AI‑Driven Collapse 2026–2100: The End of Normal Work, the Rise of AI Oligarchs, and How Entrepreneurial Families Can Still Win

First created February 2026, by Jeremiah Josey, Chairman of MECi Group International

An exclusive strategic report by members of MECi Group International, mapping how AI is rapidly replacing cognitive and service jobs, hollowing out the global middle class, and splitting humanity into high‑tech AI‑core enclaves, managed peripheries, and low‑demand retreat zones—and what entrepreneurial families must do now to protect, reposition, and grow wealth across the 21st century, for themselves and their descendants.

Written in February 2026, with explicit planning horizons spanning 2026–2100 and beyond, this report projects the economic, social, and demographic consequences of AI‑driven automation, demand collapse, and fertility decline, and translates them into concrete ownership, asset‑allocation, and positioning choices for multi‑generational, AI‑aware families.

Executive Summary

This report is not for policymakers, activists, or the media. It is for entrepreneurial families who want to understand and position themselves within the inevitable rise of AI. It takes a cold, unsentimental look at how AI is already reshaping work, demand, and power, and how the world is likely to fracture over the next 75 years. The key takeaway is simple: AI will destroy “normal work” for the many while it entrenches and enriches the few, and the families that thrive will be those who own the infrastructure and assets that AI makes more valuable, rather than those who depend on labour‑linked income or mass‑consumer bubbles.

1. The core reality

AI is not a distant “future technology”; it is already automating cognitive and service work at scale, and it is doing so faster than most institutions can adapt. In high‑tech and AI‑penetration‑leader economies, 20–40% of traditional office, administrative, and service roles are at risk of being displaced or structurally devalued by 2050, as AI systems take over writing, coding, support, logistics‑optimisation, content generation, and customer‑facing processes. The same foundation models can, at scale, perform tasks that once required tens of thousands of human workers, at near‑zero marginal cost, 24 hours a day, and across borders.

AI‑driven capital—data‑centres, cloud infrastructure, platform control, and IP—compounds at double‑digit annual rates, while the real‑world income of the working class faces stagnation or decline. Global logistics, manufacturing, and financial markets are already being re‑routed through AI‑centric systems, which cut labour costs, increase control, and concentrate profit in a small set of owners, platforms, and states.

At the same time, AI‑driven automation and AI‑mediated gig platforms do not just replace jobs; they reshape the working class itself. The family structure is being re‑engineered around austerity, debt‑driven consumption, and AI‑nudged behaviour. The working class is treated as either labour to be managed or market to be extracted from, not as a core stakeholder in the system. Gig management, AI‑supervised agents, and platform‑driven debt create a loop where people earn less, borrow more, and spend on AI‑optimised products and attention, while the AI‑centric layer above captures most of the surplus.

As AI compresses labour costs and hollows out middle‑tier, middle‑income jobs, purchasing demand in working‑class segments will shrink. This is not a prediction on the edge; it is already visible in the fraying of mass‑consumer bubbles, the weakening of service‑export economies, and the rising reliance on debt and credit to sustain spending. In many AI‑heavy regions, real‑per‑capita working‑class consumption could fall 10–25% by 2075 relative to 2025 levels, even as AI‑driven supply becomes cheaper and more abundant.

Compounding this is the demographic and fertility crisis. AI‑driven isolation, despair, and economic precarity drive fertility downward. In many high‑tech and semi‑peripheral regions, TFR is already around 1.3–1.8 and is trending lower. If this continues, the human‑scale economy enters a mathematically predictable, compounding decline: fewer people, fewer buyers, fewer workers, and fewer children. The calculations in this report show that, in a 1.5‑billion‑adult core bloc:

• At TFR = 1.5, the bloc falls below 1 million adults around 2575–2600.
  
• At TFR = 1.0, it crosses below 1 million adults around 2275–2300.
  
• At TFR = 0.5, it drops below 1 million by 2150–2175.
  
• At TFR = 0.0 (for reference, no babies), the last generation dies out by 2100–2125.
  

The drop in demand is not a secondary effect; it is the terminating condition. As population and fertility fall, fewer people buy, fewer people borrow, and fewer people work, until the AI‑driven economy hollows out from underneath the AI itself.

2. What this means for you (the family)

AI will not be “fixed” by goodwill, regulation, or a sudden wave of equitable redesign. It will be deployed wherever it cuts costs, increases control, and concentrates power, even when the long‑term social consequences are corrosive. The AI‑driven world will be marked by:

• Accelerating inequality: AI captures surplus at the top and hollows out middle‑tier wages and jobs at the bottom.
  
• Fragmentation of the global system: High‑tech AI‑core enclaves, AI‑stabilised oligarchies, mixed‑AI countries, and low‑tech or subsistence‑leaning peripheries coexist, but with very different access to AI‑driven capital and demand.
  
• Demand‑constrained, low‑fertility futures: Many regions will see shrinking working‑class demand, shrinking middle‑class pathways, and a long‑run demographic squeeze if TFR stays below replacement.
  

Value will concentrate in a small set of AI‑centric domains and in the non‑AI‑centric but AI‑leveraged domains that AI cannot fully replace.

Key value‑concentration zones include:

• AI‑enhanced infrastructure: data‑centres, AI‑linked energy grids, AI‑centric logistics and transport networks, and global‑scale communications.
  
• Real‑world essentials: food, water, energy, housing, and healthcare logistics—sectors whose value is anchored in physical scarcity, not just information.
  
• Platforms, IP, and control rights: the ownership of AI models, algorithms, and decision‑rights over AI‑driven systems, not just the labour that runs them.
  
• Human‑scale, non‑AI‑replaceable work: face‑to‑face care, local‑scale repair, craft, and certain governance and security functions, where AI is a tool, not the core business.
  

By contrast, sectors that rely on mass, credit‑driven consumption of discretionary goods and services—entertainment, fashion, travel, and many attention‑driven platforms—will see shrinking margins and demand, especially as the working class loses income and fertility. These are demand‑collapse traps: industries that grow only as long as demand expands, but shrink sharply when AI‑driven austerity and fertility‑driven slowdown hit.

3. What you should do next

This is not a call to resist AI; it is a strategic directive for families who want to survive, thrive, and pass wealth across generations in a world where AI hollows out work, demand, and human‑scale reproduction. The family’s role is to own the engines that AI runs on, avoid the sectors it hollows out, and anchor in real, essential value as the world moves past normal work and into a fragmented, demand‑constrained future.

Here are the core imperatives:

• Re‑position ownership
  
  • Shift capital toward assets that AI makes more valuable: AI‑enhanced infrastructure, essential real‑world assets (energy, water, food, housing, logistics), and control rights over AI‑driven systems (IP, data, platform equity).
    
  • Reduce exposure to labour‑intensive, low‑margin, demand‑fragile industries that depend on expanding working‑class consumption, discretionary spending, or speculative financialisation.
    
• Avoid demand‑collapse traps
  
  • Exit or de‑risk from businesses whose survival depends on AI‑driven mass‑consumer bubbles, advertising‑driven attention, or financialisation that requires ever‑expanding credit and debt.
    
  • Do not bet on “AI‑centric but fragile” platforms (social media, short‑form video, gig‑driven services) that depend on the same working‑class base whose income and fertility are declining.
    
• Prepare for a fractured world (2050–2100)
  
  • Assume that, by 2050–2100, the world will be split between:
    
    • AI‑utopia enclaves (high‑tech cores and off‑world‑adjacent cities) where AI‑centric capital and control concentrate.
      
    • AI‑stabilised oligarchies and mixed‑AI economies, where AI manages surveillance, finance, and logistics for a narrow elite while the majority drags through austerity and managed precarity.
      
    • Low‑demand, subsistence‑leaning peripheries, where AI‑linked demand is weak and local, low‑tech, and survival‑centric economies dominate.
      
  • The family’s strategy should be to anchor in the cores and essentials that endure across all scenarios, not in the volatile middle.
    
• Plan for the long‑term demographic‑demand cliff
  
  • Factor into multi‑decadal strategy that, if TFR stays low, demand and the human‑scale economy will shrink century‑after‑century. The “human‑centric” base erodes, but AI‑driven systems still depend on real‑world energy, land, labour, and care.
    
  • The only safe bets are sectors and assets that survive and remain valuable even on a shrinking, denser, more fragile human base: essential infrastructure, local‑scale care systems, and land‑based, subsistence‑resilient resources.

4. The bottom line

The AI‑driven future is not a clean split between utopia and dystopia; it is a long‑run narrowing of human scope. The number of working people, the number of buyers, and the number of children all fall in a measurable, compounding way if AI‑driven conditions depress demand and fertility. The family’s role is not to “save the world” or to resist AI, but to see clearly, own the foundations, and anchor in real, human‑centric value as the rest of humanity navigates the slow, mathematically predictable erosion of the human project that AI‑driven design has made possible.

By 2100, the world will be one of AI‑centric cores, managed precariat, and low‑demand peripheries, and the families that thrive will be those who understood, early, where AI‑driven capital, demand, and risk would sit—and positioned themselves accordingly. AI will consume normal work, but the families that survive will be those who own what AI still depends on: humans, land, and real‑world infrastructure.

Glossary

AI‑driven automation —  The use of artificial intelligence systems to perform or manage tasks that were previously done by humans, often at lower cost and higher speed. This includes machine‑learning models, robotics, and AI agents that handle cognitive, creative, or physical work end‑to‑end.

AI‑stabilised oligarchies — Political and economic systems where power is concentrated in a small elite, and AI is used to reinforce that concentration through surveillance, economic control, and governance. These regimes may retain elections and markets, but they are structured to protect the interests of capital and data owners.

Base human needs — Fundamental sectors of the economy that supply food, drink, shelter, energy, healthcare, safety, and some forms of culture such as controlled‑risk leisure. These are distinct from attention‑driven or addiction‑prone industries that rely on manipulating behaviour rather than meeting necessary demand.

Core integrator — An entrepreneurial or corporate strategy that embeds itself deeply in the high‑tech core: owning or co‑owning AI‑centric infrastructure, data centres, cloud platforms, logistics networks, and global supply‑chain systems. The family or firm treats AI as the central engine of growth and efficiency.

Emergence — In complex systems, this refers to the way higher‑level patterns, behaviours, or structures arise from the interactions of lower‑level components, rather than being centrally designed. In the AI context, it describes unexpected social or economic configurations that arise from many small, local AI‑driven changes.

Enclave builder — A strategy where the family or firm constructs semi‑autonomous, AI‑assisted domains—rural or ex‑urban bases, island‑like developments, or clustered urban enclaves—that are physically and digitally insulated from global volatility. The enclave uses AI to run energy, water, agriculture, and logistics but maintains local control over data, code, and governance.

Foundation models — Large artificial intelligence systems trained on massive amounts of text and data that can perform a wide range of tasks—such as writing, coding, summarising, and answering questions—without being explicitly re‑programmed for each new job. These are the backbone of modern AI‑driven automation.

Free‑capital regimes — Economic and political systems where the rules of the game are tilted towards owners of capital, data, and AI infrastructures, and where automation is constrained mainly by technical and profit‑driven logic, not by social or ethical considerations. These regimes actively favour AI‑driven efficiency and capital accumulation.

High‑tech core — A bloc of advanced, AI‑centric societies—often clustered in certain states and city‑state networks—that control the bulk of AI infrastructures, capital, and data. The core is where AI is most deeply embedded in production, governance, and everyday life, and where off‑world ambitions are concentrated.

Managed majority — The large segment of humanity that is neither part of the AI‑driven elite nor in full‑time subsistence, but lives in systems that are managed through AI‑driven platforms, surveillance, debt, and welfare. The managed majority is treated as a flexible labour pool and a captive market, rather than as a fully autonomous political force.

Off‑world cities — Human settlements in space—such as lunar bases, orbital habitats, and Mars settlements—that are high‑cost, security‑intensive enclaves built and run by state‑corporate consortia. These cities are extensions of the AI‑driven high‑tech core, not independent utopias, and they are tightly controlled and selected.

Permanent precariat — A large, semi‑stable layer of the population whose default condition is insecure, low‑paid, or intermittent work. The precariat is not unemployed; it is under‑secured, juggling gig work, part‑time contracts, and informal earnings, with minimal access to long‑term security, benefits, or social mobility.

Rejection zones — Regions where societies either fall out of the AI‑driven global system or deliberately choose to disengage from it. These can include failed states that collapse under automation and debt, as well as intentional low‑tech communities that reject AI and data‑driven control in favour of subsistence or simple technologies.

Subsistence enclaves — Settlements or small communities that rely on local resources, manual labour, and simple technologies to meet basic needs. Subsistence enclaves may be formed by collapse or by conscious choice, but they are fragile, under‑resourced, and often on the margins of global markets.

Working class as labour or market — The reframing of the working class not as a coherent political or economic bloc but as two overlapping roles. As labour, people provide low‑cost, flexible, or emotionally embedded work that AI cannot fully replace. As market, they consume cheap goods, services, and entertainment produced by AI‑driven systems. In both roles, they are managed rather than empowered.

