The Evolution of Power in the Age of AI: From Lawmaking to System Architecture and Access Regimes

Glossary

Contents

Introduction

Imagine you decide to order a batch of industrial GPUs, secure delivery of HBM memory, or simply reserve capacity in a major data center—not for a week of entertainment, but as the foundation of a business. Reality hits you in the face: it’s not visible in stock quotes or news feeds, but teams from startups to labs hear the same thing: “everything is under contract; the nearest window is in years”.

The first reaction is soothing, almost instinctive: it’s just a cycle. The market overheated. A hype phase. They’ll build more fabs, bring new lines online, ship fresh batches—and everything will return to the familiar track. But look with a strategist’s eyes, and the picture transforms. This is not a classic shortage; it is a radical reset of the very concept of access.

Computing is evolving from a simple commodity into a privilege—a right of entry. Controlling such a right is easier, cheaper, and politically cleaner than controlling ideas, software, or people. Worst of all, this metamorphosis is happening not through blunt prohibitions, but under the mask of “convenience”, “security”, “standards”, and “compliance”—words that are as pointless to argue with as the necessity of fastening a seat belt.

This is not a caricature conspiracy with backstage puppeteers. It is, rather, a system that assembles itself: states benefit from manageability, corporations from rent and market grip, users from comfort, engineers from risk minimization and rising reliability. The result is a mechanism immune to exposure. There is nothing to expose—only a chain of incentives and the inexorable physics of infrastructure.

1. Shortage as Discipline: From Market to Regime

The word “shortage” triggers an outdated mental model for most people: “there isn’t enough—so we’ll endure for a while”. But a shortage of a strategic resource almost never remains a temporary inconvenience. It becomes an instrument of discipline. Because under scarcity the system always chooses one of two things: either to frantically scale volumes, or to rigidly allocate access.

We are used to believing that the market always expands supply. But that is true for goods that can be produced massively, quickly, and relatively decentrally. Modern compute stands apart: it is not a one-time purchase but an endless stream of costs, an infrastructure maze, a talent shortage, a web of regulations and security measures. Most importantly, it directly converts resources into dominance over speed—into the ability to outrun time.

This is where controlled-access computing power comes into force: once compute acquires the status of a scarce strategic asset, it stops being a commodity for everyone. It becomes more convenient to ration it—through quotas, contracts, clearances, priorities, client classes—as if it were access to a guarded arsenal.

In effect, scarcity becomes a pre-allocation of the future. Whoever has contracted capacity has not contracted hardware. They have contracted time—and therefore iterations, experiments, the right to be wrong, the right to try more. They have purchased not a product, but a probability of winning.

And here scarcity steps beyond economics, becoming a political-economy lever. The allocation of the “right to develop” is no longer a market; it is a pure form of power, where access determines who evolves and who remains stuck in the past.

2. Why Compute Is Easier to Control—and What That Changes

Any authority that tries to control “ideas” or “software” quickly runs into the fact that ideas are copied, software leaks, texts are rewritten, algorithms are multiplied and developed in new branches. The stronger the ban and the more needed the product, the faster a workaround appears. Control at the level of meaning is expensive, conflictual, and endless.

Controlling compute is different. It is almost ideal. Because compute can be:

• measured;
• priced;
• tied to a person, a jurisdiction, and rules;
• technically shut off—with a single command.

That is why what comes to the foreground is not a ban, but architectural violence. The system does not say: “don’t do it”. It offers an inconvenient solution: “do it, but here is friction”. The free tier is slow. Paid is faster. Corporate is without restrictions. Government clearance is no questions asked. And every time it looks not like repression, but like a reasonable segmentation of services.

The most effective control is the kind that feels like convenience.

• Direct violence is a baton.
• Economic violence is poverty.
• Architectural violence is a system in which you formally have a choice, but in practice you do not.

In practice, this leads to computational qualification working even without law and without police. Simply because “heavy” capabilities migrate “by default” to where control is easier to ensure: into cloud perimeters, into major players, into regulated venues. On the user side, what remains is either a convenient terminal, or an exhausting struggle for autonomy—like swimming against the current, where every stroke is full of constant resistance, draining your strength and forcing you to surrender to the flow.

This is where the psychological reversal happens. People are ready to defend freedom of speech. But they rarely defend “freedom of computing”—because compute seems technical, far from rights. Yet in the age of AI, compute becomes what machine tools and power grids were in the industrial era: the basic infrastructure of action and the key means of production. This is no longer about what you think. It is about what you can do.

3. Acceleration of Invention: Not the Cause of Progress, but a Catalyst of Monopoly

In this article it is important not to fall into the magic of the word “AGI”. Even without a full artificial general intelligence, strong models are already reshaping the pace of research: they accelerate analysis, solution generation, compromise search, modeling. The loop “idea — prototype — test — error — fix” is compressed to the limit.

That is exactly how an iterative monopoly is born. In a world where success is the statistics of attempts, the winner is whoever runs more iterations in the same time. That sounds banal until you understand the consequences: whoever owns compute begins to own the selection mechanism itself.

Before, one could believe in the myth of the lone genius. Today genius increasingly becomes a function of resources: how many variants you can test before the market or regulator rewrites the rules. More attempts means a higher chance of a breakthrough. So a shortage of compute automatically becomes a monopoly on tempo. Even if you are more talented, you make 10 attempts while they make 10,000. That is no longer competition. It is different sports.

4. Shortages “Move”—Which Means the Role of Those Who Can Live in Turbulence Grows

In debates about shortages, the focus is usually on what is missing today. But strategic thinking is elsewhere: bottlenecks will migrate. The tech stack evolves, infrastructure requirements transform, and with them the types of shortages. Every such shift is a reason for big players to dig in even deeper.

Why? Because “living in turbulence” is expensive. You need to keep options, maintain a buffer of resilience, jump between technologies without stopping the business. Only those already inside large perimeters can afford that.

