After Taiwan, the scale had to be broadened. The island concentrated an extraordinary portion of global technological power, but it was only one node in a much larger system. To understand 21st-century China, it was no longer enough to look at its GDP, its military, its exports, its cities, or its technology companies. It was necessary to look at the infrastructure it was building around itself and sometimes very far away.

Ports. Railway lines. Industrial zones. Roads. Energy networks. Cables. Telecommunications. Logistics platforms. Financial agreements: each of these elements seemed to belong to a different domain. Yet, viewed together, they formed a single architecture. China was not merely seeking to participate in the global system. It was progressively seeking to be able to act within that system without being entirely dependent on the infrastructure, standards, and institutions built by others.

This is a fundamental distinction. A power can be rich and remain dependent. It can export enormously but depend on maritime routes it does not control, on currencies it does not master, on semiconductors it cannot produce at the most advanced level, on foreign software, on standards decided elsewhere.

Wealth, therefore, is not yet sovereignty.

Sovereignty begins when one has sufficient control over the conditions of one's own continuity.

This is precisely the logic of resilience I had observed in Ukraine and Taiwan. China applied it on a continental and perhaps global scale. The New Silk Roads must be viewed from this perspective. Their very name is remarkable. It evokes long history, caravans, exchanges between East and West, a China situated at the heart of ancient networks. Contemporary infrastructure thus acquires a genealogy.

A new railway line is presented as a continuation of a centuries-old route. It is a political gesture as much as an economic one. Major powers like to situate their projects within a timeframe that extends beyond that of a single government. This creates an impression of historical necessity. The present becomes an episode in a continuum. But behind the narrative lie very concrete objects: modernized ports, highways, railways, power plants, logistics zones.

These infrastructures clearly address real economic needs. It would be a caricature to reduce them to a single strategic conspiracy. Countries need ports, electricity, and roads. Chinese companies need markets. Banks finance projects. Local governments negotiate according to their own interests.

However, when enough of these relationships accumulate, they create political depth. Infrastructure can produce a lasting relationship. A road can be built in a few years and used for fifty. A port alters a country's trade flows. A power plant creates technical, financial, and industrial dependence.

That is why infrastructure is one of the most patient forms of power.

It doesn't always require a flag to be planted. It organizes the possibilities.

Power is not only the ability to prohibit. It is also the ability to define the paths that others take. This idea seems essential to me. For centuries, empires controlled trade routes, straits, and ports. In the 21st century, this remains true, but the nature of these routes has changed. They are both physical and digital. A submarine cable is a route. A payment system is a route. A cloud is a route. A technical standard is a route.

The one who defines the protocol partly determines how others will have to operate. It's an extraordinarily discreet form of power because it appears technical. Yet, technology is often political in retrospect. Take a telecommunications standard. At first glance, it's about efficiency, frequency, and compatibility. But the company whose patents, equipment, and expertise it holds, the one that participates in defining international standards, acquires lasting influence.

Once millions of devices are deployed, changing standards becomes expensive. The infrastructure thus creates a lock-in effect. This is exactly what I understood with databases and digital standards: when enough players adopt an architecture, it ceases to be simply a technical choice. It becomes an environment. We no longer decide each morning to use the protocol. We live in it.

Perhaps this is one of the most powerful forms of sovereignty: becoming the technical environment of others.

This logic is obviously not inherently Chinese. The United States has wielded it with exceptional power for decades. The internet, software, operating systems, global finance, technological standards, platforms: a huge part of the contemporary digital and financial architecture is linked to American institutions or companies.

American power, therefore, does not reside solely in its aircraft carriers. It also resides in the invisible architecture of the world. China has perfectly understood this. This is precisely why the Sino-American competition cannot be reduced to a military competition. It is a competition of architectures.

Who will provide the networks? Who will provide the processors? What currency will be used? What satellites? What platforms? What cybersecurity standards? What artificial intelligence models? A power that loses all these layers could still maintain a large military and yet remain dependent on an environment built by its rival. The fight for standards thus becomes a fight for future autonomy.