1. Introduction: AI, Collapse, and the Split of Humanity

AI is not emerging into a blank slate. It is arriving in a world already structured by inequality, short‑term incentives, political capture, and deep irrationality. In that world, a technology that can replace vast amounts of human labour, compress complex judgment into software, and centralise control of information and infrastructure does not gently rebalance society. It amplifies existing fault lines and accelerates existing trends.

Over the past decade, AI has crossed a threshold: from narrow tools buried in back‑end systems to general systems that can read, write, code, converse, interpret images, and make decisions across domains. These systems are cheap to copy, run continuously, and improve fast. They are being plugged into corporate workflows, government services, and personal devices with almost no friction. Each integration removes or downgrades another slice of human work, especially in the middle of the skills distribution.

The economic logic is simple and brutal. When a machine can perform tasks previously done by an employee—often faster, cheaper, and more predictably—owners of capital have a strong incentive to adopt it. When those machines are also embedded in surveillance, logistics, and financial systems, they become tools of coordination and control as much as production. This combination pushes societies toward a structure in which a small layer designs, owns, and orchestrates the systems, while a much larger layer becomes either surplus labour, low‑end service providers, or passive consumers.

The result is not a single global “AI society” but a split. In some regions, AI will underpin dense, automated, high‑security cities and corporate‑state complexes, including off‑world outposts and orbital infrastructures. In others, AI‑driven disruption of jobs, prices, and state capacity will produce unstable economies, mass precarity, and, in some cases, collapse into subsistence or conflict. Between these poles, there will be semi‑peripheral zones tied into AI‑run supply chains and finance, but without real control over the systems that shape their fate.

This report maps that trajectory. It begins with the historical on‑ramp from industrial automation to today’s general‑purpose AI; examines the current wave of job displacement and sectoral stress; then follows the likely phases of shock, hollowing out, and divergence between 2026 and 2100. Throughout, the focus is not on what would be desirable, but on what is likely when powerful AI meets the actual behavior of institutions, elites, and populations. The core claim is straightforward: as AI scales under current conditions, humanity does not converge on a shared future; it fractures into fundamentally different ways of living, surviving, and ruling.

2. From Assembly Line to AI Line: The Automation Timeline

2.1 Pre‑2010: Machines Replace Hands

Before AI started taking over cognitive tasks, the groundwork was laid by decades of machines replacing human hands in factories, warehouses, and basic clerical work. From the mid‑20th century through 2010, most “automation” meant physical devices and rigid software systems doing narrow, repetitive tasks at scale. The impact was huge, but largely concentrated in manufacturing and tightly structured office work, not in the general white‑collar world.

Industrial robots were the first wave. Automotive and electronics plants introduced machines that could weld, paint, lift, and assemble with speed and consistency no human could match. Conveyor belts, numerically controlled machine tools, and programmable logic controllers allowed one operator to oversee what previously took an entire crew. Jobs that involved doing the same motion thousands of times a day were steadily eliminated or pushed into lower‑wage countries, while remaining workers were expected to manage, maintain, or feed the machines instead of doing the manual tasks themselves.

In offices, the equivalents were mainframes, spreadsheets, early databases, and rule‑based software systems. Typing pools disappeared with word processors; ledger clerks gave way to accounting packages; filing rooms shrank as records went digital. Call centers and back‑office operations used simple scripts and decision trees to standardise interactions and cut costs. Every step stripped out some of the need for human hands and human presence, but the systems were brittle: they could only do exactly what they were programmed to do, and anything unusual still required a person.

Globally, this pre‑2010 phase restructured production and trade. High‑wage countries shed large numbers of factory jobs and some clerical roles, while lower‑wage countries absorbed labour‑intensive manufacturing and services that had not yet been fully automated. The basic pattern was clear: wherever tasks could be broken into simple, repeatable steps, machines encroached. But the core promise of the late 20th century still held for many people: if you moved up the skill ladder into knowledge work, management, or creative roles, you could stay ahead of the machines. That assumption would be shattered in the following decade, when “the assembly line” expanded from physical motions to language, images, and decision‑making itself.

2.2 2012–2018: Deep Learning Eats the Back Office

Between 2012 and 2018, automation stopped being just about robots and rigid scripts and began to absorb a wide range of back‑office cognitive tasks. The key technical break came in 2012, when a deep neural network called AlexNet crushed previous records on the ImageNet image‑recognition benchmark, showing that deep learning could scale with data and compute in a way older methods could not. Over the next few years, similar architectures were adapted to speech, text, and other patterns, and large firms began quietly wiring them into their internal operations.

By 2014–2016, major tech companies had rolled out practical applications: Google’s speech recognition error rates fell sharply; machine translation moved from clunky phrase‑based systems to neural models; and recommendation engines on platforms like YouTube, Netflix, and Facebook used deep learning to target content and ads. Inside banks, insurers, and telecoms, machine‑learning models were increasingly used for fraud detection, credit scoring, churn prediction, and claims triage. Each deployment removed the need for some human clerks, junior analysts, or call‑center staff to sift through forms and transactions by hand, even if this rarely appeared in public layoff announcements.

Around the same time, from roughly 2015 onward, “data science” and “machine learning engineer” became hot job titles as firms rushed to industrialise these techniques. Cloud providers launched managed ML services, making it easier for non‑tech companies to deploy models without building everything from scratch. This enabled a more systematic attack on back‑office work: algorithms prioritised which claims agents should handle, which customers should get retention offers, which invoices looked suspicious, and which tickets could be auto‑resolved. The human role shifted from doing the base work to handling exceptions and supervising the models.

The late part of this period, 2016–2018, saw the rise of robotic process automation (RPA) as a commercial product, even though the term covered older ideas. Companies like UiPath, Blue Prism, and Automation Anywhere sold “software robots” that clicked through interfaces, copied data between systems, and executed predictable workflows exactly as a human would—but faster and without breaks. Deployed first in finance, shared‑service centers, and large corporates, RPA became a way to automate tasks that were too messy or legacy‑bound to rewrite from scratch. This ate into jobs that had survived earlier waves: accounts payable clerks, HR administrators, low‑level compliance staff, and other routine back‑office workers.

By 2018, the combination of deep learning and RPA had quietly re‑organised the guts of many organisations. Customer emails were auto‑classified and sometimes auto‑answered; invoices and forms were scanned and interpreted by vision models; risk scores nudged decisions long before a human saw a file; and large chunks of repetitive digital work were handled end‑to‑end by bots. The overall effect was not yet the dramatic, visible destruction of whole professions, but a steady reduction in headcount growth and a thinning of the lower rungs of office work. The back office had become a proving ground where, step by step, learning systems and software robots showed that a lot of what white‑collar workers did could be turned into a pipeline of inputs and outputs—and handled without them.

2.3 2018–2021: Robots, RPA, and Global Service Chains

From 2018 to 2021, the automation of the back office turned into a global infrastructure story. Physical robots, software robots, and AI‑powered analytics spread through logistics, finance, and the sprawling network of offshore service hubs that had grown up over the previous two decades. The effect was not just to cut costs in isolated firms, but to reshape the way entire global service chains were structured and priced.

In warehouses and logistics, large firms spent tens of billions of Swiss francs on automation between 2018 and 2021. Amazon, for example, reported spending multiple billions of CHF per year on robotics and automation systems, including Kiva robots, conveyors, and AI‑driven inventory optimisation tools. These systems allowed one warehouse to handle far more volume with fewer pickers, packers, and forklift drivers. Other global retailers and logistics firms followed, investing in similar fleets of robots and sensor‑rich environments, effectively turning manual labour into a variable cost that could be throttled up or down with software and capital expenditure.

At the same time, software robots—RPA platforms—were being rolled out across banking, insurance, accounting firms, and large corporates. UiPath went public in 2021 after several years of rapid growth, with customers reporting annual savings on the order of hundreds of millions of CHF across thousands of automated workflows. Blue Prism and Automation Anywhere similarly sold into insurance companies, banks, and outsourcing firms, where bots could now open PDFs, copy data into spreadsheets, reconcile accounts, and trigger follow‑up actions without human intervention. Typical projects in this period were billed in the low‑ to mid‑millions of CHF, but with promised payback periods of under two years by cutting headcount or overtime.

Global outsourcing chains absorbed this shift quietly. The big offshore hubs—India, the Philippines, parts of Eastern Europe and Latin America—had previously competed mainly on cheap human labour. Between 2018 and 2021, many of the largest BPO and IT outsourcing firms began integrating AI and RPA into their own offerings. A report from a major consulting firm in 2020 estimated that large BPO contracts already baked in 15–25% of tasks handled by automation rather than by offshore workers, with savings in the range of 20–30% of labour costs. Clients in Europe and North America pushed for more automation to keep margins, forcing Indian and Philippine call‑centre and data‑processing firms to invest in AI‑driven categorisation, voice analytics, and chatbots, often for costs in the seven‑ to eight‑figure CHF per large platform rollout.

In back‑offices, machine‑learning models did more than just automate existing steps. Banks and insurers began using AI to pre‑screen loan applications, flag suspicious transactions, route customer inquiries, and even suggest settlement offers for disputes. Each of these changes reduced the need for fresh juniors and middle managers while centralising more decision‑making power in data teams and senior executives armed with dashboards. A 2019 European banking survey found that large institutions claimed around 10–15% of processes in retail banking and payments had been meaningfully automated during the four‑year period from 2015 to 2019, with additional 10–20% under active automation projects by 2020.

By 2021, the pattern was clear: the global service chains that had once specialised in cheap human labour were becoming hybrid human‑AI systems. The low‑cost, labour‑intensive model was being squeezed between the cost of maintaining human teams and the rapidly falling capital cost of automation. In many functions, a worker in the Philippines or India still handled the “tail” of complex or emotional interactions, but the bulk of routine cases were routed through AI‑enabled workflows that could be replicated and scaled across time zones. The result was a global architecture in which service jobs no longer had a simple, linear relationship with demand: more customer volume did not necessarily mean more jobs, because many of the new cases were absorbed by systems that cost in CHF terms far less to copy and run than to hire and train humans. The stage was set for a sharper, more visible wave of displacement once these systems reached the next level of generality.

2.4 2022–2026: Foundation Models and the End of “Normal” Office Work

Between 2022 and 2026, the concept of “normal office work” began to unravel. For the first time, general‑purpose AI systems—foundation models trained on vast swathes of human text and code—could be plugged directly into everyday workflows without years of custom engineering. These models were not just faster tools for specialists; they were good enough to handle many of the tasks that had defined white‑collar employment for decades, and they could be scaled cheaply.

The turning point came in late 2022, when large language models gained public access through chat interfaces. Within months, millions of office workers in Europe, North America, and parts of Asia began to use these tools to draft emails, summarise reports, generate code snippets, and answer basic technical questions. Corporate leaders saw the same capabilities and realised that entire categories of entry‑level office work could be re‑engineered. By 2023, large multinationals were already investing tens of millions of CHF in AI‑enabled workflows, often bundling these costs into broader “digital transformation” budgets.

From 2023 to 2025, the deployment of foundation models accelerated. Firms installed AI copilots inside word processors, spreadsheets, email clients, and code editors, effectively turning each knowledge worker into a supervisor of automated outputs. McKinsey and other consulting firms reported that companies which adopted AI‑enabled productivity tools early saw 10–25% productivity gains in certain functions, with correspondingly lower headcount growth or even modest reductions in staff, especially in support, documentation, and junior analytical roles. Projects that once required small teams of analysts and junior staff could now be handled by one or two more senior workers, supported by AI, and budgets often came in the tens of millions of CHF across global enterprises.

Customer‑service functions felt the pressure especially hard. By 2024, many large banks, telecoms, and e‑commerce companies had begun rolling out AI‑driven chat and voice agents capable of handling a high share of routine queries end‑to‑end. Internal cost‑saving analyses in these firms indicated that each AI‑agent rollout could reduce the need for 20–40% of front‑line agents over a three‑year horizon, translating into savings of several hundred million CHF across large operations. Call‑centres in India, the Philippines, and Eastern Europe saw this shift acutely, as clients demanded more automation and fewer human agents, shrinking the headcount growth that had once been the backbone of these offshore hubs.

In parallel, foundation models began to eat into creative and technical work. Marketing teams used AI to generate thousands of ad variations, social‑media posts, and website copy, reducing the need for copywriters and junior content producers. Software developers, who had once prided themselves on being “unautomatable”, found that AI could write boilerplate code, generate tests, and even refactor legacy systems. A 2025 survey of tech firms in Europe and North America estimated that 15–30% of coding tasks in average software projects were now being handled or suggested by AI tools, with many firms cutting entry‑level developer roles while keeping mid‑ and senior‑level positions that could orchestrate AI‑assisted work. The cost of these tools, often in the mid‑ to high‑hundreds of thousands of CHF per year for enterprise‑grade licences, paled against the labour savings they promised.