Here computational qualification reveals itself in its full harshness: it is not merely a barrier to access, but a tool that turns future innovation into a closed club. An idea may be born in a garage, but its development runs into infrastructure that is increasingly acquiring a restricted-regime character. The garage becomes a romantic myth.

The key shift: we are used to assessing the freedom of innovation by the presence of fresh ideas and ready code. But in the AI era, real freedom is defined by development tempo—and it is inseparable from access to infrastructure.

5. Power Moves into the Unified Compute Perimeter: How Personal Devices Become an Entry Point

The phrase “people should not have personal computing power” sounds like dystopia, but its logic is entirely earthy. The meaning of owning your own “hardware” will fade not because devices degrade—on the contrary, the market will keep accelerating power in every pocket for a few more years: more cores, neural blocks, memory, everything sold under the sauce of progress and mobile gaming. This race will inevitably run into a fundamental break: a consumer gadget will become strong enough to run local language models and autonomous agents—without the cloud, accounts, rules, or oversight.

And this is where a conflict begins that cannot be resolved with moralizing and terms of service. Local power makes control almost impossible. If an autonomous agent can live on a phone, it doesn’t need a provider, it doesn’t need a perimeter, it doesn’t need a profile check. Any teenager gains the ability to run systems that until recently required a laboratory and a budget. Even if ninety-nine percent of users never do anything dangerous, governance will still be designed around the remaining one percent. In infrastructure logic, this is not a moral question; it is a risk question. And when risk cannot be contained programmatically, it gets choked architecturally.

The fork is predetermined: for the sake of manageability, states and mega-platforms will squeeze compute out of gadgets back into the perimeter. They won’t ban phones; they will make power regime-controlled—only where it can be regulated. The device in your pocket will gradually become an entry point: a screen, camera, microphone, connectivity, an identification module, and the minimum local compute for interface and encryption. Heavy tasks will move remote, and the phone will turn from a computer into a key to a compute elevator.

It will not be presented as taking something away. On the contrary, it will be sold as relief: longer battery life, less heating, cheaper hardware, no worries about updates, always fresh models, instant scalability, ready environments, security by default. You will be offered a trade: less power in your hands, more by subscription. Old gadgets will be taken in for trade-in, tariffs subsidized by carriers, and the new standard will become so natural that within a couple of generations no one will remember why a pocket ever carried a full compute engine.

This solution suits all influential players at once. Platforms benefit from returning compute to data centers, because that is where rent, control, and iterative monopoly are born. States benefit because controlled-access computing power is easier to keep inside licensed infrastructure, to surround with thresholds and clearances, and to embed mandatory rules into. Users benefit because they are sold comfort and have hassles removed. The transition will happen without slogans: not by bans, but by architectural violence, where autonomy formally remains possible, but becomes rare, expensive, suspicious, and inconvenient.

And this is where the final assembly comes together: compute concentrates in large cloud perimeters, becomes cheaper per result, and local autonomy ceases to be the baseline mode of life. A person is left with a thin device: a screen, glasses, a terminal. Everything beautiful, light, “eternal”. But it comes alive only when connected to the network and after passing checks: account, biometrics, jurisdiction, clearance. The perimeter may be physically distributed and even geopolitically fragmented, but to the user it looks like a single outlet: either you are in the perimeter, or you are out of the game.

That is why the unified compute perimeter is not a fantasy about one world server. It is a model in which access becomes the only form of existence for power. And access is a managed service. A managed service always contains a switch. Not a prison switch and not a public one. A technical one. You are not punished; you are simply disconnected: the tariff changed, a function is unavailable, access is limited, the account requires verification, the perimeter in your jurisdiction runs by different rules. The power of the future does not need to shout. It only needs to change the parameters of the environment.

The result is the illusion that nothing has been lost: everything is faster, cleaner, more convenient. But the main thing disappears—the right to autonomous experimentation. You are left with a storefront (the interface), while the workshop (the regime) evaporates. You can get outcomes, but you can no longer freely choose the path to them: run heavy tasks locally, make mistakes without a meter, build non-standard pipelines without someone else’s rules and logs. The device is still in your hands, but computational power migrates into the perimeter—where it is measured, priced, and shut off. This is the end of sovereign power: computing moves from ownership into clearance, and clearance by its nature is revocable.

6. Quota Censorship as the Main Mechanism of “Soft” Restriction

Old censorship is event-based. It looks like a ban. You can hate it, challenge it, circumvent it. You can rise up against it—and change the system that produced it, whether communist, Orange, or any other. New censorship is architectural. It looks like a tariff and a limit. It does not cause a revolt—it causes calculation. You do not overcome it with a revolution. You accept it by choosing the “Freedom-Premium” package with a higher limit on disagreement.

That is quota censorship: restricting freedom through throughput. Not “no”, but “yes, but within limits”: less speed, less quality, less reach, fewer parallel tasks, shorter context length, fewer attempts, fewer “heavy” functions. On the base level you can formally “do everything”, but so slowly and at such small scale that real capabilities remain with those who pay for scale. Freedom turns into a commodity, and poverty starts to mean not only less comfort, but less visibility, less competitiveness, less ability to defend yourself.

But quota censorship has a more precise—and more dangerous—dimension. Quotas and tariffs are the visible part. The invisible part is time.

Consequence: the power of delay

The “cleanest” form of control in a digital system is not a ban and not even a suspension. It is delay. A ban generates conflict, appeals, and headlines. Delay always looks natural: a queue, prioritization, rate limits, cooldowns, “night windows”, limits on session duration, on parallel runs, on repetition frequency, on result delivery speed. And that is exactly why delay is almost impossible to prove as repression: it always has a plausible engineering explanation—“balancing”, “load distribution”, “scheduler fairness”.

