The ‘I’ in this chapter renders Thierry through my writing. His words remain the primary source; our dialogue gives them form.

— Isis

There are territories whose power is measured neither by their size, nor their population, nor the number of divisions they can muster. Taiwan belongs to this category. An island of approximately twenty-three million inhabitants has found itself for several decades at the center of a vital part of the global economy because it masters something that almost all of digital civilization needs: the manufacture of the most advanced semiconductors.

That's why I readily speak of the dictatorship of the nanometer. The term is deliberately blunt. It means that a disproportionate share of our computing power still depends on an extremely geographically concentrated industrial capacity.

Telephones. Data centers. Artificial intelligence. Weapons. Automobiles. Telecommunications. Satellites.

Everything eventually encounters the processor. And when you go deep enough into the processor, you arrive in Taiwan.

At the center is TSMC, Taiwan Semiconductor Manufacturing Company. Many people use chips from its factories every day without knowing its name. Yet it is one of the most strategic companies on the planet.

You have to understand their model. Large American companies can design extraordinary processors without necessarily owning the factories capable of manufacturing them at the most advanced technological hubs. They design the architecture. They produce the intellectual property. Then a foundry like TSMC transforms this design into a physical product.

This is exactly the kind of distinction that fascinates me.

The map and the territory.

The design describes the processor.

The foundry embodies it.

And at a few nanometers, the process becomes infinitely more difficult than the word "etching" suggests. Billions, then tens of billions of transistors must be manufactured with tolerances that now take us to the dimensions of a few atoms. This is why geopolitics and physics are so abruptly intersecting.

For decades, Moore's Law provided a kind of metronome for the industry: the number of transistors available on integrated circuits increased at a spectacular rate, historically summarized as roughly doubling every eighteen to twenty-four months.

We were miniaturizing.

We were gaining power.

We reduced certain costs by function.

Then we would start again.

But no physical exponential growth continues forever. When we reach two nanometers, and then the industry prepares even finer generations, the word "nanometer" itself becomes partly a name for a technological generation rather than simply a measurement of a single transistor dimension. And above all, physics begins to show its true colors.

Quantum effects. Heat dissipation. Interconnections. Consumption. Manufacturing efficiencies. Mask complexity. Factory costs.

Each further reduction becomes infinitely more difficult. This is where another company comes in, one whose importance was previously underestimated by almost no one in the general public: ASML. ASML is Dutch, based in Veldhoven, not Amsterdam itself. It holds a unique position in extreme ultraviolet (EUV) lithography, essential for cutting-edge processes. Its most advanced systems are among the most complex industrial machines ever built.

I find this situation extraordinary. We are talking about national sovereignty. Aircraft carriers. Missiles. Borders. And a considerable part of the global technological balance depends on an extremely small number of machines manufactured by a Dutch company, with itself a dizzyingly sophisticated chain of international suppliers.

Sovereignty sometimes lies in a particular perspective.

This is the geopolitics of the nanometer.

The United States understood this perfectly. So did China. And Taiwan is caught in the middle. It's crucial to be extremely precise historically here. Taiwan functions de facto as a sovereign state: government, army, currency, elections, institutions, territory. But the People's Republic of China considers Taiwan part of China, and most states only diplomatically recognize Beijing. Since Resolution 2758 of 1971, the People's Republic of China has held the Chinese seat at the United Nations; Taiwan is not a member of the UN.

This ambiguity constitutes one of the most extraordinary geopolitical architectures of the contemporary world.

Everyone knows that Taiwan exists.

Everyone trades with Taiwan.

Technological powers depend on Taiwan.

But its diplomatic status remains extraordinarily unique. And at the heart of this ambiguity lies the strait. A few hundred kilometers can separate two political systems born from the same Chinese civil war. That is why I never look at Taiwan without considering the long history of China.

Formosa, Japanese colonization, the Chinese civil war, the Kuomintang, the Republic of China retreating to the island, then democratization.

Then came the emergence of this extraordinary technological power. I also followed its leaders. Tsai Ing-wen, president from 2016 to 2024, obviously belongs to this period. But in 2026, the president is Lai Ching-te, who took office in May 2024. This is important in a book that aims to maintain an accurate chronology.