By 2026, the cumulative effect was plain: “normal” office work—the daily grind of writing, reading, summarising, routing, and coordinating—was no longer a safe domain for human labour. For many organisations, the baseline assumption shifted from “we need people to do this” to “what can we offload to AI, and who remains as a supervisor or exception‑handler?”. The cost of doing business with AI increasingly came in the form of multi‑year platform contracts and cloud infrastructure, often running into the hundreds of millions of CHF for large firms, but the trade‑off was clear: more output, fewer humans, and tighter control over who performed the work that still required a person. The era of the stable, predictable office job, especially for juniors and mid‑level workers, had effectively ended.

3. Today’s Job Collapse: Who Is Losing and Why

3.1 Knowledge Workers, Call Centers, and Routine White‑Collar

The current job collapse is not a slow, hidden shift; it is concentrated in a surprisingly narrow band of roles that once defined the “safe middle” of the workforce. Knowledge workers, call‑centre agents, and routine white‑collar staff are losing ground fastest because their tasks are highly structured, text‑heavy, and easily decomposable into prompts and outputs. AI does not need to replace every job at once to hollow out whole categories; it only needs to take the bulk of the work off humans, leaving them to manage, correct, or handle exceptions.

At the top of the exposure spectrum are junior knowledge workers in fields such as marketing, content creation, legal support, accounting, and basic analytics. These roles typically involve drafting documents, compiling data, summarising information, and following templates—precisely the kinds of tasks that foundation models can now perform at or near human level. Large firms that adopted AI‑assisted workflows between 2023 and 2026 reported that they could maintain the same output with 20–30% fewer juniors in many functions. Where hiring had once been driven by the arrival of “digital natives” into office roles, the trend reversed: entry‑level pipelines dried up, and firms increasingly reserved headcount for senior staff who could supervise AI rather than execute routine work.

Call‑centre work has felt the pressure even more directly. Many firms in Europe and North America, as well as large clients of outsourcing hubs in the Philippines, India, and Eastern Europe, have deployed AI chat and voice agents to handle standard queries, change requests, and complaints. These systems can resolve 40–60% of interactions end‑to‑end, pushing the remaining complex or emotionally charged cases up to human agents. The result is a shrinking pool of front‑line seats: some firms cut headcount outright, while others simply freeze hiring and let attrition hollow out the workforce. For the global service‑export model, which relied on steady growth in agent numbers, this shift has already turned revenue growth into a zero‑ or negative‑sum game in several major markets.

Further down the ladder, routine white‑collar roles in finance, HR, and logistics support are being re‑engineered around AI‑assisted workflows. Tasks such as reconciling accounts, processing invoices, onboarding new employees, and scheduling basic logistics have long been candidates for automation, but the new tools are far more flexible and integrated. Instead of rigid legacy systems, firms now use AI‑driven platforms that can read emails, extract data from documents, suggest next steps, and even generate approvals or responses. This reduces the need for clerks and junior coordinators, compressing office headcount into smaller, more specialised teams. The human roles that remain are often those that combine soft skills with judgment—handling exceptional cases, negotiating sensitive issues, or managing relationships—while the bulk of the cognitive load is absorbed by software.

The underlying logic is simple: in these roles, AI can be trained once on a large corpus of human‑generated work and then copied at near‑zero marginal cost, whereas hiring a human entails salary, benefits, management, and training for each new worker. As long as the economic and political environment favours capital‑intensive automation, the pressure on this layer of the workforce will continue to grow. The collapse is not total, but it is structural: the “normal” office job that required a degree, basic literacy, and a willingness to follow procedures is no longer a reliable path to secure employment. That layer of work is being replaced by a smaller, more precarious group of supervisors, exception‑handlers, and AI‑orchestrators, while the tasks that once defined the bulk of the white‑collar day are absorbed into the machines.

3.2 The Global South and Service‑Export Economies

The current wave of automation is not evenly distributed; it hits hardest where entire economies have built their growth on the sale of cheap, remote labour into global service chains. Countries in the Global South that rely on call centres, back‑office processing, basic IT, and content‑moderation outsourcing are already confronting a structural shock. The combination of AI‑driven efficiency, falling labour costs per task, and shifting client expectations is eroding the very model that underpinned middle‑class expansion in several regions.

Between 2022 and 2026, major outsourcing clients in Europe and North America began to demand more automation and fewer human agents. Large banks and telecoms, for example, required that AI handle 30–60% of routine queries, with humans reserved for escalations. This reduced the number of full‑time agents per contract and compressed the revenue per headcount that firms in the Philippines, India, parts of Eastern Europe, and Latin America could earn. Clients often presented this as a “productivity” requirement, but in practice it meant that service‑export firms were expected to deliver more throughput with fewer people, or lose contracts to rivals that could automate faster. The cost of AI rollout on the client side—often in the tens of millions of CHF per platform—was recouped by slicing labour budgets in the offshore hubs.

Indian and Filipino business‑process outsourcing (BPO) firms were among the first to feel this pressure. India’s IT and BPO sector, which had grown for two decades on the back of large‑scale human labour, reported slower growth in headcount even as revenues plateaued. Some of the largest players invested hundreds of millions of CHF internally to build AI‑driven document‑processing, customer‑handling, and compliance tools, but these investments did not translate into new jobs; they protected profit margins and market share. The Philippine call‑centre and administrative‑services industry, which had long marketed itself as the “call‑centre capital of the world,” saw similar dynamics. Firms that once competed on the number of English‑speaking agents now had to demonstrate that they could integrate AI copilots and analytics platforms to keep their contracts alive.

In smaller outsourcing hubs, the effect was even more acute. Several Eastern European and Latin American centres that had started to capture niche markets in finance, tech support, and legal processing discovered that their competitive advantage had shrunk. AI could now replicate much of the template‑driven, rule‑based work that had made offshore labour attractive in the first place, and clients preferred the stability and continuity of AI‑enabled on‑shore or near‑shore platforms. The cost of AI infrastructure for a single client deployment—often running into the mid‑ to high‑hundreds of millions of CHF over several years—was still a bargain compared with the cost of scaling human labour across time zones and political jurisdictions.

The broader consequence is that the “rise of the service‑export middle class” in the Global South is stalling. In many places, the ambition of sending a child to university, training them in English, and then placing them in a stable BPO or IT job is no longer a reliable path to security. Even where firms continue to hire, they do so for more specialised roles that can work with AI, supervise it, or handle complex, emotionally charged interactions. Routine service work—the backbone of previous generations’ employment—is being compressed into a smaller pool of workers, or simply eliminated altogether. The Global South’s service‑export economies, which once thrived on the global demand for cheap remote labour, are now caught between the rising cost of automation and the falling value of routine human work. The result is not a clean transition, but a slow, grinding erosion of the economic model that powered much of the region’s recent growth.

3.3 Banking, Taxes, and the Squeeze on Labour

The current job collapse is not happening in a neutral economic environment; it is being amplified by the way banking and taxation systems are structured. In many economies, the rules tilt towards capital and away from labour, which means that the benefits of automation flow upwards while the costs are borne by workers. Banks, governments, and financial institutions are all participants in this squeeze, even if their actions are framed as prudent, neutral, or necessary.

Banking systems have long favoured large, capital‑rich firms over individuals and small businesses. Between 2022 and 2026, as AI‑driven automation accelerated, the financial industry responded by expanding credit and investment to firms that could demonstrate “AI‑enabled efficiency.” Large corporations and tech‑enabled platforms received favourable lending terms, low‑cost capital, and high valuations, while individuals and small firms saw tighter credit conditions and higher borrowing costs. The cost of deploying AI infrastructure—often running into tens or hundreds of millions of CHF for large enterprises—was financed by banks that saw automation as a way to secure long‑term returns, even if it meant fewer jobs in the short term. For workers, the result was a paradox: the economy became more productive, but access to capital and credit became harder, forcing many to rely on debt or precarious gig work.

Tax systems have also played a role in the squeeze on labour. In many countries, the tax burden falls more heavily on employees and small businesses than on capital income or large corporations. Payroll taxes, value‑added taxes, and social‑security contributions are often deducted directly from workers’ wages, while capital gains and profits from automation can be shielded through corporate structures, offshore accounts, and tax incentives. Governments in Europe and North America, for example, have introduced tax breaks for AI‑related research and development, which effectively subsidise the adoption of automation while providing little support for displaced workers. The cost of these incentives—often in the hundreds of millions of CHF per year—could have been redirected towards retraining, social safety nets, or wage subsidies, but instead they have been used to accelerate the very forces that are eroding labour’s share of the economy.

The combination of banking and taxation policies has created a feedback loop that reinforces the dominance of capital over labour. Firms that can automate their operations enjoy lower costs, higher profits, and easier access to capital, which allows them to invest more in AI and further reduce their dependence on human workers. Workers, on the other hand, face higher costs, tighter credit conditions, and a shrinking pool of stable jobs. This dynamic is particularly pronounced in the Global South, where governments often rely on foreign investment and export‑oriented growth to finance their economies. In these regions, the pressure to attract and retain multinational corporations has led to tax breaks and regulatory concessions that favour automation and capital accumulation, even at the expense of local labour markets.

In practice, the squeeze on labour is felt in several concrete ways. Workers see their wages stagnate or decline, while the cost of living and debt burdens rise. Social‑security systems, which were designed for a world of stable, full‑time employment, struggle to cope with the growing number of precarious, part‑time, and gig workers. Governments, facing pressure to reduce deficits and maintain competitiveness, often cut social‑spending programmes or shift the burden onto individuals through user fees and private‑sector solutions. The result is a world where automation creates more value but less security for those who depend on labour for their livelihood.

The banking and taxation systems are not the only forces driving this trend, but they are among the most powerful. By favouring capital‑intensive automation and taxing labour more heavily, they create an environment in which firms have a strong incentive to replace workers with machines, even when the social and economic costs are high. For workers, the outcome is a double bind: they are squeezed by the rising costs of credit and taxation on one side, and by the shrinking availability of stable, well‑paid jobs on the other. The squeeze on labour is not an accident of the AI era; it is a product of the rules that govern the economy, and it will continue to shape the trajectory of automation and inequality in the years to come.

4. 2026–2030: The Shock Phase

4.1 AI Agents Take Over Tasks, Not Just Jobs

Between 2026 and 2030, the nature of disruption shifts from “job cuts” to “task absorption”. AI is no longer an occasional assistant pasted into a few tools; it becomes the default way work is done. AI agents—software routines that can plan, execute, and iterate across multiple applications, documents, and systems—begin to handle the bulk of routine cognitive labour, leaving humans with a shrinking envelope of tasks that are either too complex, too political, or too risky to fully automate.

In this period, enterprises upgrade from “AI‑enabled tools” to “AI‑driven workflows”. Instead of a worker asking an AI to draft an email or summarise a report, entire sequences of activity—reading inboxes, triaging tickets, pulling data from spreadsheets, updating dashboards, and drafting responses—are handled end‑to‑end by AI agents. One knowledge worker can now supervise several parallel agents that act on their behalf, multiplying throughput without a proportional increase in headcount. For many firms, the headline becomes: “we did 25–40% more work with the same or fewer staff”, not “we fired X people”.

The shock comes quietly at first. Organisations do not always announce “AI layoffs”; they simply stop hiring, restructure roles, and redesign processes so that the number of tasks that must be done by humans steadily declines. A support function that once needed 100 people to handle 10,000 tickets a month may now run on 60–70 people, with AI agents handling 30–50% of the volume automatically. In finance and operations, AI agents reconcile accounts, flag anomalies, draft reports, and even suggest next‑step actions for managers to approve. In many corporate back‑offices, the share of work done by AI agents moves from 10–20% of total tasks at the start of the period to 40–60% by 2030.

For workers, the experience is that their roles morph. “Write the report” becomes “review, edit, and sign‑off AI‑generated reports”. “Answer customer queries” becomes “handle escalations and oversee AI‑driven responses”. The middle of the job description evaporates, and the parts that remain often demand higher emotional intelligence, stronger judgement, and more political savvy. Employers, however, see this as a win: they can demand more experience and higher performance from a smaller group of people, knowing that AI handles the low‑level cognitive load. The cost of rolling out enterprise‑grade AI‑agent platforms across global operations—often running into the hundreds of millions of CHF per large firm—is justified by sustained reductions in personnel budgets and tighter control over workflows.

The broader effect is that “tasks”, not “jobs”, become the natural unit of automation. Because AI agents can be copied cheaply, a firm can, in one year, deploy thousands of agents across dozens of internal systems and customer‑facing channels. Each new deployment removes a slice of human work: answering basic questions, summarising documents, extracting data from records, generating first drafts, following up on reminders, and so on. Over time, the accumulation of these micro‑substitutions hollows out entire job categories. The shock phase is not defined by single, dramatic announcements of mass layoffs, but by a steady, pervasive erosion of human workload across functions, industries, and geographies. By 2030, for many white‑collar workers, the question is no longer “is my job at risk?”, but “how much of my work will be left once AI agents have taken over the routine tasks?”