This idea directly relates to my history with the internet. During the years we were building the digital architectures for Groupe Serveur, it was clear to me that a protocol is never neutral in its consequences. Choosing an open or closed, centralized or distributed, proprietary or interoperable architecture alters the balance of power. Behind an engineer's decision sometimes lies a policy that no one has explicitly formulated. The network creates both possibilities and prohibitions.

China is thinking precisely on this scale. It doesn't simply want to produce more goods; it wants to occupy more layers. This begins with manufacturing, then moves up to components, research, standards, digital infrastructure, satellites, and artificial intelligence. This move up the value chain is probably one of the most significant economic transformations of our time.

For a long time, parts of the West viewed China as the world's factory: immense industrial capacity, competitive production costs, and integration into Western supply chains. This image has become inadequate. A country that manufactures for others learns. It develops engineers, suppliers, logistics chains, universities, and capital. Gradually, it no longer simply wants to assemble other people's technology. It wants to produce the technology itself.

This is an almost natural trajectory for industrial power. It would be a mistake to believe that a country would indefinitely accept a subordinate position in a value chain if its capabilities allow it to move beyond it. China has therefore invested massively in sectors considered strategic: telecommunications, electric vehicles, batteries, solar energy, space, artificial intelligence, and semiconductors.

These choices are not merely industrial. They determine which players will own the infrastructure of the next economy. The battery, for example, might seem like a simple component of the electric car. It also provides energy storage, grid management, industrial autonomy, and access to minerals. Behind every technology lies a geology. Lithium, cobalt, rare earth elements, copper: the digital and energy world rests on a new map of resources.

We have sometimes spoken of the energy transition as if it freed us from the geopolitics of oil. It simply creates other dependencies.

Every technology produces its own geography of resources.

Coal built one geopolitical order. Oil built another. Critical metals, batteries, and semiconductors are building a new one. The dream of a completely immaterial economy collapses yet again.

The more digital we become, the more we need to extract, refine, and transport extremely precise physical materials. Information technology doesn't eliminate matter altogether. It demands a more organized form of matter. It's almost a law of technological evolution: value is concentrated less in mass than in the degree of organization. A kilogram of rock is worth little; certain grams, transformed, purified, and structured, become strategic.

This transformation of matter through knowledge is deeply connected to alchemy as I have always understood it. The alchemist does not create ex nihilo. He transforms, purifies, separates, and recombines. Contemporary industry accomplishes this with gigantic precision. A battery factory, a semiconductor plant, a data center are modern-day crucibles. They transform matter and energy into functional power.

China has understood that a state that controls these industrial hubs possesses far greater autonomy than one that merely buys their products. This also explains its presence in regions that may seem peripheral when viewed solely on large military maps: Africa, Latin America, and Central Asia. These areas contain resources, markets, ports, voting rights in international institutions, and logistical corridors.

A comprehensive strategy cannot be built solely on military alliances. It is also constructed through the accumulation of economic relationships. Here again, we must avoid oversimplification. Every partnership is negotiated. Some countries acquire infrastructure they could not otherwise afford. Some projects succeed, while others encounter problems with debt, profitability, local opposition, or environmental issues. The reality is diverse. But together, they produce something: a presence.

Presence is the first step towards influence. One cannot exert lasting influence in a region without businesses, diplomats, investments, or local knowledge. China has patiently built this presence. It has often offered a relationship presented as less prescriptive than that of Western powers: trade and infrastructure without such explicit political conditioning on internal institutions.

This approach has naturally attracted some governments. For others, it has also raised questions about Chinese dependence or influence. This is yet another ambivalent relationship. No major power project is perceived in the same way by those it affects. Infrastructure that represents an opportunity for one party can become a source of dependence for another.

A port built as a commercial tool can acquire new strategic value if the political context changes. This is why these assets must be analyzed across multiple timeframes. Current civilian use does not exhaust all possible future uses. This duality has become characteristic of many contemporary infrastructures. A satellite can be commercial and produce intelligence. A civilian port can accommodate military vessels. A telecommunications cable can carry both commercial and governmental data.

The distinction between civilian and strategic interests is becoming blurred. This ambiguity forces states to view investments that would once have been considered purely economic through a security lens. Globalization is changing its nature when every port terminal acquisition becomes a geopolitical issue. But caution is necessary: if any economic relationship with a rival is immediately perceived as a threat, the system will tip toward fragmentation, with each group seeking to rebuild a completely autonomous bloc.