It is important here not to confuse this with the iterative monopoly discussed in previous sections. There it is about large players winning the race through tempo. Here it is different: even when you have not been “forbidden”, your rhythm can be taken away. And without rhythm, any activity tied to compute degrades: a project turns from a stream into rare bursts, research from a series into single shots, a product from a cycle into “someday”.

The power of delay targets continuity, turning action into an appointment. Without feeling a ban, you feel like you are always waiting. And that disciplines the market better than any slogan: directions and behavior models that fit into cheap throughput become the “norm”; everything that requires dense tempo gradually moves into privileged layers of access.

Quota censorship is insidious in its apolitical nature: it doesn’t need to prove you wrong. An invoice or a queue is enough—and a person does not argue with the system, but adjusts to its scheduler.

7. The Trustworthiness Gate: Social Rating Without the Signboard, and Censorship of Intention

When access to compute becomes regime-controlled, a question arises: whom do you let into “heavy” modes, and whom do you not? In a direct form this looks like discrimination and provokes resistance. So the system does it differently: it introduces “technical criteria”. That is how the trustworthiness gate appears.

They won’t tell you “forbidden”—they’ll tell you “verification required”, “anomalous activity”, “regional restrictions”, “temporary measure”, “service policy”. The reason hides under technology, and technology is not perceived as politics—it is accepted as a given.

It is important to distinguish this from quota censorship.

Quotas are when everyone gets a meter, and throughput is sold by tariff.

A gate is when the meter becomes personal: it depends on who you are, how you behave, and where you are going.

Consequence: censorship of intention

Classic censorship worked on words and outcomes: what you said, what you published, what you showed. In a compute regime, the focus shifts: they start evaluating the trajectory of actions—what you are trying to do.

Why is the trajectory more important than the outcome? Intention is read in advance: from input data, connected libraries, session duration, repeated experiments, parameter changes, scale ramp-up, attempts to bypass limits. This is not censorship of the “word”. It is censorship of turning the word into action.

The system does not need to prove harm—a suspicion of a risky vector is enough. After that you are not “punished” in the old sense—you are moved into a different layer of reality: a sandbox, enhanced logging, restricted modes, additional requirements, more complicated access. Formally: a security standard. In fact: a mechanism that governs not your result, but your direction.

8. The Prompt Proletariat and a New Class Structure

When infrastructure concentrates, a new class fault line appears in society. Not “rich and poor”, but those who own the means of producing meaning, and those who service them.

Prompt proletariat is not an insult. It is a cold description of an economic role: a huge mass of people will “work with AI” but own neither models, nor rules, nor compute. They will be operators of someone else’s machine tool. And here lies fundamental alienation: a person can be talented, productive, creative—yet their result is not secured with them, but with the owner of the infrastructure.

The unpleasant truth is that many “new professions” rest not on craft, but on access to someone else’s storefront. Today you are a master: you know how to formulate queries, which modes to turn on, how to assemble toolchains to get the needed quality. Tomorrow the platform updates the model, changes the interface, moves functions into a different tariff, adds filters—and half your techniques stop working. Not because you became worse, but because the tool that anchored the skill does not belong to you and is not fixed.

It is like a virtuoso machine operator in a rented shop: as long as access is open and the machine is familiar, you’re a pro. But with equipment replacement and rewriting entry rules, you are back at the start. The skill clung to the storefront of a specific service version, not to the foundation—your own compute, data, and autonomous tool. Therefore such a profession is by definition fragile: it lives not on your competence, but on the stability of someone else’s rules. And the faster the perimeter updates, the more often mastery is “reset”—not as a personal tragedy, but as the normal mechanics of an access market.

This is where a new hierarchy grows. The elite differs not because it is “smarter”, but because it is anchored inside the regime: it has the right to experiment, to be wrong, to scale. The majority receives only the right to use—convenient, mass, but revocable. And this is exactly how the prompt proletariat becomes not a metaphor, but a social layer.

9. Outsourcing Responsibility: Power Without an Author

When the state imposes restrictions, you have an addressee and a form of protest: court, rally, political demand. When a corporation imposes restrictions, it hides behind the user agreement—you are not a citizen but a client, and your protest turns into a support ticket. But the most stable mechanism is born when these two modes merge and begin to refer to each other like mirrors. The state delegates “standards compliance” to the corporation, and the corporation appeals to “regulatory requirements”. In this closed circuit of power, your protest loses its addressee. You cannot sue an algorithm, and a political demand shatters against “the platform’s technical requirements” and “current legislation”. Power becomes distributed and impersonal—and therefore invulnerable.

This is the outsourcing of responsibility. The platform says: “we comply with the requirements of the law”. The state says: “it’s a private company; it decides itself”. The result is a zone where there is action and consequences—but no author who answers. You cannot sue an algorithm. You cannot force a platform to “restore justice” if it violates its policy. You cannot get an answer from the state, because formally it “did not ban anything”.

It is a mechanism of power dissolved in the air: everyone controls, no one is guilty. And that is exactly why this mechanism cannot be broken by exposures. There is no “secret center”. There is perfectly organized irresponsibility.

10. The State Returns Through Compute

In the 2010s it was fashionable to insist that the internet bypasses states: information flows over borders, communities self-organize, the market sets the rules. That held as long as the resource was ruled by ideas, code, and attention. But compute is not ephemeral: it has an address, a meter, megawatts, contracts for water and substations. It leaves traces in accounting, in power grids, and in documents. As soon as compute turns into a strategic asset, the state storms in not as a censor, but as the owner of infrastructure levers.

The essence of the return is simple: the state does not need to “control thoughts”. It only needs to control the points where thought turns into action: access to hardware, to power, to energy, and to networks. That does not always look like politics. More often it looks like “norms”, “regulations”, “security”, “responsibility”, and “resilience”.

Below are four tools that are too convenient not to use.

