The most useful way to think about American competition with China is not through the language of moral theater, which is how almost every domestic political question is now presented, but through the older and less flattering language of power. Civilizations rise because they accumulate capacity, and they decline because they lose it. Capacity means energy, capital, military competence, technical knowledge, industrial depth, infrastructure, institutional confidence and, increasingly, control over artificial intelligence. None of these things is mystical. They are built, maintained, financed and defended, and if one society decides that maintaining them is vulgar while another decides that doing so is essential, the long-term outcome is not difficult to predict.

This is why the strange visual culture of early Covid deserves to be remembered, especially the dramatic videos from China of citizens suddenly collapsing in the street as the virus struck them like an absurd monster in a low budget horror film. The West demonstrated how rapidly fear, imagery and social imitation could synchronize entire populations before the underlying facts were well understood. Videos of apparent collapses in streets, sealed apartment buildings and men in protective suits circulated globally with the emotional grammar of a disaster film, and governments, institutions and ordinary citizens reacted not merely to epidemiology but to a mediated image of catastrophe. Whatever one thinks of the eventual merits of particular policies, the strategic lesson was obvious: highly networked liberal societies are extremely vulnerable to narrative cascades.

An intelligent adversary does not need to create every weakness inside a rival society. Usually it is enough to recognize the weaknesses that already exist and learn how they behave. If a population can be induced to move rapidly from image to outrage, from outrage to moral certainty, and from moral certainty to policy, then the cost of influencing that society falls dramatically. One does not need millions of paid agents, because most of the work is performed voluntarily by people who sincerely believe they are acting from independent conviction. In fact, sincerity is preferable to conspiracy. The activist who requires no payment, no instruction and no coordination is a more efficient political instrument than the conscious operative.

This is the mechanism that matters now in the politics of artificial intelligence. The United States possesses enormous advantages in capital, research, software, entrepreneurship and advanced computing, but all of those abstractions eventually collide with physical reality. AI requires chips, and chips require fabrication capacity. Compute requires enormous amounts of electricity. Electricity requires generation and transmission. Training and inference require data centers, transformers, cooling systems, land, water, permitting and industrial construction. The supposedly immaterial digital economy eventually becomes very material indeed, and once it does, it becomes vulnerable to the ordinary American system of veto, litigation, protest, delay and moralized obstruction.

This creates an obvious strategic asymmetry. China does not have to become more capable at every margin if the United States can be persuaded to become less capable at its own. Relative power increases either because one party rises or because the other descends. The second method is often cheaper, and it has the additional advantage of allowing the declining society to interpret its own decline as virtue.

This is the recurring temptation of modern Western politics. Energy abundance becomes environmentally suspect. Industrial construction becomes socially disruptive. Infrastructure becomes ugly. Corporate scale becomes immoral. Semiconductor fabrication consumes too much water. Data centers consume too much electricity. Transmission lines injure viewsheds. Nuclear power is frightening. Fossil fuels are wicked. The objection changes, but the institutional result is remarkably stable: less capacity, built more slowly, at greater expense.

The most sophisticated version of this politics does not openly say that the West should become weaker. It says almost the opposite. It insists that one particular factory, one particular data center, one particular transmission line or one particular energy project is uniquely unnecessary, dangerous or unjust. Each individual objection can sound modest and reasonable. The aggregate effect can nevertheless be civilizational self-reduction.

This is where the activist class becomes strategically important. Its members need not be stupid, and they certainly need not be consciously disloyal. What they are often unable to perceive is the system-level consequence of their preferred method of politics. Modern activism is exceptionally good at generating concentrated pressure against discrete projects while remaining almost completely insulated from responsibility for the cumulative consequences. The activist can oppose a power plant without becoming responsible for electricity shortages, oppose a housing project without becoming responsible for housing scarcity, oppose a data center without becoming responsible for reduced compute capacity, and oppose industrial development without assuming ownership of industrial decline. The moral credit is immediate, while the material cost arrives later and is distributed across everyone.

That is an extraordinarily attractive political technology, because it converts obstruction into virtue while separating intention from consequence. It is also a technology that foreign competitors would be foolish not to study.

The Chinese strategic interest here requires no elaborate theory. If the United States constrains energy while China expands it, China benefits. If the United States makes large-scale computing infrastructure harder to build while China makes it easier, China benefits. If American political culture teaches its citizens to regard national technological dominance as morally embarrassing, China benefits. None of this proves foreign direction of any particular protest movement, and serious people should resist the temptation to turn every domestic disagreement into espionage. The more unsettling point is that direct control may not be necessary at all. A rival power can gain from tendencies that are entirely indigenous.

The Covid episode should therefore be remembered less as a story about secret authorship than as a demonstration of susceptibility. The extraordinary fact was not that frightening images existed. Frightening images have always existed. The extraordinary fact was the velocity with which emotional imagery became institutional consensus. A society that believes itself radically individualistic revealed an almost herd-like capacity for synchronized reaction, and that fact is strategically useful information whether anyone intended to provide it or not.