This fragmentation would have a colossal economic cost and could itself increase tensions. We find ourselves once again facing the perennial dilemma between interdependence and sovereignty. Too much dependence creates vulnerability. Too much separation destroys the mechanisms that give each country a stake in the stability of the other. Part of the peace between great powers rests precisely on the fact that a catastrophe in the adversary's country would have a cost for oneself.

Complete decoupling can therefore reduce vulnerability and simultaneously remove an economic obstacle to conflict. This is yet another example of second-order causality. There is no choice without cost. The best strategy probably involves differentiating dependencies according to their criticality. This is exactly what a security engineer would do. You don't secure every file like a nuclear secret. You identify critical assets.

Similarly, a state must identify which dependencies are acceptable and which constitute an existential risk. This idea of classification seems fundamental to me. Without it, sovereignty becomes a mere slogan. Strategy requires an inventory: what energy? What medicines? What processors? What networks? What minerals? What substitution timeframe? We always come back to the database.

We need to know what we own, what we depend on, where it comes from, and how long it would take to replace it. Contemporary sovereignty almost begins with a well-designed relational foundation. This statement may sound provocative; it is not. A government unable to accurately map its industrial dependencies may discover its vulnerability at the moment of crisis, when correction becomes far more costly.

This has been evident with pandemics, energy, and electronic components. I start from a principle: a policy cannot be sound if the information on which it is based is false or incomplete. This is precisely what I have defended my entire life. Data is not administrative. It is strategic. That is why Artprice has never been, for me, simply a website providing prices. It is an architecture that makes visible a market whose opacity generated power for those who possessed the information.

Making things visible immediately shifts the balance of power. Globally, states are now rediscovering the same thing: whoever has precise knowledge of supply chains, production capacities, and dependencies possesses superior decision-making power. Artificial intelligence will obviously accelerate this mapping. It can analyze millions of trade flows, patents, satellite imagery, and scientific publications. It can help identify a critical supplier that no one had previously recognized as such.

Geopolitics is therefore increasingly becoming a data science. But here again, the risk is immense. Seeing a trade correlation does not automatically allow us to understand the political relationship. A company may import a Chinese component and have several alternatives. Another may import only a small quantity but depend on irreplaceable expertise. Raw data is not enough. It must be contextualized. This is the difference between intelligence and mere inventory.

China itself possesses this obsession with information. A state of this size cannot plan without massive amounts of data. The digitization of Chinese society has produced considerable capabilities in administration, payment, commerce, and surveillance. This is where our Western perspective often focuses on the dimension of political control, which is real and deserves examination, but we must also understand the administrative efficiency that data can generate.

The same technical system can facilitate daily life and increase the state's surveillance capabilities. Again, this ambivalence. Digital payment can be extraordinarily convenient and produce a complete trace. Smart cameras can streamline a city and monitor its population. I never consider technology neutral in its effects, because its architecture determines who can see what.

This is a fundamental political question for all societies, not just China. Democracies themselves accumulate enormous volumes of private data in the hands of corporations. The difference lies in the institutions, the law, the avenues for redress, and the separation of powers. This is why comparing digital models must go beyond slogans. The real issue is: who owns the data, who can access it, under what control, and with what checks and balances?

This is probably one of the great invisible constitutions of the 21st century. A country can have a political constitution and a much more centralized, implicit digital constitution. The user clicks "accept" and sometimes relinquishes an amount of data that an authoritarian state would once have dreamed of obtaining. This shows that surveillance doesn't always come solely from political power. It can come from the economic model.

China, the United States, and Europe are developing different architectures in this regard. Comparing them is fascinating because they reveal different philosophies of the relationship between the individual, the market, and the state. Europe often seeks to build sovereignty through law and regulation. The United States possesses extraordinary power through its technology companies and private innovation. China combines state power, industry, and large platforms under more direct political control.

None of these models is static. They influence and react to one another. This institutional competition is as important as military competition. Which model will produce the most innovation? The most trust? The most resilience? These are open questions. We must guard against any deterministic predictions. The models can evolve.

