Schmitt identified decision as the core of political power, capable of defining the boundary between the possible and the impossible. In the semiconductor economy, decision does not necessarily manifest as an explicit legal act, but as control over the material conditions that make some choices feasible and others unattainable. Governing access to silicon means governing, to a large extent, the perimeter of technological action of others. Sovereignty thus shifts from the normative plane to the infrastructural one.
For enterprises, this reality imposes a profound revision of decision making models. Dependence on concentrated supply chains transforms supplier choice into a long term strategic decision. It is no longer only a matter of optimizing costs or performance, but of assessing structural exposure to unstable geopolitical contexts. Individual cognitive limits make these dependencies difficult to perceive until they manifest as crisis. Conscious leadership is therefore called to develop capacities for systemic anticipation, recognizing that operational continuity is no longer guaranteed by efficiency, but by the deliberate construction of margins of maneuver.
Semiconductor chains also produce deep territorial asymmetries. Territories hosting advanced production acquire disproportionate strategic weight, attracting capital, skills, and political attention. Others remain confined to roles of assembly or consumption, with limited capacity to influence technological trajectories. This polarization is not neutral: it determines development opportunities, human capital attraction, and positioning within global value chains. In organizational terms, it corresponds to the separation between decision centers and operational peripheries, with long term effects on the cohesion of the economic ecosystem.
The development of artificial intelligence further intensifies this dynamic. Each performance leap in models requires a more sophisticated hardware base, increasing competition for access to advanced chips. Silicon thus becomes an invisible boundary separating those who can accelerate from those who must wait. Technological progress does not proceed as a uniform flow, but as a field crossed by material frictions, bottlenecks, and political decisions.
Semiconductor chains are not destined to simplify in the short term. On the contrary, they will tend to become increasingly complex and strategic. In this scenario, the challenge does not consist in pursuing an illusory technological autarky, but in recognizing and governing structural dependencies. Enterprises and institutions that integrate this awareness into their strategies do not eliminate risk, but render it legible and manageable.
Contemporary technological power is not exercised solely in code or financial markets, but in the material that makes them possible. Semiconductors represent the point of contact between digital abstraction and the hardness of the physical world. Understanding this point means accepting that the economy of artificial intelligence is not an immaterial domain, but a contested territory, in which decision, fragility, and power are increasingly intertwined.
Global AI Observatory