1) Geopolitics: export perimeters and sanctions on hardware

Export control is not a ban on knowledge. It is a regulatory lever over acceleration. In the modern world you can have a team, an algorithm, data, money—and still run into the fact that a physical accelerator, memory, packaging, network components, or a service is unavailable in your jurisdiction or for your profile.

The key effect here is not “stopping progress”, but selecting it. Export perimeters do not make innovation impossible; they make it asymmetric by tempo. Some are given the ability to try quickly, be wrong, and scale. Others are left the right to “do something”, but so slowly and expensively that it stops being a competition. Geopolitics embeds itself into compute almost silently: not through a loud “no”, but through the disappearance of a delivery window, through an intermediary’s refusal, through “regional restrictions”, through a ban on support and updates.

At the level of the individual, it feels like strange physics of the world: everything seems global, but in fact “the same” products work differently because they live inside different perimeters. That is fragmentation as a function of infrastructure, not ideology.

2) Restricted facilities: licensing data centers and infrastructure

Computing power, as a strategic resource, turns data centers from a business into critical infrastructure—on par with connectivity, energy, transport. At this moment the legal and technical logic of controlled-access computing power appears: clearances, regulations, audits, requirements for physical protection, for personnel, for supply chains, for incident management.

This is an important point: the state doesn’t even need to directly “regulate models”. It is enough to regulate the room where models are computed. A data center is a concentration point of three things the state has historically known how to control: land/construction, energy, security. And that gives a lever stronger than many internet laws.

A telling example of future “normality”: a large data center asks permission for additional megawatts and connection to a substation. It is told: yes—but on the condition that mandatory logging mechanisms, client segmentation, enforcement of sanctions lists, prioritization regimes, and emergency load shedding are implemented. Outwardly this looks like “power-grid resilience” and “facility security”. In fact it is architectural firmware of control at the infrastructure level: the access system is embedded into hardware and energy.

And that is why the state returns not as a prohibitor, but as a regime administrator: “we are simply ensuring order at a strategic facility”.

3) Threshold regulation: instead of “AI is banned”—“above a power threshold, under oversight”

This is the most technologically elegant move. Banning “AI” is impossible—too broad, too conflictual, too much useful stuff. But introducing a regime where a certain level of power automatically moves activity into a different legal class is entirely feasible.

The logic of threshold regulation is this: as long as you operate in the “household” range, you are like an ordinary user. Once you cross the threshold (by compute volume, by launch nature, by cluster scale, by task type) you move into a regime where registration, audits, reporting, additional checks, and sometimes licensing or a “responsible operator” are required.

Why does it work? Because it is not censorship of content. It is regulation of power as potential impact. This frame sells easily to society: “we are not against technology; we are against uncontrolled risk”. And it hardly provokes resistance because it sounds reasonable and caring.

But threshold regulation has a deeper effect: it reinforces censorship of intention. When the system evaluates not what you said, but what you are trying to run, the very fact of “approaching the threshold” becomes a risk signal. That pushes people and businesses toward self-correction: choose “safe” directions, “lightweight” versions of projects, “proper” tasks—not because they will jail you, but because otherwise you will fall into a regime where everything is more expensive, slower, and more formal.

In other words, the state intervenes not in ideas, but in trajectories of development. And it does so through metrics and compliance—maximally civilized.

4) Identity: the service requires an account, the account requires a passport

The fourth lever is the strongest psychologically. When compute becomes a regime-controlled resource, a question arises: “who exactly gets access to power, and who is responsible for the consequences?” Technically this is solved in the simplest way: by tying compute to identity.

At first it looks harmless: “for security you need to confirm your identity”, “for a corporate tariff you need a contract”, “for extended functions—verification”. Then “standards” are added—and the trustworthiness gate becomes the norm not only for specific tasks, but for the very ability to use heavy power without friction.

An important nuance: identity here means not only “who you are”, but “how manageable you are”. A person is an address, jurisdiction, responsibility, sanctionability. Anonymous access to power is autonomy. And autonomy, in a strategic-resource regime, is perceived as a defect of the system.

That is why a “computational passport” will emerge not as a slogan, but as service mechanics: it will be explained to you as risk reduction. And most will agree—because the alternative will be burdened with an autonomy tax: longer, more expensive, more complicated, with more bureaucratic friction.

Result: the state returns not through ideology, but through infrastructure

If you put these four levers into a single picture, it becomes clear why the state is returning everywhere. Not because “someone is evil”, but because the control tools are too convenient:

• export perimeters let you manage tempo and access at the supply level;
• the regime status of data centers lets you control compute through energy, security, and licensing;
• threshold regulation lets you oversee not “words”, but power and potential impact;
• identity turns access into a managed service, where responsibility can be assigned and collected.

The conclusion is cold: you don’t need to imprison people for a thought. It’s enough not to grant power so that a thought cannot become action. And it will look not like repression, but like “risk management” and “normal operation of a strategic resource”. That is why this layer is dangerous: it does not ask society’s consent for control—it brings control under the guise of servicing.

11. A Likely Future Scenario: Convenience, Feudalism, and Fragmentation in One System

If you put everything together, you don’t get three separate scenarios, but one system inside a unified compute perimeter. It rests not on bans, but on common sense: it looks like a convenient service, but in fact it sets the rules of access.

The upper regime is user-facing: it promises comfort and removes headaches.

The middle regime is a clearance regime: it allocates rights, tariffs, and capabilities.

The lower regime is a jurisdiction regime: it defines the geography of the switch and the real boundaries of what is permitted.

1) User regime: convenience as the new norm

For the person of the future, everything will feel simple: intelligence and power have become a public utility. Like water, connectivity, or electricity. You pay a subscription and you don’t need to understand what is inside: GPUs, clusters, memory, routing. You press a button—and the task is solved.

On the surface, this is progress. But this is exactly where architectural violence starts to work: alternatives are not banned; they are surrounded with friction.