China itself has changed profoundly several times in just a few decades. It is a consistent mistake to project a current snapshot onto a thirty-year horizon. Power is a process, not a static entity. It is precisely this capacity for Chinese transformation that strikes me. The country has been able to go from a largely rural economy to an industrial and technological powerhouse in an extremely short historical period. The scale of this transformation is staggering.

Entire cities have been built, railway networks deployed, and hundreds of millions of people integrated into urbanization. Whatever political criticisms one might level at the regime, ignoring this capacity for execution would prevent us from understanding the system's true power. A serious strategy begins with recognizing the qualities of the adversary or competitor. Underestimating them out of contempt is one of the most dangerous forms of the Blind Spot.

The 20th century offers numerous examples. We sometimes prefer to imagine that the rival system will collapse because it seems contradictory. But a system can contain contradictions for a very long time if it possesses sufficient adaptability. The reverse is also true: a trajectory of success does not guarantee its perpetual continuation.

Demographic aging, debt, real estate, social tensions, technological competition, energy dependencies: China, too, has its vulnerabilities.

A great power is never invincible. It is simply a system whose strengths and weaknesses are far greater. This is why I am wary of narratives that alternately announce China's inevitable domination or its imminent collapse. Both satisfy narrative desires. Neither constitutes a method. We must look at the indicators, the structures, the capacity for adaptation. And above all, we must be willing to update the model when reality contradicts it. This is probably the most important rule of all intelligence.

In my view, a model that can no longer recognize its errors risks becoming an ideology. This applies to humans as well as AI. This capacity for self-correction should be at the heart of any cognitive system worthy of the name. It directly relates to my Blind Spot method: asking where the assumption I no longer question lies. China is a formidable test case for this discipline because it concentrates Western projections.

Fascination with its efficiency, fear of its authoritarianism, hope for its market, military concerns, technological admiration, criticism of human rights: these perspectives overlap. Each contains a part of reality but can become blinding if it becomes all-encompassing. Analysis must therefore accept contradiction: a country can be a major economic partner and a strategic rival; a source of innovation and a source of risk; an indispensable player in the global climate and a military competitor.

The multipolar world compels us to move beyond binary moral categories without abandoning judgment. This is an extremely important distinction.

Understanding complexity never implies considering all actions as equivalent. We must be able to condemn one practice and cooperate on another issue with the same actor. Diplomacy is precisely the art of managing relationships between actors who do not share the same values or interests.

If we only agreed to speak with those who are like us, diplomacy would be pointless. This is where de-escalation becomes crucial. Between the United States and China, perhaps the greatest danger isn't the existence of competition itself. Competition is now structural. The real danger lies in the inability to define the limits of this competition. All rivalries require implicit or explicit rules.

During the Cold War, Washington and Moscow gradually built mechanisms for communication, arms control, and crisis management precisely because they understood the risk of mutual destruction. The Sino-American rivalry today has less of this accumulated infrastructure. Yet contemporary technologies reduce reaction times and expand the areas of confrontation.

Space, cyber, the sea, AI, semiconductors: the potential for incidents is multiplying. Therefore, channels must be built before a crisis occurs. It's almost an engineering principle: the backup system must exist before the fire. You don't design the emergency exit when the building is burning. Yet states sometimes do exactly that with diplomacy. They discover they no longer have a channel of trust precisely when they need it most.

Trust doesn't require friendship. It simply requires knowing that the other person will respond and that their words possess a certain stability. This minimal level can be enough to prevent misinterpretation. I deeply believe in this engineering of de-escalation. It is neither naive pacifism nor weakness. It is the architecture of security. An arsenal without communication can be more dangerous than a larger arsenal accompanied by control mechanisms.

This is a paradox that our era must rediscover. AI will further complicate this by adding systems whose workings may be partially opaque even to those who deploy them. Imagine two powers using algorithmic tools to detect hostile behavior. The same military movement could be classified differently by each model.

Each state then receives an alert generated by a machine trained on its own data and doctrines. The biases become geopolitical. An American AI and a Chinese AI could observe the same world and produce two different operational realities. This is a huge problem. We talk a lot about algorithmic biases in individual decisions. We must also consider strategic biases.

