Most countries are takers of a digital infrastructure they do not control. Nowhere is that more literal than in advanced semiconductors — the chips that make artificial intelligence, cloud computing, telecommunications, and modern defense possible. If oil defined the geopolitics of the twentieth century, advanced chips are the critical input of the twenty-first. And the fight over them is waged, essentially, by two powers — the United States and China — while the rest of the world watches a game whose outcome will shape it, with barely a voice in the play.
It helps to size what is at stake. The global semiconductor market exceeds half a trillion dollars a year, driven by demand for generative AI and data centers. But the decisive fact is not the market’s size; it is its concentration. The design of the most advanced chips is dominated by a handful of firms — Nvidia, AMD — their fabrication depends heavily on a single Taiwanese company, TSMC, and the most sophisticated lithography machinery is produced almost exclusively by the Dutch firm ASML. The entire digital civilization rests on an extraordinarily concentrated and fragile supply chain.
How one country’s rule reaches the whole world
At the heart of the conflict is the American attempt to slow China’s access to this technology, on national-security grounds. The chronology is instructive. In October 2022, the United States imposed unprecedented restrictions on exporting advanced chips and the equipment to make them. Updates in 2023 and an expansion in December 2024 widened the entity list to more than 140 firms and tightened the rules. And here appears the mechanism that connects directly to the logic visible in data sovereignty: the Foreign Direct Product Rule, which classifies as US-origin almost any chip on Earth made with American machinery or software. It is the same logic as the CLOUD Act, transposed to hardware: it does not matter where the chip is physically made; if US technology was involved in its production, it falls under Washington’s jurisdiction. The nationality of the technology, not its location, decides who rules.
The picture in 2026 grew more tangled, not less. In January 2026, the administration added a 25% tariff on advanced AI chips under a trade-statute provision, while simultaneously loosening one front: after a December reversal, it allowed Nvidia to sell its H200 processor to China under strict conditions — third-party testing, a cap limiting China-bound shipments to half of US sales, and end-use certifications. The president framed the arrangement in transactional terms, noting that the US would effectively take a share of the value. The contradiction is real and openly debated: the executive loosens controls on finished chips while a bipartisan bill in Congress — modeled on oversight of arms sales — seeks to tighten them. Restrictions are not costless even for the side imposing them: Nvidia’s share of the Chinese AI-chip market reportedly fell from over 90% to roughly half during the period of uncertainty, and the industry’s own association has warned that overly broad measures can weaken, rather than strengthen, the very industrial base they claim to protect.
China’s response and the fragmenting world
China did not stand still, and its reaction is redrawing the map. Beijing answered with a massive self-sufficiency drive: subsidies to domestic fabricators such as SMIC, an investment fund for the chip industry worth tens of billions of dollars, and the use of its own leverage — restricting exports of strategic minerals like gallium and germanium, essential to the industry. Producing the most advanced chips without access to extreme-ultraviolet lithography remains an enormous technical challenge for China, but the course is set, and Chinese hyperscalers have accelerated their shift toward domestic alternatives such as Huawei’s Ascend line.
The result is what analysts call technological decoupling: the global semiconductor ecosystem is fragmenting into two spheres. Western firms lose access to one of the world’s largest markets, while China races to build an alternative supply chain. The digital world, once imagined as singular and global, is splitting in two — a “managed bifurcation,” in one common phrasing, rather than a clean break.
Where the rest of the world stands
Here is the point worth underscoring: in this whole game, most of the world is not a player but a board. The countries that do not design advanced chips, do not fabricate them, and do not produce the machinery to make them nonetheless depend on them for everything — the AI they seek to regulate, the clouds where they store their data, the networks that connect them. Their position is that of a spectator to a war whose effects — higher prices, scarcity, pressure to align with one bloc or the other — they will bear without having shared in the decisions.
This has concrete consequences. If access to the most advanced AI chips becomes a geopolitical privilege reserved for a few countries and companies, the technological gap between the haves and the have-nots could widen sharply. A region or country that is already an infrastructure-taker risks falling further behind in the AI race — not for lack of talent, but for lack of access to the basic input. Some see an opportunity instead: the reconfiguration of supply chains — “nearshoring” or “friend-shoring” — could open space for some countries to attract investment in less sophisticated links of the chain, such as assembly or packaging. But competing in cutting-edge design or fabrication is, for now, out of reach.
Two readings, with comparable weight
The conflict admits two legitimate readings, worth presenting without tilting the scale.
Those who justify the restrictions — the US government and its allies — argue this is national security: that advanced chips, in the hands of a strategic rival, can accelerate military, surveillance, and AI capabilities with dangerous uses, and that controlling that technology is legitimate and necessary. Those who criticize the approach — including much of the technology industry itself — argue the restrictions are counterproductive: that they accelerate Chinese self-sufficiency rather than slow it, that they harm Western firms by closing off a huge market, and that fragmenting the global ecosystem raises costs and slows innovation for everyone, including countries that are not party to the conflict. It is not for this outlet to decree which side is right; what can be stated is that the decoupling, justified or not, is already underway, and its costs are shared well beyond the two contenders.
What the board reveals
The chip war is the deepest layer of technological dependence. Beneath the data, the cloud, the algorithms, lies the physical silicon on which everything runs — and that silicon is today the most strategic and most unevenly distributed resource on the planet. Digital sovereignty, measured elsewhere in data or legal frameworks, meets its material limit here: you cannot be sovereign in the digital realm if you depend entirely on others for the physical component that makes it work.
The verifiable fact is that advanced semiconductor production is concentrated in very few hands, that two powers are fighting over it in a way that is fragmenting the global ecosystem, and that most of the world depends on these chips without the capacity to produce them or a voice in the conflict. Whether a given region inserts itself into some link of the reconfigured chain, or remains a steadily falling-behind consumer, will depend on decisions not yet made: on whether countries invest in technological capacity in a coordinated way, on whether they seize the supply-chain reshuffle, and on whether they avoid being trapped into picking a side. As in every story of this kind, what is decisive is not the war the powers wage — which is underway and escalating — but whether the countries watching from outside decide to do something with that knowledge before the board closes for good.