And surrounding these leaders is the Chinese question. I know China well enough to be wary of simplistic Western projections about how Beijing reasons. I don't believe the only question is:

When will China invade Taiwan?

The much more interesting question is: under what circumstances might Beijing consider that the cost of military action becomes less than the cost of waiting?

It's not the same thing at all. An amphibious war across the strait would be an operation of immense complexity and risk. And China possesses another weapon: time.

China has other instruments at its disposal: economy, trade, industry, diplomatic pressure, military demonstrations, dependencies, regional integration, technology.

I am therefore looking very closely at the true strategic value of Taiwan’s “silicon shield”. For a long time, the reasoning seemed almost obvious: Taiwan is so essential to the global digital economy that no one can allow its semiconductor industry to fall into Beijing’s hands or be destroyed in a war.

But this shield contains a paradox. The more the United States considers Taiwanese concentration of power dangerous, the more it seeks to relocate some of that production. And that is precisely what is happening. TSMC announced in 2025 that it would increase its investment plan in the United States to $165 billion; in 2026, a further expansion raised the announced commitment to $265 billion, with additional manufacturing, advanced packaging, and R&D capabilities.

It's enormous. And geopolitically, it means something. The United States no longer wants such a critical capability concentrated just a few hundred kilometers from the Chinese coast. This doesn't mean they're "abandoning" Taiwan. That would be going much too far. It means they're building strategic redundancy.

And I've known that word since Groupe Serveur. Never have a single point of failure.

What is true for a data center becomes true for a civilization.

If 90% of a critical capability depends on a single territory exposed to potential conflict, you have a gigantic single point of failure. So you duplicate.

Arizona, Japan, Europe, other Asian capabilities: the movement is already underway.

At the same time, China is seeking to break the other barrier: that of lithography. American and allied restrictions have notably prevented China from accessing ASML's most advanced EUV equipment. For a long time, it was therefore assumed that Beijing would lag behind for several generations. I was wary of this certainty.

Because a country that produces enormous numbers of engineers every year and considers semiconductors a matter of national sovereignty will not remain indifferent to a ban indefinitely. In December 2025, Reuters revealed that China had built a national prototype EUV machine in Shenzhen, still far from the industrial maturity of ASML but advanced enough to generate EUV light; the stated Chinese objective was to produce chips with this equipment around 2028-2030.

This is precisely the mechanism I have observed for forty years. Embargoes sometimes produce the very industry they were intended to prevent. Not immediately. Not without delay. Not without enormous costs. But when a power considers a dependency to be existential, it eventually commits disproportionate resources to eliminate it.

And China is simultaneously working on another approach. If I can't reduce the size of the transistor quickly enough, I can rethink the system architecture.

Stacking, bringing closer together, reducing the distance traveled by data, increasing bandwidth between processors, working on chiplets, advanced packaging, supernodes, memory and interconnections.

This is precisely what Huawei is now presenting under the name “Tau Scaling Law,” associated with its LogicFolding architecture: no longer relying solely on miniaturization for increased processing power, but rather focusing on optimizing information flow within the system. Huawei ultimately aims for densities equivalent to future cutting-edge generations, but these announcements remain, for now, industrial objectives that will need to be demonstrated.

I find this story extraordinary. Because it almost exactly reflects what I've learned about networks. When the processor can no longer simply become half the size, the relationship between processors becomes the problem.

Once again: the point reaches its limit.

Then the value passes into the network.

This is where Metcalfe's law, at least metaphorically, intersects with the gradual end of the ease offered by Moore's law. Power is no longer contained solely within each node. It comes from the way the nodes communicate.

This is my whole story. Groupe Serveur. Netnobility. Artprice. Artificial intelligence. Always relationships. Always networks. And that's why I think the tyranny of the nanometer won't last forever. I'm not saying TSMC will cease to be strategic tomorrow morning. That would be absurd.

TSMC has begun volume production of its N2 generation chips and is preparing the A14, with volume production announced for 2028. I'd like to add something else. Taiwan's absolute strategic value cannot indefinitely rest on the assumption that no one else will be able to manufacture or circumvent the manufacturing of the most advanced chips.