A military model trained primarily on historical data where certain behaviors preceded an attack may overinterpret similar signals in a different context. This is precisely the danger of automated analogy. The system recognizes a pattern and assumes causality. This is why military AI should probably be designed not only to produce a primary hypothesis but also to generate alternative scenarios and the reasons why it might be wrong.

I would suggest that it become a machine for questioning as much as a machine for making recommendations. This is almost the opposite of how we sell technology today. We value certainty, precision, and immediate answers. Yet in some fields, the best intelligence is that which knows how to say: I don't know enough. This statement is a strength, not a weakness. The political world will have to learn this.

China itself is investing heavily in AI because it understands its power multiplier effect. But here again, the competition isn't just about models. It's about the entire technology stack: energy, data centers, processors, research, data, and talent. The word "stack" is interesting. Computer scientists use the term to describe the different layers of a system.

I believe that contemporary power operates in exactly this way. There is a geopolitical stack. At the bottom: matter, energy, resources. Then: physical infrastructure, factories, networks. Next: semiconductors, computer systems. Next: software, AI models. Then: institutions, companies, administrations that know how to use them. Finally: political doctrines and decisions.

A power that controls only the top without sufficiently controlling the lower layers can be vulnerable. Conversely, possessing mines and factories without intellectual or institutional capacity does not automatically produce dominance. Power is vertical. It stems from the integration of the stack. China is working precisely to reduce the gaps in its own stack.

The United States seeks to preserve the areas where it holds an exceptional advantage. Europe is still sometimes trying to determine where it wants to stand between these two models. This is probably one of the major European political questions: do we want to be solely a regulated market, or do we also want to possess the infrastructure for our technological sovereignty? Law is powerful, but it cannot entirely replace industry.

A European standard can influence the world if the European market is large enough. This is sometimes called regulatory power. But a standard for a technology that we no longer know how to produce is inherently fragile. We can regulate the use of foreign infrastructure, but we remain dependent on the supplier. This is why legal sovereignty and industrial sovereignty must be linked.

Here again, I see my own experience reflected. The sui generis right to databases in Europe was immensely valuable because it allowed for the protection of an informational investment. But the law would have been useless if we hadn't actually built the databases. Legal protection doesn't create the asset. It protects what already exists.

This is precisely the European problem with AI: creating rules is necessary, but so is creating companies, models, data centers, and chips. Otherwise, we risk becoming everyone else's expert innovation lawyers. This idea seems important to me because it moves beyond the false debate between regulation and innovation. We need both.

Standards must protect people and build trust; industry must empower them. A civilization that only regulates can become dependent. A civilization that imposes no limits can become dangerous to itself. The challenge is balance. This is probably at the heart of any mature technology policy.

China, for its part, has sometimes chosen a different priority: massively accelerating industrialization and then regaining control when certain private actors become too powerful or when the state considers that a sector threatens its objectives. This produces a particular relationship between innovation and political authority. It can allow for the rapid mobilization of resources, but it can also create uncertainty for entrepreneurs or investors.

Each model has its strengths and costs. No institutional system exists without trade-offs. This is an English term that should almost be retained because it is so precise: any gain in one direction often produces a cost elsewhere. More centralization can increase execution capacity and reduce the diversity of decisions. More decentralization can foster innovation and slow down coordination.

Intelligence lies in knowing when an advantage becomes a disadvantage. This applies to a state, a company, and AI. For me, China is therefore less an answer than a persistent question: to what extent can a highly centralized system maintain the flexibility necessary for innovation? And conversely, to what extent can liberal democracies maintain their capacity for strategic decision-making in systems where power and interests are dispersed?

This competition has no predetermined winner. It will be decided by the capacity for adaptation. Once again: only the mutants survive. But the Chinese mutation also has an effect on the rest of the world. When the world's first or second largest economy decides to invest massively in a technology, prices, production capacities, and global standards change.

Electric vehicles are a striking example. The rise of Chinese manufacturers is transforming the global automotive industry, forcing established automakers to accelerate their development, and prompting governments to reconsider tariffs, subsidies, and battery dependency. Technological competition is thus immediately becoming industrial policy. The clear separation between market and national security has disappeared.