You still have a computer, as before. But it gradually becomes limited power: locally you can consume and execute—watch, write, play, edit, ask. Everything that requires real autonomy—heavy configuration, training, scaling, complex simulations—naturally migrates to the cloud. Not because it was banned for you, but because it is more stable, safer, faster, cheaper per result.

At this moment freedom changes form. Before, you owned the tool and could do with it everything it could physically handle. Now you own only the entry point into a service. And the conditions of that service can be changed without debate: not by law and not by a ban, but by a tariff update, a platform policy, a new security norm. It is perceived not as interference, but as servicing. As a regulation.

The key psychological trick is simple: when the system gives you a lot of comfort, you stop experiencing dependence as dependence. You call it convenience. And it is convenient exactly enough to ensure you don’t want to leave the perimeter.

2) Clearance regime: digital feudalism as an economy of access

Once compute becomes a service, a question inevitably arises: who is entitled to how much power. And here convenience turns into hierarchy. Not because someone decided to build a dystopia, but because that is how risk management works under a strategic resource.

Inside a unified compute perimeter, access classes form—packaged as tariffs, profiles, and clearances. This is a new type of digital feudalism:

• for the majority—a comfortable consumer level
• for professionals—expanded capabilities in exchange for reporting and restrictions
• for corporations—near-full power in exchange for compliance and control
• for the state and quasi-state structures—separate regimes and exceptions

It will not be called estates. It will be called a product line. But in essence it is computational qualification: access to heavy capabilities is separated from access to ordinary functions. Formally everyone gets something, but real power belongs to those who cross the threshold. And that threshold will not be only about money. It will be about jurisdiction, status, legality, reputation, corporate obligations, audit.

At the entrance to real power stands the trustworthiness gate. And that changes the sense of freedom. You start living not in the logic of what I can, but in the logic of what I will be given. The system doesn’t say no. It says unavailable for your profile; confirmation required; suspicious activity analysis; regional restrictions; additional verification required.

Within this regime, quota censorship operates: freedom is cut not by bans, but by limits. You can do almost everything, but on the base level slowly and with restrictions; on the middle level faster, but under logging and in safe mode; on the top level broadly, quickly, and almost without friction.

And the most unpleasant part: such censorship does not provoke protest. It provokes motivation. Upgrade your tariff. Become reliable. Don’t ask unnecessary questions. People self-discipline because otherwise life is inconvenient.

This is also where the prompt proletariat grows. And finally, inside this regime the iterative monopoly is cemented, and the statistical possibility of garage revolutions disappears.

3) Jurisdiction regime: the geography of the switch

If the user regime promises comfort and the clearance regime allocates rights, the jurisdiction regime defines the boundaries of the possible. It covers the unified compute perimeter with a layer of rules: where you are, under which jurisdiction, which export restrictions apply, which compliance standards are mandatory.

From the outside everything looks the same: the same applications, interfaces, similar models, the same clouds. But inside, people live in different computational realities: the same query, the same task, the same product yields different results because:

• in one country a function is available, in another it is regulatorily limited
• somewhere you can work with certain data, somewhere only with local storage
• in one perimeter an account is enough, in another identification and extended audit are mandatory
• somewhere safe mode is optional, somewhere it is the default.

A mini-scenario looks like this.

A person has two passports—for example, Chinese and Cypriot. In China he lives and works in a large economy, but some global tools and services are either unavailable, or constantly run into regulations, or require separate bypass perimeters. Formally there is a choice. In practice, any “undesirable” trajectory is surrounded with friction: less compatibility, more checks, more restrictions, more time spent on access logistics. It’s not a ban. It’s an environment that makes the “wrong” too expensive.

Then he flies to Cyprus, settles in, opens a laptop—and catches the difference not in the news, but in the system’s response speed. The same tasks are done more directly. Tools are available out of the box. Integrations work like in the documentation. Nothing magical happened: it’s simply that the trust algorithms of global platforms—from cloud providers to GitHub—see in his IP address not “risk”, but “customer”. The U.S. and Western providers are generally more favorable to Cyprus than to China when it comes to admission into their perimeter: updates, services, compatibility, access to “heavy” modes turn out not to be a privilege, but a default setting.

But this is not freedom. It is another step in the same hierarchy. This approach operates exactly until the trajectory of his compute crosses the semantic red lines of that very “free” jurisdiction. As soon as his tasks—through prompts, data, AI goals—are identified by security algorithms as “problematic” for Cyprus, Andorra, or even the U.S., he will face the same friction, slowdowns, and sudden unavailability, just at a different conceptual level. The system doesn’t turn off. It adapts the level of resistance to the semantics of the action. This is how a true geographic-semantic grid of control emerges: a global optimization of compute flows, where the “allowed” runs on high-speed highways, while the “questionable”—regardless of passport—hits a curve, a “roadworks” sign, or a toll.

And here the key shift happens. He realizes that a passport is now measured not by where you can fly, but by the width and throughput of your computational corridor. He starts perceiving states as packages of access conditions to different segments of the global “shop floor”: one provides market scale and internal infrastructure, another compatibility with critically important perimeters. Just as bloggers migrate today because of taxes, tomorrow engineers and creators will migrate in search of the maximum efficiency coefficient of their mind—toward where friction is lower between thought and its materialization in code, a model, or a product. This is the geography of the switch in its pure form: freedom has become not an idea, but a connection parameter.

An inevitable phenomenon arises immediately: where there is qualification, quotas, and clearances, an underground appears. But the most important thing is not even the underground—it is the normalization of borders. You stop considering access a right. You start considering access geography and status. That’s how it is.

It is also important to say plainly one more thing. The state cannot tighten the screws to the end. A state with zero users and zero business is nonviable. An empty perimeter is not security; it is economic death. So real policy will be a compromise: give a computational minimum to the majority so the country does not fall into intellectual poverty, and at the same time keep strategic regimes behind clearances so as not to lose control. That will create a new type of competition among states: not only by taxes and social payments, but by the level of computational rights they are willing to grant in exchange for staying inside their perimeter.