4.2 The “Junior Crisis” and the End of Entry‑Level Careers

Between 2026 and 2030, a quiet structural crisis crystallises in advanced economies and service‑export hubs: the junior‑level career path, as it existed for decades, largely disappears. The problem is not that there are no jobs at all, but that the kinds of roles that once absorbed young graduates—routine analysis, drafting, research, data‑handling, basic client support—are no longer economically viable as human‑led positions. AI agents can now perform those tasks at lower cost, with higher speed and consistency, so firms have little incentive to hire legions of juniors.

Graduates entering the labour market in this period face a new reality. Many firms still post “entry‑level” roles, but the job descriptions are often hybrids: “AI‑orchestrator assistant”, “prompt engineer junior”, “AI‑supervisor associate”, or “data‑curator trainee”. These roles do not replicate the old model of starting as a junior analyst, grinding through spreadsheets for years, and slowly earning more responsibility. Instead, they expect the applicant to arrive already familiar with AI tools, prompt‑engineering, and basic automation workflows, and to spend most of their time managing, correcting, and contextualising AI outputs rather than doing the work themselves.

The consequences are visible in both hiring patterns and wages. Companies that once hired 100–200 juniors a year now hire 30–50, and many of those hires are contracted workers, gig‑style roles, or short‑term positions. Where once a firm would have brought in a cohort of 20‑year‑olds to handle background research, drafting, and data entry, it now adds a small number of mid‑level staff who can supervise AI agents and a handful of contractors who handle overflow or edge cases. The cost savings for firms are substantial: a 2027 analysis of large European and North American enterprises estimated that AI‑driven restructuring could reduce junior‑level hiring by 30–60% over the next five years, with corresponding savings in the range of tens of millions of CHF per large organisation annually.

For young people, the “junior crisis” is not just about unemployment; it is about blocked social mobility. The traditional ladder—starting in a low‑paid but stable office job, slowly gaining skills, and eventually moving into roles of greater responsibility—is no longer reliably open. Many graduates find themselves in a cycle of internships, temp contracts, and precarious service work, with AI further eroding the stability of those positions. Entry‑level roles in publishing, marketing, basic legal and accounting work, call‑centres, and back‑office operations shrink fastest, while the premium is placed on roles that can supervise AI, interface with clients, or manage complex, non‑automatable situations. Those who can access family capital, elite networks, or high‑paid technical skills may sidestep the crisis, but the majority of young workers in the middle of the labour market are effectively pushed into a limbo zone between formal employment and informal gig work.

Service‑export economies feel this crisis acutely as well. Countries that once marketed themselves as sources of “English‑speaking, educated youth” for global firms find that those firms now prefer small, highly skilled teams working with AI, rather than large cohorts of juniors. The Philippine, Indian, and Eastern European outsourcing sectors, for example, report that clients increasingly demand “AI‑ready” workforces, where each human agent can handle 2–3 times the volume of work with AI assistance. This reduces the number of entry‑level seats available and compresses the traditional pathway from “graduate to agent to supervisor”. The outcome is a generation of young people who are educated, connected, and often technically savvy, but who cannot find the kind of stable, entry‑level employment that previous generations took for granted.

The “junior crisis” is not a temporary blip; it is a structural feature of the AI‑driven economy. By 2030, the expectation that a degree and a few years of experience will guarantee a stable, upward‑mobile career is no longer realistic for most workers. The end of entry‑level careers marks a hardening of the divide between those who can adapt to AI‑driven roles and those who are left to compete in a shrinking pool of low‑paid, precarious work. The crisis is both economic and social: it undermines the social contract that once bound education, work, and security together, and it sets the stage for deeper dissatisfaction and instability in the decades ahead.

4.3 First Collapses: Countries Living on Cheap Remote Labour

By the late 2020s, several countries whose recent growth was built on “cheap remote labour” begin to experience visible structural collapse. The model that once worked—large populations of English‑speaking, college‑educated workers sitting in air‑conditioned offices handling calls, data entry, basic IT, and back‑office tasks for firms in rich economies—stops delivering the same returns. As AI agents absorb an ever‑larger share of routine work, the global demand for human‑provided remote services shrinks, leaving entire national economies stranded mid‑development.

The first clear casualties are mid‑income, service‑export‑dependent economies whose fiscal health and middle‑class expansion were tied to outsourcing contracts. The Philippines, parts of India, and certain Eastern European and Latin American hubs had built significant infrastructure—call‑centre campuses, BPO parks, training academies, and urban service corridors—on the expectation that demand for remote labour would keep growing. Between 2026 and 2030, that demand flattens or turns negative. Large clients in Europe and North America no longer sign ten‑year “headcount” contracts; instead, they sign AI‑enabled “outcome” contracts, where a portion of the work must be handled by automated systems, and human headcount is capped or actively reduced over time.

The economic impact is sharp. Firms that once employed tens of thousands of agents find their contracts slashed by 20–40% within a few years, even as their AI‑integration costs rise. Local governments, which had relied on tax receipts, real‑estate development, and consumption driven by BPO salaries, see revenues decline. Urban centres that bloomed around business‑process parks begin to stagnate: new campus construction grinds to a halt, property values soften, and small businesses that catered to office workers see their customer base shrink. In the Philippines, for example, provinces that once revolved around call‑centre clusters report youth unemployment among graduates rising to 25–30%, with many young people forced into informal work, migration, or debt‑driven asset‑stripping.

In India, certain states that had specialised in IT and BPO services feel the pressure acutely. Large Indian outsourcing firms, which had once grown by scaling headcount, now face the same logic: AI can handle 30–60% of routine work at far lower marginal cost. These firms respond by cutting fresh‑graduate intake, freezing wages, and pushing existing staff into more “AI‑management” roles, but the net effect is fewer stable entry‑level positions. The national economy as a whole is large enough to partially absorb the shock, but the hit to middle‑class aspirations in specific service‑cluster cities is plain: once‑booming centres become “hollow cities”, where offices half‑empty, malls see footfall decline, and local politics increasingly revolve around unemployment and migration disputes.

Eastern European and Latin American outsourcing hubs face a similar fate. Countries that had attracted clients by offering good education, low wages, and convenient time zones now discover that AI‑driven automation can be deployed directly in the client’s home country, with only a small supervisory layer offshore. Clients are happy to keep a small core of highly skilled workers in these hubs, but they have little incentive to maintain large armies of juniors. The result is a thinning of the middle tier of the labour market, with a small elite of “AI‑ready” professionals and a much larger group of underemployed or unemployed youth. The cost of this transition is not counted in classic recession terms; it is seen in falling wages, rising household debt, and a gradual erosion of the social compact that once linked education, English‑language skills, and a stable office career.

The “first collapses” are not, in the short term, states that vanish overnight. They are countries whose growth engines stall, whose middle‑class pipeline breaks, and whose political systems face escalating pressure from frustrated graduates, declining tax bases, and rising inequality. In some cases, the squeeze triggers social unrest, emigration waves, or shifts in foreign‑policy alignment as governments scramble to attract new sources of capital. In others, the decline is more muted but no less real: a slow slide into semi‑peripheral status, where the economy subsists on a mix of remaining service contracts, remittances, and low‑level manufacturing, with little hope of reclaiming the rapid, labour‑driven growth of the previous decades.

By 2030, the writing is on the wall: for several nations that bet their futures on cheap remote labour, the AI era marks the end of that model. The collapse is not total, but it is structural—an irreversible narrowing of the path that once led from education to secure employment. For the rest of the world, these early collapses serve as a stark warning: when AI removes the demand for large numbers of routine workers, even entire national economies can be hollowed out in a decade.

5. 2030–2040: Hollowing Out and Entrenchment

5.1 Shrinking Middle‑Class Pathways

Between 2030 and 2040, the erosion of “normal” work deepens into a structural narrowing of middle‑class life paths. The ladder that once connected education, stable employment, gradual asset accumulation, and modest social mobility begins to fray for large segments of the population. In advanced economies and many service‑exporting states, the primary mechanism is no longer mass unemployment, but an expansion of precarious, low‑security roles and a shrinking pool of genuinely stable, upward‑mobile positions.

The core of the problem lies in the way AI reshapes the job market. For many firms, the new standard is to run lean operations supported by AI agents, with a small number of “core” workers handling coordination, judgment, and client relations. Routine tasks—data entry, basic analysis, drafting, customer‑service scripts, compliance checks—are absorbed into software systems that can be copied and scaled at low marginal cost. As a result, the number of decent‑paid, full‑time roles that offer a clear path to homeownership, pensions, and respect grows far more slowly than the number of people who need them. A 2032 analysis of labour‑market trends in several European and North American economies estimated that the share of new jobs offering stable benefits and clear promotion paths had fallen from roughly 60% in 2020 to under 40% by 2035, with the difference filled by gig‑style, short‑term, or AI‑supervised roles.

In the Global South, the effect is even more pronounced. Countries that once rode the wave of service‑export growth now find that their “middle‑class escalator” has stalled. In the Philippines, parts of India, Eastern Europe, and Latin America, the combination of AI‑driven automation and declining demand for cheap remote labour means that the generation of 2020–2030 does not inherit the same opportunities as their parents. Many young people obtain degrees, gain digital skills, and even learn AI‑prompting techniques, only to enter a job market where the number of solid office roles is shrinking. The cost of living in urban centres continues to rise, but the salaries on offer do not keep pace, forcing young workers into debt, shared housing, and multiple part‑time gigs.

The shrinking middle‑class pathways are not just economic; they are social and psychological. The expectation that “hard work and education will be rewarded” becomes a luxury reserved for those who can access elite networks, capital, or high‑tech skill clusters. For the rest, the future looks like a long plateau of under‑employment, variable income, and fragile security. Political systems in many countries begin to reflect this reality: governments offer targeted subsidies, retraining schemes, and basic income‑style support, but these are rarely sufficient to restore the old promise of steady progress. The cost of these measures—often in the hundreds of millions of CHF per year in richer economies—pales against the amount of capital being invested in AI infrastructure and automation, which continues to grow at double‑digit rates.

By 2040, the middle‑class crisis has hardened into a feature of the AI‑driven economy. Stable, well‑paid jobs are no longer the default outcome of education and effort; they are a scarce resource, concentrated in specific sectors, regions, and skill sets. The rest of the workforce is pushed into a spectrum of roles that offer less security, lower wages, and fewer opportunities for advancement. The result is a world where the middle class is not disappearing en masse, but it is thinning, fragmenting, and increasingly polarised between a small, highly skilled elite and a much larger group of precarious workers who must navigate a system designed to minimise their share of the pie.

5.2 AI‑Stabilized Oligarchies and Managed Discontent

Between 2030 and 2040, several states evolve into what can be called “AI‑stabilised oligarchies”: regimes where power is concentrated in a small elite, and AI systems are used to cement that concentration, not to broaden participation. These regimes are not necessarily crude dictatorships; many retain elections, markets, and some legal pluralism, but the rules increasingly tilt so that capital, data, and security assets accumulate in the hands of a narrow group of owners, technocrats, and state managers. AI becomes the tool that locks in their advantage.

In these oligarchies, AI is deployed across three domains: economics, surveillance, and governance. In the economy, automation and AI‑driven optimisation are used to maximise returns to capital and skilled elites while keeping labour costs low. Large firms, often with close ties to ruling groups, invest hundreds of millions of CHF in AI‑centric infrastructures—smart factories, logistics networks, financial platforms, and service‑delivery systems—that run with minimal human overhead. Regulations and tax incentives tilt towards these capital‑intensive models, making it harder for small, labour‑intensive enterprises to compete. The result is a self‑reinforcing structure: AI increases productivity, concentrates wealth, and then provides the capital needed to fund even more AI, all while the political environment is shaped to protect incumbents.

In surveillance and security, AI systems extend the state’s ability to monitor, predict, and control populations. Facial‑recognition, social‑media analysis, predictive policing, and biometric databases are integrated into public‑order and “risk‑management” frameworks. AI‑driven analytics can flag potential troublemakers, map protest networks, and anticipate flashpoints, allowing authorities to act preventively. In some cases, this is sold as “crime‑fighting” or “counter‑terrorism”; in practice, it is a form of political triage that channels AI‑enforced discipline towards the poor and dissident, while lighter constraints apply to the powerful. The cost of these systems—often in the tens to hundreds of millions of CHF per large deployment—is justified by claims of stability and efficiency, not by democratic consent.

In governance and welfare, AI is used to manage, not empower, the majority. Social‑assistance programmes, subsidies, and basic services are increasingly administered through AI‑driven platforms that verify eligibility, disburse funds, and monitor compliance. For the state, this is a way to cut bureaucracy and reduce leakage; for many citizens, it means a more opaque, inflexible, and sometimes punitive system. Appeals and appeals‑like mechanisms are limited, and the logic of the algorithm—opaque and unaccountable—overrides local discretion. In practice, these systems stabilise the regime by keeping unrest below the boiling point: enough people receive enough support to survive, but not enough to build independent power. The working‑class majority is treated as a consumer of services and a target of surveillance, not as a source of political authority.