Each country is beginning to subsidize its critical industries, control certain investments, and impose restrictions. We are almost returning to a form of technological mercantilism, but with much more complex global supply chains. This contradiction is creating a unique era: everyone wants globalization when it reduces costs and sovereignty when it reveals dependencies.

Governments will have to constantly arbitrate. There won't be a single, sweeping decoupling, but likely a differentiated restructuring across sectors. Everyday goods may remain globalized. Critical technologies will be more tightly controlled. This will create an economic map with multiple levels of trust. The global network could fragment into interoperable but politically distinct zones.

The internet itself risks following this trajectory. For a long time, we talked about a single global network. Already, regulations, censorship, standards, and infrastructure are creating different experiences depending on the country. The network could become less universal. This is an issue that particularly affects me because my initial vision of the internet was deeply rooted in its ability to transcend borders.

The fragmentation of the network is therefore a historic shift. An internet divided along political lines would diminish some of its universality, but it may already be taking shape. National firewalls, data localization rules, application restrictions, cloud sovereignty: each measure can be justified individually, but their accumulation creates digital borders.

Here again is this idea: a border can emerge after the fact from hundreds of technical decisions. No one necessarily decides one day to divide the internet into blocks. Yet the system eventually fragments.

This is how many historical transformations occur. They are not always announced. They result from an accumulation.

Much like the Abode of Chaos itself. No single work is sufficient to produce the whole. It is accumulation, repetition, the relationship between gestures that ultimately creates a world. China, too, is building a world through the accumulation of infrastructure. This term "world" is important. Every great power implicitly proposes a way of organizing relations.

After 1945, the United States proposed an architecture based on international institutions, the dollar, alliances, and relatively open markets, even if practice has obviously never perfectly matched the ideal. China has not yet built a completely alternative global system, and perhaps it is not seeking to replace all existing institutions. Nevertheless, it is creating options.

New financial institutions, agreements, infrastructure, trade in alternative currencies, South-South cooperation: simply creating an option is already a form of power. A state becomes more autonomous when it has multiple suppliers, multiple financiers, multiple partners. China thus increases the room for maneuver of certain countries by offering them an additional relationship. This mechanically reduces the Western monopoly on influence.

This is probably one of the most significant transformations of the multipolar world. Countries once forced to choose from a relatively limited menu can now negotiate between several powers. This competition can give them greater autonomy. It can also create new dependencies. But it changes the relationship.

The Global South then becomes less a bloc than a collection of states capable of playing between several centers. It is a common Western mistake to ask: are they with us or with China? Many implicitly answer: with us. They can buy weapons from one country, build a port with another, trade with a third. This is precisely the logic of multipolarity.

The alignment becomes partial. This is yet another sign of the end of simple blocs. The graph reappears. India is a fascinating example: a partner of the United States in certain Indo-Pacific architectures, a historical relationship with Russia, competition with China, and a policy of strategic autonomy. It would be absurd to try to categorize it simply into one camp. The 21st century will be filled with these pivotal powers.

Indonesia, Turkey, Saudi Arabia, Brazil, South Africa: each will likely seek to maximize its room for maneuver. The system thus becomes less hierarchical, but perhaps more unstable. The more independent decision-making centers there are, the greater the number of possible relationships. Mathematically, the network becomes very complex very quickly. This is where AI can become almost indispensable for analysis.

The human mind can grasp a few major relationships. Maintaining hundreds of economic, military, and technological dependencies simultaneously becomes far more difficult. AI can help map this complexity. But it should never turn the graph into destiny. A strong economic relationship does not automatically guarantee an alliance. A vote at the United Nations can shift depending on the issue.

The multipolar world therefore demands contextual intelligence. This is precisely what rigid classifications lack. The Abode of Chaos seems to have anticipated this multiplicity because it never definitively categorizes a character. A portrait can be close to several contradictory signs. A political figure can be an ally here, an adversary there, an economic partner elsewhere.

The wall accommodates this contradiction far better than a two-dimensional painting. This is probably art's strength in the face of overly neat models: it can allow contradictions to coexist without artificially resolving them. Reality needs this space. China itself is full of it: market economy and Party control, globalization and sovereignty, open trade and protection of strategic sectors, nationalism and dependence on global commerce.