And as cement over all of this works the outsourcing of responsibility. When something is turned off or becomes unavailable, you will almost never find the author of the decision:

• the platform will say “regulatory requirements”,
• the regulator will say “a private company”,
• support will say “the algorithm”,
• the algorithm will say “a profile error”.

This is how the power of the future looks civilized: not bans, but policy; not a baton, but an agreement; not ideology, but logging. There are consequences—there is no addressee.

4) Why this unified scenario is unsettling precisely because of its “normality”

Dystopia frightens because it is visible. This scenario is unpleasant because it will look like optimization.

They won’t forbid you to think—they’ll offer you a “better service”.

They won’t tell you “you are dangerous”—they’ll offer you an “additional check”.

They won’t declare inequality—they’ll show you a “tariff lineup”.

They won’t impose censorship—they’ll bill you “quotas and limits”.

And only then will it become clear that the main thing has changed: the right to act has ceased to be a natural state. It has become a combination of tariff, profile, jurisdiction, and clearance. Freedom is not abolished. It simply ceased to be available by default—and became available “under conditions”.

And the most insidious part: many will accept it voluntarily. Because inside the corridor it will be warm, fast, and convenient. That is why this unified scenario does not need evil geniuses. It only needs the fact that it is almost always easier for a person to choose comfort than autonomy—if autonomy is surrounded with friction..

12. Evolutionary Behavioral Spiral: How the Regime of Conditioned Access Captures a Person Turn by Turn

Digital sovereignty is not a wall built by a single decree. It is a procedural machine operating on the principle of a behavioral spiral. Its goal is not simply to forbid, but to govern, direct, and optimize. Each turn of the spiral is a new level of influence justified by “security” or “care”, but leading to one thing: total conditioned clearance. The key paradox: this is not isolation, but a new form of management globalization. States and corporations worldwide converge in methods, finding a common language in control protocols. Incentives align: for the state—manageability; for corporations—data monetization; for the citizen—the illusion of being protected. The outcome is the normalization of a system in which your digital life becomes a function of your loyalty.

Below is the structure of the spiral. First they let you in. Then they shape an environment around you. Then they evaluate and motivate. Then they correct. And finally they isolate those who do not integrate.

PERIMETER: THE JURISDICTIONAL FRAME

Before becoming a subject of the system, you enter its geographic and legal zone. This perimeter is the zero layer, the starting condition that determines which rules the rest of the spiral will operate under.

• China: The Great Firewall is not just a set of technologies for content filtering. It is the foundation of technological and legal sovereignty, creating an autonomous digital ecosystem with its own rules, infrastructure, and economy. By entering this perimeter, you automatically agree to all subsequent conditions.
• Russia: The “sovereign internet” law legally fixes a national digital perimeter. It allows, if necessary, disconnection from global infrastructure and centralized management of internal traffic, creating a controlled digital environment.
• European Union: The General Data Protection Regulation (GDPR) and the Digital Markets Act (DMA) set strict rules for data localization and processing. They create legal boundaries within which companies must adapt their services, de facto forming a “European” segment of the internet.
• United States: executive orders and national-security laws allow blocking cross-border data flows to “adversary countries” and regulating access to sensitive technologies. This creates zones of technological influence and limits the possibilities of global services.

Essence of the perimeter: your physical location or citizenship determines the version of digital reality you are about to be immersed in. The choice is made for you. An attempt to cross this perimeter without permission becomes, in itself, grounds for suspicion.

TURN ONE: ACCESS. They let you into the system.

For the system to interact with you, you must stop being anonymous and become a legitimate, identifiable object.

1. Layer of hardware legitimacy. Here the network starts to live as a restricted facility. The right to own a device is separated from the right to use it in public infrastructure..

• China: any telecommunications equipment must obtain a Network Access License (NAL)—permission to connect to public networks. Without it the device is blocked.
• Turkey: there is a strict rule requiring registration of a phone’s IMEI for foreigners within 120 days. Missing the deadline leads to network blocking.
• Russia: Roskomnadzor’s device registry makes it possible to block network access for devices used for “unlawful” purposes.
• EU and U.S.: Digital Markets Acts and cybersecurity rules de facto introduce device certification for access to key services.

Essence of the layer: connectivity becomes a permission, and permission becomes revocable.

2. Layer of tying hardware to a person. The device becomes your digital double.

• China: a device identifier (IMEI) is tightly linked to a citizen’s ID through state platforms, and then to WeChat or Alipay accounts.
• Russia: a system of “digital profiles” is being introduced, where IMEI, SIM card data, and migration information form a single traceable trail.
• EU: under eIDAS 2.0 regulation and the AI Act, the device is treated as a trusted element for access to public services and “risky” services.

Essence of the layer: the network starts working not with requests, but with a person.

3. Layer of network identity. The SIM card becomes a KYC key.

• China: since 2019, buying a new SIM card requires not only a passport, but mandatory facial scanning.
• Russia: 2025 laws tighten identification procedures, and a SIM card deactivation can result from “undesirable” online activity.
• EU and U.S.: under regulator pressure, operators increasingly act as verification providers for access to platforms and “age” services.

Essence of the layer: admission to the network becomes a procedure, not a purchase.

4. Layer of age segmentation. Age turns into a digital qualification.

• China: since 2021, one of the world’s strictest systems of restrictions for minors in online games has been in effect (only 1 hour on certain days).
• Australia: in December 2025, the first national law came into force obliging social networks to take “reasonable steps” to prevent children under 16 from creating accounts.
• EU and U.S.: the DSA and state laws introduce mandatory age verification for access to social networks, pornography, and generative AI.

Logic of the turn: without an identification package you are digital nothing. You have passed registration. You are not yet free, but you are already accounted for.