The same mechanism now appears throughout contemporary politics. A morally compelling narrative emerges, a villain is identified, a simplified causal story is circulated, and thousands of people adopt the story because doing so signals membership in the correct social coalition. The details often matter less than the pattern. A person can know almost nothing about grid capacity, transformer shortages, energy markets, model training, semiconductor supply chains or the strategic economics of AI and nevertheless become intensely certain that a particular data center represents an intolerable injustice. The absence of technical understanding does not weaken the political conviction; often it strengthens it, because technical complexity interferes with moral clarity.

This habit is particularly dangerous in a period of technological competition because industrial power is cumulative. A country cannot simply wake up ten years later and decide that it would now like to have an electricity grid, a semiconductor industry, a hyperscale computing sector and a deep bench of AI infrastructure. These systems are path-dependent. They require years of investment, supply chains, permitting, skilled labor, capital formation and institutional learning. Lost time is not easily recovered.

Europe provides a useful warning. It remains wealthy, educated and scientifically accomplished, yet it has become strikingly better at regulating many frontier technologies than at producing the companies that dominate them. This is not because Europeans lack intelligence. It is because intelligence is not sufficient. A civilization also needs institutional permission to build. If every major industrial project becomes an extended negotiation among regulators, activists, local veto players and political intermediaries, the civilization may eventually discover that it has preserved every procedural value except the capacity to shape the future.

The United States is now flirting with the same mistake at precisely the moment when AI may become the most important general-purpose technology since electrification.

There is, however, an irony here that should not be missed. The major technology companies helped create much of the political environment now turning against them. For years, parts of Silicon Valley treated ideological activism as something between a moral ornament and a useful instrument. Companies funded organizations, adopted activist vocabularies, incorporated political doctrine into corporate bureaucracy and sometimes participated in a broader culture of suppressing disfavored views. They apparently assumed that this political machinery would remain permanently directed at other people.

This was naïve.

Political movements organized around identifying systems of domination do not retire when yesterday’s enemies have been defeated. They require new objects. Wealth, scale and power are themselves suspicious categories, and eventually the technology companies that imagined themselves progressive allies become merely another concentration of wealth, scale and power. The patron discovers that patronage did not purchase immunity.

There is some justice in allowing the technology industry to experience this lesson. It may be useful for executives who once regarded ideological coercion as harmless to discover what ideological coercion feels like when directed at infrastructure they consider essential. It may be educational for people who tolerated caricatures of other industries to hear their own work described through caricature. Political pain can teach what abstract argument does not.

But the lesson has to stop short of destruction. A serious country cannot afford to confuse revenge with strategy. The United States needs its technology companies, its engineers, its capital markets, its semiconductor firms, its laboratories and its data centers. If the ideological machinery that Silicon Valley once tolerated now succeeds in crippling American AI capacity, the result is not poetic justice. It is national stupidity.

The correct response is therefore alignment, although that word needs to be rescued from both corporate jargon and authoritarian fantasy.

Alignment does not mean suppressing dissent, declaring every protester an agent of Beijing or demanding obedience to whichever faction currently holds power. A country that protects itself from Chinese authoritarianism by imitating Chinese authoritarianism has missed the point in a rather spectacular way.

Alignment means something more basic: establishing agreement about the continued existence of the civilization itself as a first-rank power.

Americans can disagree about taxation, environmental standards, corporate concentration, labor policy, copyright, privacy, model safety, water use, land use and the distribution of gains from AI. They should disagree. A free society is supposed to contain conflict.

What cannot remain infinitely negotiable is whether the United States intends to possess the underlying capacity at all.

A functional political order would begin from several propositions that are treated not as partisan preferences but as civilizational requirements: the United States intends to remain wealthy, technologically advanced, militarily capable, energy-abundant and competitive in artificial intelligence. Within those boundaries, politics can rage.

Without those boundaries, politics becomes a process by which every faction extracts moral satisfaction from dismantling some portion of the common inheritance.

China does not need Americans to admire China. It does not need them to become communists, embrace censorship or celebrate authoritarian government. It merely benefits if Americans become unable to build the things required to remain powerful.

That is the more elegant form of competition, because the loser performs the work himself.

A civilization can be conquered from outside, but it can also be persuaded that capacity is oppression, confidence is arrogance, hierarchy is injustice, industrial strength is pollution and national interest is morally suspect. Once those propositions become sufficiently embedded, decline acquires an almost ceremonial quality. Each individual act of self-limitation is celebrated as progress, while the aggregate result is a society that possesses ever fewer instruments with which to determine its own future.

That is the alignment problem in its most serious form. The choice is not between becoming authoritarian and becoming virtuous. The choice is between retaining enough power to remain politically independent and discovering, after the fact, that moral prestige does not compensate for strategic irrelevance.

The countries that understand this will build.

The countries that do not will explain why they were too enlightened to do so.