None of these pairs cancel each other out. They coexist. The strategic question is how long the system can maintain their tension and how it transforms it. This is precisely what a living power is: an organized tension. Chaos never means an absence of structure. It means a plurality of forces whose stability derives from their relationship.

Here again, I find myself at the House. From the outside, some see a contradiction: the global headquarters of a data company within a post-apocalyptic Total Work of Art. For me, it is precisely this contradiction that creates the coherence. Data center and rust, artificial intelligence and alchemy, the Art Market and vanity.

Contemporary China also exhibits these clashes of eras: temples and megacities, Marxism and billionaires, digital surveillance and global trade, Confucian tradition and quantum engineering. Great civilizations never entirely replace their ancient layers; they reconfigure them. This historical depth also helps explain why China sees itself less as an emerging power than as a return to the past.

The Western narrative often speaks of the “rise of China.” The Chinese narrative can be one of recovery after a period of historical weakness. This shift in perspective is fundamental. A power that sees itself as returning to its rightful place does not view its growth as aggressive ambition but as a correction of history. This can make certain concessions more difficult because they are perceived as a return to a past humiliation.

This is precisely why memory is strategic. A negotiation never simply pits present interests against each other. It sometimes pits centuries of narratives against each other. Those who ignore these narratives can provoke reactions they will not understand. This is the same lesson as with Russia, Iran, or the Middle East. Modernity never erases memory. It gives it new tools.

China today uses the most modern technologies to pursue objectives that can be expressed in a very old historical language: unity, sovereignty, national rejuvenation. This overlap seems to me to be one of the characteristics of the contemporary world. Daesh used Twitter in the name of an imaginary caliphate. China uses AI in a narrative of civilizational rebirth. The United States mobilizes Silicon Valley in a political tradition stemming from the 18th century.

The technological future is always imbued with ancient myths. This is why an engineer alone cannot understand technology. The historian, the sociologist, and the philosopher are also needed. An AI model never emerges in a vacuum. It enters a society, a legal system, a shared imagination. The same technology will therefore produce different institutions depending on the civilization.

Perhaps one of the great errors of digital globalization is believing that technology will automatically homogenize societies. It standardizes certain tools and can simultaneously reinforce political differences in their use. The smartphone is almost the same object everywhere; what a state can do with its data varies radically. The processor is universal; the rules governing its use are not.

This is why the battle of standards is so important. Technical standards can be global, while legal standards diverge. We may enter a world where machines communicate seamlessly with each other, while the political systems surrounding them become increasingly distinct. It's a fascinating paradox. The more technology connects, the more politics may seek to reassert borders.

The global network simultaneously produces integration and sovereignty. This is precisely the dialectic that China embodies. It has benefited from globalization like few states before it, and then uses the accumulated resources to reduce some of its dependence on globalization. This is not an accidental contradiction. It is a strategy of empowerment.

Every power eventually seeks to transform the dependence that allowed it to grow into the autonomy that will allow it to endure. The United States itself has historically followed this trajectory in relation to Europe. Therefore, nothing is exceptional in principle. What is exceptional is the scale and speed of this transformation in China. This transformation directly raises the question of the next world order.

If multiple technological, financial, and military architectures coexist, will they be sufficiently interoperable to prevent total fragmentation? Can there be multiple centers of power within a single global system? History sometimes answers yes for long periods, sometimes no. The problem is not multipolarity itself. It is the absence of recognized rules between the poles.

A distributed network can be extremely resilient if its protocols are common. Without protocols, it becomes a chaotic mess of incompatible systems unable to communicate. Climate change is the most obvious example: China and the United States may be strategic rivals, but no global climate policy is possible without them. Viruses don't consider military alliances. Global warming doesn't obey sanctions. Some threats remain universal.

The 21st century will therefore have to invent a capacity to be simultaneously competitors and cooperators. This idea seems contradictory only if we still imagine the world as divided into two camps. In a graph, two nodes can have several different relationships. This is precisely what we must learn politically: to cooperate on climate, to compete on AI, to negotiate on trade, and to deter each other militarily.