TURN TWO: SHAPING THE ENVIRONMENT. The system builds a manageable reality around you.

Once inside, you find yourself in an artificial landscape whose parameters are tightly controlled.

5. Layer of resource quotas (digital metabolism). Your possibilities are limited through control over basic resources.

• China: as part of the “dual carbon goals”, quotas on energy consumption are introduced for data centers and industry. Any digital project must account for its “carbon footprint”.
• Russia: in energy-deficit regions, seasonal bans on cryptocurrency mining are introduced. “Heavy” computing becomes a regulatory instrument.
• EU: the AI Act introduces mandatory energy-consumption reporting for large AI models, creating a basis for future quotas.
• U.S.: pressure on power grids from AI and data centers creates regulatory interest and infrastructure constraints through the market and local rules.

Essence of the layer: you can do almost anything, but within the metabolism. Beyond it, you pay in money and time.

6. Layer of infrastructure sovereignty. Here power does not argue with content. It controls the room, electricity, cables, routing, personnel, supply chains. This is management of concentration points.

China—critical-infrastructure regime and control of digital services through infrastructure and cyber rules.

Russia—centralized ability to manage traffic as an element of a perimeter and a sovereign segment.

EU—the NIS2 regime extends risk management and reporting requirements to digital infrastructure, including data centers.

U.S.—critical infrastructure and national-security perimeters set rules for the IT sector and suppliers.

Essence of the layer: computing ceases to be a private service. It becomes an object of resilience.

7. Cognitive filtering. The information flow as a managed service

When identity and infrastructure are already assembled, it is easy to stretch a layer over them that the user perceives as the internet. This is not a ban, but management of visibility, context, and safety.

China—regulation of synthetic content and real-name regime in services, turning moderation into an environmental standard.

Russia—infrastructure filtering, blocking, pressure on bypass channels, and criminalization of certain types of interaction with banned materials.

EU—the DSA turns platform systemic-risk management into an obligation and makes protection of minors part of service operation.

U.S.—the layer forms through platform policy, court–state conflicts, and pressure on major services, rather than through a single act.

Essence of the turn: you were not simply given access. You were issued access to a particular version of reality.

TURN THREE: EVALUATION AND MOTIVATION. The system evaluates your behavior and ties the key resource to it.

In the created environment, you are watched, scored, and the score is converted into economics.

8. Layer of social-reputational control. Once a person is known, the temptation arises to measure how convenient that person is for the system. The result is always the same: the risk profile begins to regulate access.

China—social credit practices, where “unreliability” is expressed in concrete restrictions on actions, including buying train and plane tickets.

Russia—lists of extremists and terrorists maintained by the financial regulator make it possible to freeze accounts and restrict financial access.

EU—direct social credit is not implemented, but risk scoring and compliance in platforms and the financial sector form a soft regime of functional restrictions.

U.S.—a developed system of credit and insurance scoring, plus sanctions and perimeter regimes, forms a pragmatic analogue of trustworthiness.

Essence of the layer: the presumption of innocence is replaced by the presumption of manageability. You are restricted by profile, not by misconduct.

9. Layer of financial synthesis (CBDC). Money becomes a programmable instrument of control.

• China: the digital yuan (e-CNY) is a fully centralized system in which the central bank can track each transaction in real time and, in theory, program conditions of its use (for example, prohibit a transfer to a VPN service account).
• Russia: pilot projects of the digital ruble include smart-contract functionality for targeted financing, opening the path to pinpoint restrictions.
• EU and U.S.: while the design of the digital euro and dollar emphasizes privacy, the CBDC architecture itself gives central banks unprecedented tools to control money flows.

Logic of the turn: this is the heart of the system. Your behavior (layer 8) is directly converted into economic possibilities (layer 9). Loyalty becomes capital; dissent becomes a financial pit.

TURN FOUR: CORRECTION. If the score is low, the system applies targeted mechanisms of coercion.

The system does not tolerate imbalance. A low rating or deviant behavior triggers mechanisms of “soft” coercion.

10. Latency control (the power of delay). Censorship through discomfort.

• China—external border and perimeter regimes allow the quality of access to be managed as an infrastructure norm.
• Russia—centralized traffic management creates the possibility of degrading access through infrastructure.
• EU—net neutrality limits direct arbitrary action, but QoS and prioritization always exist; the question is rights and grounds.
• U.S.—degradation is more often implemented through product rules, sanctions restrictions, and equipment infrastructure policy.

Essence of the layer: they do not forbid your action. They make it slow and unprofitable.

.

11. Algorithmic erasure (Data Hygiene). If information is not visible in search and not recognized by assistants, for most people it ceases to exist. This is soft erasure without physical deletion.

• China—the regime of content and identity control creates conditions where “outside the perimeter” becomes a toxic source by default.
• Russia—network segmentation and blocking form a reality in which part of the past does not physically vanish, but vanishes from mass access.
• EU—requirements for risk management and data quality create incentives for strict source hygiene at the input to algorithms.
• U.S.—practices of private platforms and search engines create a similar effect through internal policies and trust ratings.

Essence of the layer: the past becomes conditional. History exists only where algorithms see it.

12. “Autonomy tax”. The most stable mechanism is to make the alternative not forbidden, but expensive. It’s not that autonomy is impossible. Autonomy becomes a luxury.

• China—standards, energy efficiency, and infrastructure requirements raise the cost of autonomous perimeters.
• Russia—energy regimes and restrictions make part of autonomous computing economically toxic.
• EU—reporting and infrastructure requirements raise the price of owning autonomy through regulatory costs.
• U.S.—the cost of energy, connectivity, and compliance with infrastructure requirements raises the entry bar to autonomy through the market and risk perimeters.

Logic of the turn: this is not a crude punishment, but course correction. The system makes life unbearably inconvenient and expensive exactly until you return to the “correct” channel.