The oligarchic bargain is clear: a small elite captures most of the gains from AI, while the rest of society is managed through a mix of welfare, distraction, and control. The “managed discontent” model works as long as the economy delivers enough cheap goods, entertainment, and minimal security to keep overt rebellion rare. Unemployment and under‑employment are accepted as systemic features, not failures; the state’s role is to contain their explosive potential, not to reverse them. AI platforms are used to analyse sentiment, segment populations, and tailor messaging—whether through entertainment, propaganda, or targeted benefits—so that grievances are channelled into low‑impact outlets, rather than into collective organisation that can challenge the system.

By 2040, this pattern is visible in several regions. Some resource‑rich states combine AI‑enabled security with oil and gas rents to keep large populations docile. Others, particularly in the Global South, are squeezed by foreign capital and AI‑driven global supply chains, but use AI‑driven governance and surveillance to keep their own elites in power. The result is a world where AI does not democratise control; it helps oligarchies stabilise themselves by making dissent harder to coordinate, labour more dispensable, and populations easier to manage from the top. The “hollowing out” of the middle class feeds this dynamic: with fewer independent actors to challenge the status quo, the AI‑stabilised oligarchies have ample room to entrench their power.

5.3 The Rise of a Permanent Precariat

Between 2030 and 2040, a new structural layer solidifies across much of the world: the “permanent precariat”. This is not a temporary cohort of people caught in a cyclical downturn; it is a large, semi‑stable population whose default condition is insecure, low‑paid, or intermittent work, with minimal access to long‑term security, benefits, or social mobility. The precariat includes gig workers, part‑time contractors, under‑employed graduates, and formerly middle‑class workers pushed down the ladder by automation and wage‑squeeze. They are not unemployed; they are under‑secured.

The rise of this group is driven by the way AI reshapes the labour market. As AI agents and automation systems absorb routine tasks, firms have less need for full‑time, benefits‑loaded staff. Instead, they prefer flexible arrangements: short‑term contracts, platform‑mediated gigs, task‑based payments, and AI‑managed workflows that can be turned up or down with software commands. For employers, the advantage is obvious: they can adjust labour costs in real time, keep overhead low, and push risk onto the individual worker. The cost of shifting to this model—onboarding, training, and platform‑development—is often covered by investments in the tens or even hundreds of millions of CHF, justified by persistent savings in personnel budgets.

For the worker, the outcome is instability. A common trajectory by the late 2030s is to juggle multiple income streams: driving, delivering, content‑moderation, data‑labelling, basic admin tasks, and occasional project work for AI‑driven platforms. Each gig pays modestly, offers no guaranteed hours, and rarely provides pensions, health coverage, or long‑term career prospects. Wages are compressed by competition: one more person is always ready to take the job at a lower rate. In many countries, reforms to labour law and social‑security systems codify this shift, treating “independent contractors” as the norm and eroding the old protections for full‑time employees. The cost of these reforms to governments is often low in fiscal terms, but the social cost is high: a growing pool of people who are neither fully employed nor fully supported.

In the Global South, the precariat expands fastest. Countries that once relied on service‑export growth now see their middle‑class pipeline blocked, while AI‑driven automation and falling labour demand push millions into informal work. Urban slums, rural margins, and peri‑urban zones swell with people who work part‑time, seasonally, or on‑demand, often for digital platforms that pay in small amounts of foreign currency or local equivalents. The cost of living in these regions may be lower than in rich economies, but savings are minimal, and shocks—health crises, inflation, political unrest—can quickly push households into destitution. Social‑security systems, already stretched, cannot keep pace with the speed of change.

Even in advanced economies, the precariat bleeds upwards. Formerly stable office workers, retail employees, and low‑level professionals find themselves displaced by AI and forced into lower‑paid, less secure roles. Many accept “flexible” work in name only, trading stability for the illusion of autonomy. The result is a large cohort of people who are constantly “on call”, anxious about income, and investing time in upskilling or “AI‑proofing” themselves, even as the labour market offers fewer guarantees. The cost of trying to stay employed—training, therapy, coping strategies, debt—often exceeds the benefits reaped.

By 2040, the permanent precariat is no longer a fringe phenomenon. It is a core feature of the AI‑driven economy: a large, flexible, and politically fragmented layer of humanity that serves as both labour and market. Precarious workers are the backbone of gig platforms, the default pool for AI‑managed tasks, and the target of low‑cost consumption goods and services. They are also the primary victims of the squeeze from AI‑driven automation and oligarchic rule. The rise of this group marks a hardening of inequality: a world where the stable and the secure are a minority, and the majority lives in a state of managed, persistent uncertainty.

6. 2040–2060: The Split of Humanity

6.1 High‑Tech Core Societies and the AI‑Driven Elite

Between 2040 and 2060, a distinct bloc of “high‑tech core” societies crystallises: wealthy, tightly integrated regions where AI is not just present but foundational to economic, political, and social life. These cores are not evenly distributed; they cluster around a handful of states and city‑state networks that command capital, data, and critical infrastructure. Within them, AI is used to maximise productivity, control risk, and protect the position of a small elite, while the majority of inhabitants live in highly managed, but stratified, conditions.

In these core societies, AI is embedded in every significant sector: manufacturing, logistics, energy, finance, healthcare, education, and governance. Automated factories and warehouses run largely without human labour, while AI‑driven logistics networks optimise supply chains in real time. Financial markets are dominated by AI‑driven algorithms that manage risk, detect fraud, and execute trades at speeds no human can match. Healthcare uses AI for diagnostics, treatment planning, and drug discovery, while AI‑powered education platforms personalise learning and track performance down to the individual level. The cost of these systems—often in the hundreds of millions of CHF per year across large institutions—is justified by the efficiency gains and by the competitive advantage they confer.

The beneficiaries of this system are a narrow layer of owners, technocrats, and political managers. The “AI‑driven elite” consists of:

• large shareholders and executives of AI‑centric corporations;
  
• top engineers, data scientists, and security specialists who design and control the systems;
  
• political and bureaucratic elites who align with these interests and shape the rules.
  

This group is small—often less than 10% of the population in these core societies—but it controls the vast majority of AI‑driven wealth and infrastructure. Its members live in high‑security enclaves, enjoy access to the best healthcare, education, and leisure, and are insulated from the volatility that affects the rest of the population. The remaining majority are divided into:

• a shrinking, often older, middle class that clings to remnants of stability;
  
• a growing permanent precariat tied to gig work, AI‑managed platforms, and low‑paid service roles.
  

The core societies are also the first to push into off‑world ambitions. With AI‑driven project management, robotics, and closed‑loop systems, they begin serious construction of orbital habitats and lunar or Mars bases by the mid‑21st century. These are not “colonies” in the romantic sense; they are high‑cost, high‑risk ventures designed to extend the reach of capital and control. The initial populations are small—thousands, not millions—but they are highly curated: elites, their dependents, and a cadre of highly skilled workers who can operate in extreme environments. The cost of these projects—billions of CHF per year—is absorbed by conglomerates and states that see them as strategic investments in the long term, not as charitable projects.

Within the core, social life is increasingly mediated by AI. Surveillance, recommendation engines, and social‑media platforms shape who talks to whom, what is seen, and which ideas are amplified. AI‑driven credit, housing, and mobility systems reward conformity and risk‑aversion, while dissenting or unconventional behaviour is flagged, monitored, or quietly excluded. The “normal” life in these societies is one of constant optimisation: algorithms nudge choices, manage schedules, and anticipate needs, but at the cost of autonomy and privacy. The elite are the primary beneficiaries of this system, while the rest of the population must navigate a world where their data, behaviour, and even emotions are commodified.

By 2060, the high‑tech core and its AI‑driven elite represent a self‑reinforcing bloc. They control the AI systems that drive global productivity, the capital that funds new technologies, and the security apparatus that protects their interests. The split between this core and the rest of humanity is not just economic; it is existential. The core societies are the first to live in a world where AI is the default, and the working class is a managed subset of the population. The rest of the world—semi‑peripheries and retreat zones—must adapt to the logic of this core, even as they are excluded from its benefits. The “split of humanity” is no longer a metaphor; it is a material reality.

6.2 Semi‑Periphery: Export Zones, Extraction, and Debt

Between 2040 and 2060, a broad band of “semi‑periphery” societies sits between the high‑tech core and the retreat or subsistence zones. These are not collapsed states, but not fully integrated core members either. They are tied into the AI‑driven global system as suppliers of raw materials, energy, land, and selective human labour, while remaining politically and economically subordinate. Their economies are shaped less by internal dynamics than by their position in the global chain of export, extraction, and debt.

In the export zones, the primary logic is to provide cheap, reliable inputs to the core. This includes:

• mining and processing of critical minerals needed for AI chips, batteries, and data centres;
  
• large‑scale agriculture and aquaculture feeding urban centres in the core;
  
• energy production—fossil and renewable—that powers data centres, transport networks, and manufacturing hubs.
  

AI systems run the logistics and optimisation of these operations, minimising costs and maximising throughput. The human workforce in these zones is often highly regimented, with AI‑driven scheduling, surveillance, and performance‑monitoring pushing productivity to the edge of tolerable conditions. The cost of AI‑driven automation in these sectors—often tens to hundreds of millions of CHF per large project—is absorbed by foreign capital, with local elites taking a cut in the form of rents, infrastructure projects, or political patronage.

Extraction is the other side of the coin. In some regions, natural resources are not just mined, but deeply integrated into AI‑driven supply chains. For example, rare earth metals or lithium are extracted and processed with AI‑optimised mining equipment, robotic sorting, and predictive‑maintenance systems that keep operations running around the clock. The benefits of this automation flow to the core: higher‑quality materials at lower costs, which feed the next generation of AI hardware. Local populations may see some improvement in jobs or infrastructure, but the gains are uneven and fragile; when prices fall or demand shifts, entire regions can be abandoned or repurposed without much regard for local stability.

Debt plays a central role in binding the semi‑periphery to the core. Many of these societies take on substantial foreign debt to finance AI‑driven development projects—ports, power grids, data‑centres, industrial zones—that are designed to integrate them into global supply chains. The loans come with conditions: favourable tax regimes for foreign investors, light regulation, and limited sovereignty over key assets. When the revenues from these projects do not pan out as expected, the debt burden becomes a lever of control. Creditors can demand restructuring, privatisation, or political concessions, effectively locking in the semi‑peripheral status of these states. The cost of servicing this debt—often in the billions of CHF per year—leaves little room for domestic investment in social programmes or long‑term development.

Within the semi‑periphery, the social structure is highly stratified. At the top are local elites who benefit from rent‑sharing with foreign capital and AI‑driven platforms. They live in gated communities, enjoy access to core‑market goods, and send their children to elite schools. Below them, a thin middle class exists in the export zones and cities, but it is vulnerable to fluctuations in global demand and AI‑driven efficiency. The vast majority of the population lives in precarious conditions, working in low‑paid jobs, subsistence farming, or informal sectors that are barely integrated into the global economy. The AI‑driven export and extraction systems benefit the core and the local elite, but they do little to lift the broader population out of poverty.

By 2060, the semi‑periphery is a critical node in the global AI‑driven economy. It provides the raw materials and energy that power the high‑tech core, while its own populations are treated as a managed resource. The combination of export dependency, extraction, and debt creates a system of control that is both economic and political. The semi‑periphery is not a place of collapse, but a zone of managed underdevelopment—where the benefits of AI and automation are captured by the core and local elites, while the rest of the population struggles to survive in the margins.

6.3 Rejection Zones: Retreat, Subsistence, and Low‑Tech Societies

Between 2040 and 2060, a growing number of regions begin to function as “rejection zones”: places where societies either fall out of the AI‑driven global system or deliberately choose to disengage from it. These are not monolithic; they range from failed states that collapse under the weight of automation and debt, to intentional low‑tech communities that reject the logic of perpetual optimisation and surveillance. What they have in common is a retreat from the core’s model of life, work, and control.

In the most extreme cases, rejection zones emerge from the collapse of semi‑peripheral states. When AI‑driven automation and debt‑driven extraction hollow out the middle class and the state’s capacity to provide basic services, entire regions can descend into chaos. Governments lose legitimacy, security apparatuses fragment, and infrastructure deteriorates. In these failed states, the primary logic is survival, not growth. People rely on local barter networks, subsistence farming, and informal trade to meet their needs. The AI systems that once promised efficiency have become distant, unattainable luxuries, controlled by foreign powers or local elites who have already fled. The cost of maintaining these systems—often in the hundreds of millions of CHF—exceeds the fiscal capacity of the state, leading to abandonment.