The relationship is becoming multidimensional. This is more difficult to explain to the public than a Cold War narrative, but much closer to reality. Political language will have to become more complex. This is probably one of the greatest challenges facing democracy. Societies want simple narratives (friend or foe), whereas strategy often requires being a competitor, partner, or adversary, depending on the domain.

This ambiguity is not a weakness. It is precision. This is precisely the precision I seek throughout this section of the book: to reject fictitious blocs. Iran is not Shiism. Daesh is not Islam. China is not a homogenous entity. The United States is not a single will. Every system contains institutions, businesses, factions, and interests.

A great power is itself a graph.

The Chinese Communist Party is centralized, but it must administer provinces, businesses, the military, the population, and the economy. Washington has a different structure: Congress, the White House, states, businesses, courts, and public opinion. Comparing the two therefore requires understanding their modes of coordination.

A Chinese decision can be implemented quickly when it is centrally prioritized. An American decision may be challenged, delayed, but this dispersion can also generate more independent innovation. Again, the trade-off. There is no optimal structure in all circumstances. War favors certain architectures, innovation others. The challenge is to be able to change modes without losing one's identity.

That's precisely the definition of resilience, and I keep coming back to the idea of transformation. I believe this entire geopolitical landscape converges on that word. Empires don't simply die because they become weak; they often die because they can no longer transform their institutions as quickly as the world around them. China is transforming. The United States is transforming. Europe must transform.

AI is accelerating the speed at which this mutation becomes necessary. The world to come will likely not leave much room for rigid systems. This is where the Total Work of Art becomes almost a conceptual model. The Abode of Chaos is never finished. It maintains its structure while continually changing. It absorbs the event without losing its identity. This is precisely the property I call mutant permanence.

A 21st-century institution will likely need to function in this way: possessing a stable doctrinal core and forms flexible enough to adapt. For a democracy, the core might be the rule of law, dignity, and checks and balances; administrative and technological forms may change. For a business, the core might be its mission, while its tools evolve. For a state, sovereignty and institutions must outlast constantly evolving infrastructures.

China is currently building its own model of enduring, ever-changing power. The question remains: how far can it go? And behind this question lies another, broader one: can we build a global architecture in which multiple models of power coexist without necessarily seeking to destroy the others? This will likely be the defining geopolitical question of the artificial intelligence age.

Because AI adds another dimension: it could significantly amplify the power gap between systems that master it and those that don't. A system with better models, more energy, data, robots, and industrial capabilities could accelerate its own development. We may be entering positive feedback loops where technological advancement generates more ways to increase that lead.

If this proves true, competition will become much more intense. States will seek to avoid falling below a threshold of cognitive dependence. This is a strange but essential expression: depending on another to calculate, analyze, translate, predict, and design. We have experienced energy dependence; we may now experience cognitive dependence.

A country relying exclusively on models developed elsewhere may end up outsourcing part of its way of understanding the world. Even without deliberate manipulation, the model carries with it choices of data, categories, languages, and standards.

Cognitive sovereignty thus becomes a new layer of sovereignty.

This does not mean that each country should manufacture its own giant model; that would be impossible and often unnecessary.

But it must understand the tools it depends on, be able to audit them, have alternatives, and maintain its own knowledge base and skills. Here again: manage dependence rather than fantasize about self-sufficiency. China has understood this. So have the United States. Europe is beginning to understand it.

The next chapter must therefore delve into this rivalry, not only through physical infrastructure, but also through what flows above it: models, data, computing power, artificial intelligence. For after the road, the port, the chip, and the network comes the real challenge: who produces the systems capable of interpreting all the data in the architecture? Therein lies the beginning of a new form of power.

A power that no longer merely transports goods or information, but can begin to understand them on a superhuman scale. The next chapter must therefore delve into what I consider the most profound strategic shift since the advent of nuclear weapons: not simply the arrival of a new tool, but the emergence of an industrial cognitive capacity capable of permeating economics, warfare, science, and politics.

The United States, China, and Europe: three different architectures will now attempt to define the place of this intelligence in their societies. And behind the industrial competition will emerge a much more dizzying question: what becomes of sovereignty when the capacity to understand the world itself becomes an infrastructure that can be owned, rented, controlled, or lost?