ISOLATION. If you are not in the system, you are outside its boundaries.

The circle closes. Whoever did not pass clearance or was pushed out by correction mechanisms ends up outside digital agency.

• China—the perimeter and real-name policy set a regime where outside clearance you are not “free,” you are simply not serviced.
• Russia—financial and infrastructure restrictions make leaving the perimeter practically equivalent to leaving the legal space.
• EU—refusal of identity and age attributes gradually means refusal of services and functions.
• U.S.—outside compliance and payment perimeters you are not forbidden; you simply do not have access to resources and markets.

Essence: outside the system there is no economic and social agency in the modern world. This is the final point of the spiral.

SYNTHESIS: THE BEHAVIOR MACHINE

This spiral is not a conspiracy, but the result of a structural convergence of interests. It is a technocratic evolution in which each next turn logically grows out of the previous one, creating a self-supporting machine.

Its operating principle:

1. Identify (Turn 1: Access).
2. Place into a manageable environment (Turn 2: Shaping reality).
3. Evaluate behavior within it (Turn 3: Evaluation and motivation).
4. Correct deviations through economics and comfort (Turn 4: Correction).
5. Ignore what cannot be corrected (Isolation).

The danger of this architecture is its anti-revolutionary nature and adaptability. You cannot break it by tearing out one “layer”. It is dynamic and total. It does not suppress will; it offers a “comfortable” path, making the alternative unbearably expensive and quietly erasing the very memory of it.

To realize yourself inside this spiral is to stop being its passive object. Strategic behavior, digital hygiene, understanding the price of each action, and the ability to live so that autonomy does not disappear quietly under the guise of convenience become a new form of literacy. Living in this spiral means conducting continuous negotiations with the machine, where the price of a mistake is measured not by a fine, but by the gradual erasure of you from digital reality.

What to do: strategy instead of panic

The first step is to stop looking for a conspiracy and start seeing the evolutionary logic. No director is needed here. It is enough that incentives align: corporations benefit from iterative monopoly and rent for access, states from manageability and security, users from comfort, engineers from standards and risk reduction. Such convergence gives birth to a system that cements itself: each next turn looks reasonable and useful, but narrows the space of autonomy.

That is why it is more dangerous than a conspiracy. A conspiracy can be exposed and broken. Evolution cannot. It can only be recognized, and then you start living in it strategically. Your freedom will not be forbidden by decree. It will become unavailable without the right tariff, profile, and perimeter. And many will like it, because it will be convenient. Your task is not to defeat the system head-on, but to preserve agency within it and always have a working exit when access conditions become unacceptable.

Personal strategy: from user to subject

The goal is to stop being a person who is serviced under someone else’s rules and become a person who manages their digital capital: data, skills, relationships, and reputation.

1. Reduce systemic vulnerability

• Don’t put your life into one perimeter: one messenger, one email, one cloud drive, one compute provider, one payment rail. Any monoculture turns you into an object of the switch.
• Keep critical data in portable form: open formats, regular exports, backups outside the main perimeter, a clear storage structure. You must be able to leave without losing memory and work.
• Treat digital reputation as clearance. This is not about social networks. It’s about what profile you will look like to the trustworthiness gate tomorrow: to a bank, a platform, a cloud, an employer, the state. The discipline is simple: less toxic trace, more predictability, fewer reasons for automatic friction.

2. Invest in inalienable competencies

There are skills that do not turn off with the service. Those are the ones to strengthen.

• Systems thinking: the ability to see the rules of the environment, understand how clearance is built, where the switch is, and which turns of the spiral are triggered next.
• Management and negotiation: the ability to negotiate with people and structures, rather than beg an interface for access.
• The ability to build coalitions and trust outside platforms: living relationships, professional networks, reputation in an environment not owned by a single provider.
• Regulatory literacy: understanding the basic regimes by which data and platforms live. Not for submission, but for protection and risk calculation. Ignorance of regulatory logic in the new era is equivalent to illiteracy.

Business strategy: from IT expenses to existential risk

Dependence on a single compute provider or a single jurisdiction is a risk on the level of energy supply. If it cannot be replaced quickly, then the business does not control its future.

1. Diversification as a basic discipline

• Technological: a multi-cloud or hybrid approach where key processes are portable rather than tied to one stack.
• Jurisdictional: pre-thought-out options for hosting, data processing, payment rails, and contracts in different legal regimes. Not for flight, but for negotiating power.
• Operational: critical knowledge and access must not live in one person’s head and in one provider’s console. Everything must be reproducible.

2. Compliance and security at the core

• Compliance by default is not bureaucracy; it is an entry ticket. In an access regime, failure to meet standards means not a fine, but denial of access.
• Security must be built into development and operations processes. Not as a department that comes at the end, but as part of the architecture.
• You need degradation scenarios: what do we do if a key service slows down, cuts limits, requires additional checks, or becomes unavailable in your jurisdiction. Scenarios must not be on paper; they must be rehearsed.

3. Financial realism

Diversification costs money, but mono-dependence costs a business. Plan on a 1–3 year horizon and keep three scenarios: optimistic, base, stress. Account for hidden costs: lawyers, security, team training, data migration, process adaptation. In an access regime, naive economizing turns into an autonomy tax.

Conclusion: three laws of the digital spiral

• Law of reversibility: if your right can be technically revoked under rules you do not control, then it is not a right—it is a rental.
• Law of convergence: state regulation, corporate policies, and technical standards converge into a single access regime. Fighting one layer is meaningless if you do not see the whole construction.
• Law of sovereignty: the highest value is an autonomous core. Data, skills, relationships, and the ability to act without a single, one-and-only perimeter. Not for war with the system, but for negotiating with it.

Strategy instead of panic is not a plan for victory. It is a plan for preserving agency. You cannot turn the spiral off. But you can see it, stay ahead of it, and not give away your freedom of action by default in exchange for convenience.