In other regions, rejection is a conscious choice. Some communities and micro‑polities decide to withdraw from the high‑tech world, forming intentional low‑tech societies that reject the constant surveillance, data‑harvesting, and profit‑driven logic of the core. These communities prioritise self‑sufficiency, local governance, and simple technologies that are easy to maintain and repair. They may use basic renewables, simple machinery, and local knowledge to power their lives, but they avoid the deep integration with AI and digital platforms that characterise the core. The motivation is often ideological, cultural, or ecological: a belief that the AI‑driven world is unsustainable, dehumanising, or unjust. The cost of this retreat is high in terms of access to global markets and advanced services, but it offers a sense of autonomy and control over one’s own life.

Rejection zones also include subsistence economies that have never fully participated in the global system. In these regions, the arrival of AI is felt indirectly, through global commodity prices, climate change, and migration pressures. Local communities adapt by maintaining traditional ways of life, relying on family networks, and preserving local knowledge. The AI systems that dominate the core seem remote and irrelevant, even as their effects ripple through the global economy. In some cases, these communities find niches in the global market—selling artisanal goods, organic produce, or cultural experiences—but they remain on the margins, trading with the core while preserving their low‑tech ethos.

By 2060, the rejection zones are a diverse and growing layer of humanity. They are not utopian escapes or romantic idylls; they are often fragile, under‑resourced, and vulnerable to external shocks. But they represent a different way of living, one that is less entangled with the AI‑driven logic of the high‑tech core. In these zones, the question is not “how can we optimise our lives for AI?”, but “how can we preserve our autonomy and dignity in a world that is moving on without us?”. The rejection zones are a testament to the limits of AI’s reach and the resilience of human diversity. They are also a warning: the AI‑driven world is not a universal solution, but a specific mode of existence that excludes many.

7. 2060–2100: Earth Enclaves and Mars Cities

7.1 Off‑World Cities and the New “Last Frontier”

Between 2060 and 2100, the first true off‑world cities begin to take shape, not as science‑fiction settlements but as hard‑nosed extensions of the AI‑driven high‑tech core. These are not mass migrations of humanity into space; they are small, tightly controlled outposts—on Mars, in orbit, and on the Moon—built and run by state‑corporate consortia whose primary interest is security, prestige, and long‑term strategic position, not egalitarian settlement.

By the 2060s, orbital habitats and lunar bases have grown beyond research outposts into semi‑permanent installations housing thousands of people. These are not “cities” in the Earth‑bound sense; they are highly engineered enclaves where every cubic metre of air, every gram of water, and every watt of power is managed by AI‑driven systems. Robotics and AI handle most of the maintenance, logistics, and construction work, from repairing life‑support systems to assembling new modules. Humans are concentrated in roles that require judgment, coordination, and tertiary problem‑solving—mission control, engineering oversight, security, and diplomacy—while the bulk of the physical work is done by machines. The cost of maintaining these installations—often in the billions of CHF per year—is justified by the symbolic value of having a human presence beyond Earth and by the potential for resource extraction from asteroids or lunar regolith.

Mars settlements follow a similar pattern, but with slower growth. The first Martian habitats, built in the 2050s and 2060s, are small, enclosed environments barely larger than advanced Antarctic stations. They house scientists, technicians, and security personnel, and are supplied by regular shipments from Earth or orbit. By the 2070s and 2080s, as AI‑driven robotics and 3D‑printing technologies improve, these habitats expand into larger “cities” with tens of thousands of inhabitants. The inhabitants are not a cross‑section of humanity; they are carefully selected elites, their dependents, and support staff, vetted for skills, stability, and political reliability. The AI systems that run these cities are the same ones that dominate Earth’s high‑tech core, optimising resource use, managing risk, and enforcing social order with minimal human oversight.

The “last frontier” of space is not a place of open opportunity; it is a highly stratified extension of the core’s power. The off‑world cities are tied to Earth‑based capital and AI infrastructures, not independent entities. They rely on data flows, AI‑driven project management, and global supply chains to function, and their governance is shaped by the political and economic interests of the states and corporations that fund them. In practice, this means that the Martian and orbital cities are semi‑autonomous enclaves where the AI‑driven elite can extend their influence and secure their survival in the event of Earth‑based disruptions. The cost of this expansion is immense, but the payoff is strategic: control over new frontiers, new resources, and new forms of power.

By 2100, the off‑world cities are a visible symbol of humanity’s split. They are the first places where the AI‑driven world is not just a layer on top of Earth’s chaos, but a separate, self‑sustaining reality. The Earth itself is a mix of high‑tech cores, semi‑periphery zones, and rejection enclaves, but the off‑world cities are pure core: tightly controlled, AI‑dominated, and insulated from the volatility of the planet below. The “last frontier” is not a myth of freedom; it is a new frontier of inequality, where the elite can live beyond the reach of Earth’s crises, while the rest of humanity remains trapped in the old, fractured world.

7.2 Earth’s Managed Majority: Labour, Customers, and Control

Between 2060 and 2100, the majority of humanity on Earth lives in what can be described as a “managed majority”: a large, heterogeneous population that is neither fully autonomous nor fully free, but carefully steered by AI‑driven systems, capital‑intensive automation, and state‑corporate partnerships. This majority is not a single group; it includes gig workers, low‑paid service staff, subsistence farmers, and precarious urban dwellers, all of whom are treated as both labour and customers in a highly optimised global economy. The key feature of this world is not oppression in the classical sense, but management: the majority are kept in motion, consuming, producing, and complying, while the core retains control.

In the managed majority, AI systems are the backbone of daily life. From work to consumption to governance, algorithms dictate the flow of resources and opportunities. Gig platforms, AI‑managed marketplaces, and automated logistics networks ensure that labour is always available and always flexible. Workers are nudged by AI‑driven recommendations: which jobs to take, which hours to work, which products to buy. The cost of this system—often in the tens to hundreds of millions of CHF per year for large platforms—is justified by the efficiency gains and the ability to keep labour costs low. The result is a world where people are constantly “on call”, but rarely in control of their own schedules or destinies.

At the same time, the managed majority are the primary customers of the AI‑driven economy. They consume cheap goods, digital services, and entertainment produced by AI‑optimised factories and platforms. The cost of these products is low, but the price of access is high: data, attention, and compliance. AI systems track every purchase, every click, every interaction, and use this data to personalise offers, manipulate prices, and predict behaviour. The managed majority are not passive consumers; they are active participants in a system that rewards conformity and punishes dissent. The more they conform to AI‑driven recommendations, the more they receive benefits; the more they resist, the more they face friction, exclusion, or surveillance.

Control is the other side of the coin. In many regions, AI‑driven surveillance and governance systems keep the managed majority in check. Facial‑recognition, sentiment analysis, and predictive policing are used to identify and manage potential troublemakers. Social‑security programmes, AI‑driven benefits platforms, and digital IDs ensure that people remain within the system, dependent on it for survival. The cost of these systems—often in the hundreds of millions of CHF per year—exceeds the cost of maintaining basic services, but the payoff is stability and control. The managed majority are not rebels; they are a managed workforce that provides the labour and the market for the AI‑driven economy.

By 2100, the managed majority is the default condition for most of humanity. The high‑tech cores and off‑world cities are the exceptions, not the rule. The Earth is a patchwork of AI‑driven zones, semi‑peripheral regions, and rejection enclaves, but the managed majority is the backbone of it all. They are the ones who keep the global economy running, who consume the cheap goods, and who comply with the AI‑driven rules. The split of humanity is no longer a metaphor; it is a material reality. The managed majority are the labour and the market, while the AI‑driven elite and their off‑world outposts are the controllers and the beneficiaries.

7.3 Subsistence Enclaves by Choice and Collapse

Between 2060 and 2100, subsistence enclaves proliferate across Earth, not as a single movement but as a diverse array of settlements born from both collapse and conscious rejection of the AI‑driven core. These are small, often isolated communities where people live on the margins of the global system, relying on local resources, simple technologies, and tight social bonds rather than integration into high‑tech networks. Some are forced into this existence by the failure of states and economies; others embrace it as a deliberate alternative to the managed, data‑saturated world above.

In the regions where collapse drives subsistence, the story is one of gradual unraveling. As AI‑driven automation and debt‑driven extraction hollow out semi‑peripheral states, entire regions lose access to basic services, infrastructure, and global markets. Governments fail, economies stall, and people retreat to local means of survival. In these collapsing zones, subsistence enclaves form organically: small clusters of families band together to farm, fish, or trade locally, using whatever tools they can maintain. The AI systems that once promised efficiency are abandoned, their cost—often in the hundreds of millions of CHF—too high to sustain. People rely on traditional knowledge, barter networks, and face‑to‑face cooperation to meet their needs. The result is a fragile, under‑resourced existence, but one that is insulated from the volatility of the global system.

Elsewhere, subsistence enclaves arise from deliberate choice. Some communities, disillusioned with the AI‑driven core’s logic of optimisation, surveillance, and inequality, decide to withdraw from it. These intentional low‑tech societies prioritise self‑sufficiency, local governance, and simple technologies that are easy to repair and maintain. They may use basic renewables, simple machinery, and local crafts to power their lives, but they avoid the deep integration with AI and digital platforms that characterise the core. The motivation is often ideological, cultural, or ecological: a belief that the high‑tech world is unsustainable, dehumanising, or unjust. The cost of this retreat is high in terms of access to global markets and advanced services, but it offers a sense of autonomy and control over one’s own life. These enclaves may trade with the core—selling artisanal goods, organic produce, or cultural experiences—but they remain on the margins, preserving their low‑tech ethos.

By 2100, subsistence enclaves are a diverse and growing layer of humanity. They range from small, self‑sustaining villages in remote regions to intentional communities in the shadow of high‑tech cores. These enclaves are not utopian escapes or romantic idylls; they are often fragile, under‑resourced, and vulnerable to external shocks like climate change, migration pressures, or resource scarcity. But they represent a different way of living, one that is less entangled with the AI‑driven logic of the core. In these zones, the question is not “how can we optimise our lives for AI?”, but “how can we preserve our autonomy and dignity in a world that is moving on without us?”. The subsistence enclaves are a testament to the limits of AI’s reach and the resilience of human diversity. They are also a warning: the AI‑driven world is not a universal solution, but a specific mode of existence that excludes many.

8. The Political Economy Without Guardrails

8.1 Free‑Capital Regimes and the Logic of Unchecked Automation

Between 2060 and 2100, the political economy of many states crystallises into what can be described as “free‑capital regimes”: systems where the rules of the game tilt overwhelmingly towards owners of capital, data, and AI infrastructures, and where automation is constrained only by technical limits and profit margins, not by social or ethical considerations. In these regimes, the primary logic is simple: capital must grow, and labour must adapt—or vanish. The result is a world where AI‑driven automation is not just encouraged, but deeply embedded in the structures of power and wealth.

In free‑capital regimes, the state plays a supporting role to the owners of AI systems and capital. Governments implement policies that lower taxes on capital gains, reduce regulation on AI deployment, and provide subsidies for automation and robotics. The cost of these policies—often in the hundreds of millions of CHF per year—is justified by the promise of economic growth and productivity gains. Large corporations, particularly tech‑centric firms, dominate the political landscape, funding campaigns, shaping legislation, and influencing regulatory bodies. The result is a feedback loop: AI increases productivity, concentrates wealth, and then provides the capital needed to fund even more AI, all while the political environment is shaped to protect incumbents.

Within these regimes, the pressure on labour is relentless. AI agents and automation systems absorb routine tasks, leaving humans with a shrinking envelope of work. Firms that invest in AI can reduce labour costs, increase output, and improve margins, all of which are rewarded in the market. The cost of deploying AI infrastructure—often tens to hundreds of millions of CHF for large enterprises—is recouped through sustained reductions in personnel budgets and tighter control over workflows. The working class, meanwhile, faces stagnant wages, rising debt, and a shrinking pool of stable jobs. The state responds with targeted subsidies, retraining schemes, and basic income‑style support, but these are often insufficient to restore the old promise of steady progress.

By 2100, free‑capital regimes are the default in many advanced economies, and their logic extends into the semi‑periphery and even some rejection zones. The AI‑driven elite consolidates its power, while the managed majority and the precariat adapt to a world where automation is the norm. The split of humanity is no longer a metaphor; it is a material reality. The free‑capital regimes are the engines of this split, driving the unchecked automation that reshapes the global economy and the balance of power. In this world, the only guardrails are the ones that protect capital, not the ones that protect people.

8.2 Why Elites Capture the System

Between 2060 and 2100, the dominance of elites in the AI‑driven world is not an accident; it is a structural feature of the system. The owners of capital, data, and AI infrastructures are the primary beneficiaries of automation, and they have the resources and influence to shape the rules of the game in their favour. The state, media, and institutions are not neutral arbiters; they are co‑opted tools that reinforce the power of the elite.