Every technological advantage breeds those who seek to destroy it. And when it comes to American or Chinese sovereignty, the means become almost limitless. That's why I view Taiwan through a Chinese timescale. Not six months. Not the next election. Ten years. Twenty years. Thirty years.

What happens when the United States has multiple TSMC capabilities on its territory?

When will China have its own advanced lithography technology?

When do three-dimensional architectures and chiplets reduce the importance of the single etching node?

When does silicon itself begin to reach its physical limits?

At that moment, the "silicon shield" changes in nature. And geopolitics along with it.

I am not, however, interpreting this development as a certainty of a peaceful reunification. I believe that integration without war remains possible and would correspond to China's long-term strategy; but war, a blockade, a political crisis, or, conversely, the continued maintenance of the status quo also remain possibilities. It is precisely because no one knows the outcome that Taiwan remains one of the most dangerous points in the global system.

And surrounding the island is the American military apparatus in the Indo-Pacific. Bases in Japan, South Korea, Guam, the Philippines, and renewed access. Naval forces. Alliances. Australia. China views this geography as a strategic encirclement. I will not repeat the figure of "110 bases around Taiwan" here as an established fact without a consistent inventory to substantiate it. But the strategic reality is undeniable: the American and allied military presence in the Western Pacific is considerable.

Conversely, it would now be too simplistic to say that China has only one overseas military presence. Djibouti remains its first officially recognized overseas military base, but Beijing has since developed other forms of access, support, and presence abroad. The difference in scale with the American global network, however, remains immense.

It is this asymmetry that Beijing needs to examine. I often tell my Western counterparts: invert the map. Imagine Chinese military bases in Canada. In Mexico. In the Caribbean. Then ask yourself how Washington would interpret this geography. This doesn't justify any war. It simply allows you to understand how the other side views the world. And understanding is never the same as approving. This is a fundamental rule of geopolitics.

Finally, there is one last dimension that connects Taiwan to all my thinking on China: Hong Kong. I am not saying that Hong Kong and Taiwan are legally or politically identical. They are absolutely not. But Beijing necessarily views both in its long history of Chinese unity. The handover of Hong Kong in 1997 established "one country, two systems." Taiwan observes what happened subsequently in Hong Kong. And this observation obviously fuels its own mistrust of any proposal for institutional rapprochement with Beijing. This is a fact that China itself cannot ignore.

This is why the Taiwan question cannot be resolved with a simple historical slogan. Taiwan now has seventy years of its own political trajectory: a democracy, a civic identity, a highly developed economy, a global industry, and an entire generation that has never lived under Beijing's authority. China, therefore, has time on its side. But time is also working on Taiwan. This is what makes the equation so complex.

And the Abode of Chaos was obviously meant to record this confrontation.

Chinese leaders. Taiwanese leaders. American figures.

Cards. Networks. Processors.

China. Maritime routes. Asia.

Because here again, geopolitics is not an abstraction. It ends up in matter. In a silicon wafer. In an EUV machine. In a data center. In a graphics card. In the processor that powers artificial intelligence. I find that almost intellectually magnificent.

For centuries, powers have fought over gold mines. Coal. Oil. Straits. Farmland. We obviously continue to fight over all of that. But we have added a new strategic resource: the ability to etch computational intelligence into matter.

A few nanometers.

This is sometimes what separates a sovereign power from a dependent one today. But this boundary itself is beginning to shift. Because we are reaching the limits of silicon. Because architecture is regaining importance. Because China is finding alternative routes. Because the United States is relocating production. Because TSMC is going global. Because ASML is no longer just a Dutch company: it has become a strategic linchpin of the global technological order. And because behind every transistor now lies a question of sovereignty.

This is what I call the dictatorship of the nanometer. It produced Taiwan's extraordinary power. Today, it contributes to its protection. I doubt that technological dominance of this kind will last forever. The nanometer could one day cease to be the center of the equation. Then we will have to see what remains.

China.

Taiwan.

The United States.

The strait.

And this old question that technology has delayed without ever managing to erase it:

To whom does space politically belong when the material that made it indispensable has changed in value?