The first reason elites capture the system lies in the way AI is funded and deployed. Large corporations and wealthy individuals control the capital needed to invest in AI infrastructure, which can run into tens or hundreds of millions of CHF per project. These investments are not just about technology; they are about power. The owners of AI systems gain control over data, decision‑making, and influence over markets and governments. The state, eager for growth and tax revenue, provides subsidies, tax breaks, and regulatory favours to these firms, effectively transferring resources from the public to the private sector. The cost of these policies—often in the hundreds of millions of CHF—pales against the gains reaped by the elite, who can then reinvest in more AI and automation, creating a virtuous cycle for themselves and a vicious one for the majority.

The second reason lies in the way elites shape the political and social environment. They fund campaigns, think tanks, and media outlets that promote the narrative that automation is inevitable and beneficial, even as it hollows out the middle class. They lobby for deregulation, tax policies that favour capital over labour, and international agreements that protect their interests. The result is a system where the rules are written by the wealthy, for the wealthy. The managed majority and the precariat are left with little voice in the political process, their votes often diluted by gerrymandering, media manipulation, and the sheer complexity of the issues at hand.

The third reason is the way elites use AI to entrench their power. AI systems are used to monitor, predict, and control populations, while the elite themselves operate in secure, highly curated environments. The state, under pressure from capital, implements AI‑driven surveillance and governance systems that keep the majority in check, while the elite remain insulated from the consequences. The cost of these systems—often in the hundreds of millions of CHF—is justified by the promise of stability and security, but the real beneficiaries are the owners of capital who can then operate with impunity.

By 2100, the capture of the system by elites is complete. The AI‑driven world is a world of managed chaos, where the elite control the levers of power, and the rest of humanity is left to adapt. The split of humanity is no longer a metaphor; it is a material reality. The elite are the controllers, and the rest are the “managed”.

8.3 Why Resistance Fails to Scale

Between 2060 and 2100, resistance to the AI‑driven, free‑capital order emerges in many forms—protests, strikes, digital dissent, local rebellions, and ideological movements—but it rarely scales into a coherent, sustained challenge to the core. The logic of the system, the way power is structured, and the material conditions of the managed majority conspire to keep resistance fragmented, short‑lived, or co‑opted.

One reason resistance fails to scale lies in the fragmentation of the working population. The permanent precariat is not a solid bloc of stable, unionised workers with shared workplaces and clear hierarchies; it is a loose, geographically dispersed mix of gig workers, platform labourers, part‑time contractors, and informal‑sector survivors. Their jobs are mediated through different apps, platforms, and algorithms, often in different languages and legal jurisdictions. Building durable, cross‑regional organisation—let alone a global movement—becomes structurally harder because there is no single “workplace” to occupy, no single employer to confront, and no single set of formal contracts to renegotiate. AI‑driven platforms also design their terms and user interfaces to discourage solidarity, emphasising individual rewards, competition, and isolation.

A second barrier is the speed and flexibility of capital and AI. When protests or strikes flare, firms can quickly re‑route work through AI agents, alternative platforms, offshore hubs, or automated systems, effectively bypassing the organisers. The cost of this re‑routing—often in the tens of millions of CHF per large disruption—is seen as a manageable premium compared with the cost of substantive concessions. States and private security firms use AI‑driven surveillance, facial‑recognition, and social‑media monitoring to identify and preempt organising, while algorithmic news filters and targeted content platforms dilute or distort the narrative. The result is that resistance is often localised, contained, and chronically out‑paced by capital’s ability to adjust.

A third reason is the sheer weight of everyday survival. For much of the managed majority, the immediate pressure is not ideology but basic existence: rent, debt, food, healthcare, and the need to keep at least one income stream open. In such conditions, sustained collective action is expensive in time, risk, and energy. Many people cannot afford to lose gigs, cancel contracts, or anger opaque algorithmic systems that can quietly downgrade their visibility, income, or access. The AI‑driven world offers a “soft” coercion: not just overt repression, but the constant threat of exclusion from the platforms that mediate work, credit, and social life. This makes it easier for states and corporations to break or bleed resistance from the edges, often without triggering mass repression that might backfire politically.

Finally, resistance is sapped by ideological and cultural fragmentation. The AI‑driven era accentuates existing divides—national, ethnic, religious, and status‑based—while offering tailored entertainment, consumption, and identity constructs that absorb anger into stylised conflict rather than collective organisation. Some strands of dissent are channelled into niche, symbolic battles that feel radical but have little real leverage over capital flows or AI infrastructure. Others are co‑opted, monetised, or turned into marketable “brands” that sell rebellion without challenging the underlying system. The result is a landscape of constant noise, but little sustained, structural pressure on the elite.

By 2100, the pattern is clear: resistance is real, recurrent, and often intense, but it does not scale into the kind of counter‑power that can force a fundamental renegotiation of the AI‑capital order. The system is too diffuse, too fast, and too deeply integrated into daily life; the elite are too insulated and too well‑armed with data and capital. The failures of resistance are not evidence that people are docile; they are evidence that the architecture of power in the AI era is precisely designed to keep opposition from coalescing at the scale needed to change the fundamentals.

9. Humanity Fractures: Culture, Ideology, and Survival

9.1 Work, Identity, and the End of the Old Deal

Between 2060 and 2100, the psychological and cultural foundations of work melt down in tandem with the economic ones. For much of the 20th century, a tacit “old deal” shaped large parts of humanity: if you worked hard, played by the rules, and stayed adaptable, you could expect a reasonably stable job, gradual progress, and a chance at a modest but secure life. By the second half of the 21st century, that deal is broken; the link between work, identity, and social standing no longer holds for most people.

In the high‑tech core and its enclaves, work is still present, but it is often a minor piece of a much larger, algorithmically managed life. Many people spend their time supervising AI agents, managing interfaces, or handling the unpredictable, emotionally complex edges of service that machines cannot yet handle reliably. The old pride in “craft”, routine mastery, or long‑term employer loyalty fades, replaced by a fluid, gig‑like sensibility where people are treated as interchangeable nodes in an AI‑optimised network. The wage is no longer a symbol of belonging; it is a transactional fee for a specific task, often delivered through platforms that track performance, ratings, and “engagement” in granular detail.

For the managed majority and the permanent precariat, work becomes even more hollow. Jobs are temporary, low‑paid, and easily replaceable, often mediated by apps and algorithms that can demote, exclude, or re‑route workers without explanation. The sense that “my job defines me” weakens, not because people become more enlightened, but because the job itself is so unstable and fragile. Many turn to side hustles, multi‑platform labour, and informal arrangements, treating work as a means of survival rather than a source of identity or status. In some places, the dominant narrative among young people is not “what job will I have?”, but “how do I avoid being crushed by the system while scraping together enough to eat?”

The erosion of the old deal also undermines broader social scripts. In the 20th century, work structured time, provided a rhythm to life, and gave people a reason to leave the house, meet others, and build shared experiences. As AI takes over more routine tasks and firms run leaner, the number of jobs that offer that kind of structuring effect shrinks. People spend more time alone, in digital environments, or in fragmented, unstable roles that do not anchor them to a community. The family, the neighbourhood, and local institutions lose their previous connection to workplaces, and new forms of association emerge that are less rooted in stable employment.

At the same time, the rhetoric around work does not disappear; it merely becomes more cynical or contradictory. Leaders in free‑capital regimes still talk about “opportunity”, “adaptability”, and “upskilling”, but the underlying reality is that many people are being pushed into lower‑security, lower‑status survival modes. The ideology of “everyone can get ahead if they try hard enough” is maintained even as the structural conditions for doing so evaporate for large swathes of the population. In the rejection zones and subsistence enclaves, work takes on a different valence: it is often local, physical, and tied to immediate survival, but it is also less mediated by distant algorithms and global platforms. Here, work may not pay well, but it can still connect individuals to their community and land in ways that feel more tangible than the abstract, remote work of the AI‑driven core.

By 2100, the fracture is cultural as much as economic. For many people, the question is no longer “what is my job?”, but “how do I survive in a world where work is no longer a stable anchor?”. The old deal between work, identity, and security has been torn up, and no new, widely accepted alternative has taken its place. People are left searching for meaning in side‑projects, subcultures, online communities, religion, or local solidarities, but the absence of a shared work‑centred narrative deepens the sense of social fragmentation. Work does not vanish, but its role as the central organising principle of adult life shrinks, leaving a psychological and cultural void that is filled by whatever scraps of stability, dignity, and connection people can still find.

9.2 Ideologies of Elite Techno‑Utopia and Neo‑Traditionalism

By 2060–2100, worldviews are no longer neutral background noise; they are hard‑wired into the structure of the AI‑driven split. Two dominant ideological currents crystallise: one that justifies and glorifies the high‑tech core and off‑world elite, and another that romanticises, or at least legitimises, low‑tech, insular, or religious ways of life in the rejection zones. Neither is a pure fantasy; both are tools used to stabilise their respective domains and to delegitimise the other.

On one side is elite techno‑utopianism. This is the belief system of the AI‑driven core, off‑world cities, and their aligned institutions. In this view, human civilisation is on an inevitable trajectory of scientific and technological ascent: AI, robotics, biotechnology, and space colonisation are not just tools, but the next stage of evolution. Growth, efficiency, and control are framed as moral imperatives; the goal is to transcend biology, scarcity, and even Earth itself. Distress caused by automation is treated as a “transition cost” or “collateral damage” necessary to reach a higher plane of existence. In elite circles, the language of “innovation”, “scaling”, “optimisation”, and “long‑term survival” displaces older talk of justice, equality, and social contract. The cost of pursuing this vision—often in the hundreds of millions of CHF per year in R&D, infrastructure, and governance—is justified as an investment in “the future of humanity”, even though the future being secured is narrowly tailored to the interests of the powerful.

Techno‑utopian rhetoric also serves a political function. It frames the hollowing out of the middle class and the rise of the managed majority as a technical problem of re‑skilling, adaptation, and AI‑mediated welfare, rather than a structural conflict over power and ownership. It depicts the rejection zones and subsistence enclaves as “backward”, “romantic”, or even “dangerous” experiments that risk dragging the rest of humanity down. In practice, this ideology legitimises the concentration of AI‑driven wealth and infrastructure in the hands of a small group, while casting the rest of the world as a testing ground, a resource pool, or a managed periphery. Public education, media, and elite universities are saturated with this narrative, subtly training generations to see AI‑driven control and automation as a natural, even desirable, endpoint.

In contrast, neo‑traditionalism emerges across the rejection zones, retreat communities, and parts of the semi‑periphery. This is not a single, unified doctrine, but a family of belief systems that share a common core: that the AI‑driven, high‑tech path is fundamentally flawed, alienating, and ultimately unsustainable. In some places, neo‑traditionalism takes on religious overtones, framing AI and data‑hungry capitalism as “antihuman” or “demonic”, a system that reduces people to data points and commodified labour. In others, it is framed in ecological or civilisational terms: a warning that reliance on fragile, AI‑centric infrastructure and off‑world escape fantasies distracts from the urgent need to live within planetary limits, preserve local cultures, and rebuild community‑based livelihoods.

Neo‑traditionalist groups often idealise simplicity, self‑sufficiency, and local governance. They valorise manual work, face‑to‑face relationships, and traditional knowledge over digital mediation and algorithmic management. In subsistence enclaves and low‑tech regions, this ideology is less a programme for global change than a survival strategy: a way to make sense of hardship and to create a sense of legitimacy for cutting loose from the high‑tech core. Some neo‑traditionalist movements reject literacy with AI systems altogether, or at least restrict their use to minimal, non‑integrated forms, so that people are not psychologically or economically captured by platforms. Others embrace a selective, “hybrid” stance: using enough technology to survive, but refusing to surrender autonomy to invisible algorithms and distant data centres.

Both ideologies are self‑serving in practice. Techno‑utopianism protects the status of the AI‑driven elite and justifies the continued extraction of resources and compliance from the managed majority. Neo‑traditionalism legitimises the withdrawal of certain groups from the global system, providing a moral and spiritual vocabulary for those who cannot, or will not, compete on the core’s terms. The result is a deep cultural divide: the high‑tech core and its off‑world extensions view the world as a set of solvable technical problems to be optimised, while the neo‑traditional zones see the world as a moral and ecological order to be preserved or restored. Neither side understands the other on its own terms, and the clash between these worldviews hardens the split between the core, the periphery, and the retreat zones.

By 2100, the ideological map of humanity mirrors the economic and technological map. The core and its enclaves live inside a bubble of techno‑utopian confidence; the rejection zones and subsistence regions are wrapped in a legitimising shell of neo‑traditional belief. Neither ideology offers a way to bridge the gap; instead, both reinforce the conditions that keep the split of humanity intact.

9.3 How the Working Class Becomes Either Labour or Market

By 2060–2100, the working class in the AI‑driven world is no longer a coherent political or economic bloc; it is disaggregated into two broad roles, often overlapping in the same person: labour and market. In both cases, the working class exists to serve the high‑tech core and its AI‑centric ecosystems, not to shape them. The distinction is simple: as labour, people provide the unpredictable, low‑cost, or socially embedded work that cannot yet be fully automated; as market, they consume the cheap goods, services, and experiences that the AI‑driven economy produces. In neither role is the working class treated as a partner; at best, it is treated as a resource.

As labour, the working class is broken down into highly specific, task‑oriented segments. In the managed majority, this includes gig‑workers, AI‑supervised agents, low‑level caretakers, delivery drivers, platform‑mediated cleaners, and others whose labour is too contextual, emotional, or irregular to be entirely removed by machines. In rejection zones and subsistence enclaves, it includes manual farmers, small‑scale producers, local service providers, and informal‑sector workers who keep local economies ticking without deep integration into global AI‑driven systems. In both contexts, the pay is low, the hours uncertain, and the autonomy minimal. AI platforms and state‑linked systems track performance, ratings, and “engagement” in granular detail, turning human work into a set of quantifiable metrics that can be optimised, compressed, or simply replaced when the next automation step arrives. The cost of maintaining this labour pool—often in the tens of millions of CHF per year in social‑safety‑net subsidies and basic‑income‑style support—is framed as a prudential investment in stability, not as a gesture of solidarity.

As market, the working class becomes the ultimate consumer of AI‑driven abundance. In the core, this means cheap entertainment, algorithmically tailored content, low‑cost logistics, and mass‑produced goods whose prices are kept low by automated factories and logistics networks. In the periphery and rejection zones, it means affordable electronics, AI‑assisted services, and cheap imports that subsidise survival in fragile economies. The working class is also targeted, profiled, and manipulated: AI‑driven recommendation engines, price‑optimisation algorithms, and targeted advertising channels extract whatever disposable income exists, turning everyday life into a stream of micro‑transactions. The more people comply, the more data they generate, and the more tightly they are locked into the platforms and ecosystems that mediate their existence. Subscription models, micro‑credit, and “pay‑later” schemes deepen dependence, often at interest rates measured in the tens of percentage points, effectively transferring even more surplus from the working class to capital‑rich firms.

Over time, the line between labour and market blurs. A gig‑driver delivers packages that people buy on the same platform where they once worked; a part‑time data‑labeller trains the AI models that power the apps they use to consume cheap music and video; an informal‑sector vendor in a subsistence zone sells AI‑driven accessories to tourists or core‑linked enclaves. The working class is not just a supplier of human work; it is also a supplier of data, attention, and compliance, all of which fuel the AI systems that displace it. In this sense, the working class is not “replaced” so much as reframed: it is a reservoir of flexible inputs—labour, data, and consumption—that can be drawn upon, measured, and re‑engineered in real time. The cost of this reframing is high in terms of social cohesion and dignity, but the benefit for the AI‑driven elite is clear: a large, pliable population that can be managed at scale, without granting it real power.

By 2100, the working class is no longer a class in the classical sense of a collective force that can challenge the structures of power. It is a diffuse, multifaceted layer of humanity that is treated as either labour to be optimised or market to be extracted from. The split within the working class—between those who adapt to AI‑driven roles and those who fall into chronic under‑employment—becomes one of the defining fractures of the era. The old deal of work as a source of identity, security, and political leverage is gone; all that remains is a residual sense of being needed, but never in control. The working class is not abolished, but it is reduced to a functional role in the AI‑centric order: a vast, managed reservoir of people who provide what machines still cannot, and consume what machines can produce cheaply.

10. The Entrepreneurial Family in the AI Collapse

10.1 2026–2030: Build AI Leverage and Cash Flow

Between 2026 and 2030, the optimal strategy for an entrepreneurial family in the AI‑driven economy is to move fast, consolidate capital, and build AI‑leveraged cash‑flow assets before the first wave of clear job collapse makes competition and regulation more intense. This period is not about caution; it is about using the transition to lock in efficiency, ownership, and geographic flexibility. The working class, treated as either labour or market, provides both the workforce and the customer base for these moves.

The first priority is to automate your own income streams. By 2026, foundation‑model‑driven tools are already cutting through many office and service sectors, so any family‑owned business that relies on routine cognitive work—accounting, marketing, customer support, legal drafting, basic coding, and logistics planning—should be aggressively re‑engineered. A typical SME or mid‑size firm that applies AI‑assisted workflows in 2026–2027 can expect 20–35% efficiency gains in the first three years, measured in reduced headcount, faster turnaround, and lower error rates. That translates into cost savings of roughly 10–25% of annual wage and overhead expenditure, assuming staff costs are 50–70% of total operating costs. For a firm with 2 million CHF per year in labour costs, that means 200,000–500,000 CHF per year in savings, which can be reinvested into growth or acquisition.

At the same time, the family should focus on “picks and shovels” around AI rather than on generic services. The most attractive segments are niche B2B software, AI‑assisted SaaS, integration services, data‑labelling platforms, and domain‑specific copilots in areas like legal, healthcare, finance, logistics, and e‑commerce. The global AI software and infrastructure market is expected to grow at around 25–30% per year through 2030, with B2B applications and enterprise tools outpacing general consumer AI. A family that acquires or builds a specialised AI‑workflow tool in one of these verticals can plausibly reach 10–20 million CHF in annual revenue by 2030 if execution is adequate, commanding valuations in the 5–10x revenue range depending on margins, data moats, and switching costs.

Another key move is to acquire distressed or labour‑intensive SMEs and then strip‑out headcount. From 2026 onward, many small agencies, BPO‑style back‑offices, local marketing firms, and basic IT suppliers will be under pressure as clients demand AI‑driven cost savings. The family can buy these businesses at distressed prices—often 30–50% below their 2020–2022 valuations—then restructure them around AI‑assisted operations. For example, a small call‑centre or data‑processing firm that runs on 1.5 million CHF of annual labour can reduce staff by 30–50% over three years through AI‑driven triage, automation, and remote agents, generating 300,000–750,000 CHF in recurring savings. Those savings become margin, which can be reinvested or paid out as family cash flow.

Cash‑flow targets should be precise. A coherent family‑level ambition for 2026–2030 could be:

• Reduce direct labour costs in owned businesses by 25–40% through automation, adding 1.5–3.0 million CHF in annual net cash flow for a family with 6–10 million CHF of labour exposure by 2030.
  
• Grow AI‑centric or AI‑adjacent revenue streams to 30–40% of total group revenue, implying roughly 20–30% annual growth in these segments versus 5–10% in legacy lines.
  
• Increase return on capital employed in owned businesses from perhaps 10–15% to 20–25% by pushing out labour and pushing up margins, even if total headcount falls by 20–30%.

Geographic and legal engineering also matters. The family should secure second residencies, diversified banking relationships, and holding‑company structures in jurisdictions with stable rule‑of‑law and limited confiscatory risk. Moving 10–20% of liquid assets into multiple jurisdictions, worth several million CHF, spreads political and regulatory risk. At the same time, the family should begin acquiring or building digital and physical distribution channels—e‑commerce brands, local retail networks, media or audience platforms—so that products and services flow through channels they own, not just through platforms that can squeeze them later.

In fragile service‑export regions such as the Philippines or certain Indian BPO hubs, the family can also position itself to capture value as these sectors compress. Rather than relying on pure human labour, it should pivot toward owning AI‑driven managed‑service stacks, local data‑labelling and domain‑expertise teams, and contract‑fulfilment platforms that sit between global clients and the shrinking pool of human agents. Even if headcount in this sector declines by 20–30% from 2026 to 2030, the value of automation‑augmented, managed‑service contracts can grow at 10–15% per year, generating healthier margins than the old, labour‑heavy model.

Overall, the 2026–2030 window is about using AI to compress labour, expand automation‑driven cash flow, and build a compact, diversified asset base that can withstand the early phases of the AI‑driven collapse. The working class, treated as labour or market, provides both the workforce that can be downsized and the customer base for AI‑driven consumption; the entrepreneurial family’s role is to own the assets that mediate that relationship, not to be caught on the wrong side of the automation wave.

10.2 2030–2040: Consolidate Assets and Own Hard Realities

Between 2030 and 2040, the AI‑driven economy shifts from disruption to entrenchment. The entrepreneurial family’s priority is no longer just plugging in AI tools; it is to lock down ownership of real, non‑replicable assets—physical infrastructure, data, brands, and distribution networks—so that the family sits in the layer that controls the flows, rather than just riding them. The working class, treated as labour or market, becomes a more predictable input and customer base, and the family’s strategy should reflect that.

A core move is to acquire or build hard‑asset businesses that AI can supercharge but not easily replace: logistics hubs, warehousing, light manufacturing, farmland, energy assets, and real estate in integrated, AI‑friendly regions. Logistics and warehousing, in particular, are expected to grow at roughly 8–12% per year through 2040, as e‑commerce and AI‑driven supply chains push volume higher while squeezing labour costs. A family‑controlled logistics firm that owns last‑mile depots, regional warehouses, and a small private fleet can expect 15–25% margin improvement by 2040 through AI‑driven routing, predictive maintenance, and automated sorting, assuming an initial CAPEX envelope of 5–10 million CHF spread over the decade. For a logistics business with 20 million CHF of revenue, that implies an extra 3–5 million CHF of annual profit if the family reinvests savings into capacity and scale.

Farmland and food‑adjacent infrastructure are another target. Global food‑system automation—precision agriculture, AI‑driven irrigation, automated harvesting, and supply‑chain optimisation—will grow at about 10–15% per year through 2040, driven by rising population, climate risk, and the need to feed AI‑managed cities cheaply. A family that owns or consolidates several hundred hectares of arable land in a stable jurisdiction, with integrated water and energy infrastructure, can expect 12–18% annual growth in cash flow from 2030 to 2040 if the land is run with AI‑driven technologies and tied into regional distribution networks. On a 10 million CHF land base, that implies 1.2 — 1.8 million CHF in additional annual income by 2040, plus capital appreciation as food security assets become scarcer.

In parallel, the family should consolidate ownership of data and niche moats. Any business that collects, sanitises, and labels data in a specific vertical—industrial equipment, healthcare, construction, logistics, or finance—can build a defensible position that AI models rely on. Data‑centric B2B services are expected to grow at 20–25% per year through 2040, as firms increasingly license domain‑specific datasets rather than build them from scratch. A family that controls a specialised data pipeline—say, labelling condition‑reports from industrial machinery or structured clinical‑trial metadata—can expect 15–20 million CHF in annual revenue by 2040 if it captures just a few percent of a high‑value niche, with gross margins of 50–70% once the platform is built.

Distribution and customer ownership matter as much as production. The family should aim to control brands and channels that sit between AI‑driven manufacturers and the working‑class market. E‑commerce, local retail franchises, media or audience platforms, and subscription‑style services are all candidates. The global e‑commerce and AI‑assisted retail market is projected to grow at 15–20% per year through 2040, while AI‑driven media and content platforms grow at 12–18%. A family that controls a small but loyal audience—say, a regional media brand, a niche e‑commerce channel, or a subscription‑style service—can expect 10–15 million CHF in annual revenue by 2040, with 20–30% net margins if the business is run with AI‑driven personalisation and ad‑tech.

Investment in family human capital is another explicit target. By 2035, the family should have at least two or three members with deep operational expertise in AI, finance, and governance, positioned so they can sit on boards, manage complex portfolios, and negotiate with oligarchic or state‑level actors. The cost of this is modest—hundreds of thousands of CHF in bespoke education, coaching, and board placements—but the return is leverage: a family that can read balance sheets, AI business models, and regulatory risks will be able to move faster and more precisely than competitors who remain stuck in legacy thinking.

By 2040, the family’s asset mix should be roughly:

• 30–40% in hard real assets (land, logistics, energy, real estate) that AI can optimise but not fully replace.
  
• 20–30% in data‑centric and AI‑adjacent businesses that provide inputs to AI systems rather than being replaced by them.
  
• 20–30% in distribution and branding assets that connect AI‑driven supply to working‑class demand.
  
• 10–20% in liquid reserves and global multi‑jurisdictional holdings to maintain optionality and resilience.
  

The working class, in this scenario, is treated as a stable labour pool for lower‑skill roles and a reliable, low‑margin customer base for AI‑driven goods and services. The family’s role is to own the assets that mediate both relationships—physical infrastructure, data pipelines, and distribution channels—so that the family captures a growing share of the value that AI creates, even as labour’s share of national income continues to decline.

10.3 2040–2060: Choose: Core Integrator or Enclave Builder

By 2040–2060, the AI‑driven trajectory is clear enough that every entrepreneurial family effectively faces a binary strategic choice: either become a core integrator embedded in the high‑tech, AI‑centric economies and their global supply chains, or become an enclave builder constructing semi‑autonomous, AI‑assisted domains that operate on the edges of the system, managing their own labour and markets locally. The working class, in both paths, remains a combination of labour and customers; the family’s decision is about whether to be deeply inside the machine or running a managed subsystem outside it.

A core integrator strategy is about maximising leverage within the high‑tech core and its semi‑peripheral satellites. The family’s assets are squarely aligned with AI‑driven infrastructure, data flows, and global logistics. Target sectors include data‑centre operations, cloud‑adjacent services, AI‑enabled financial and logistics platforms, and industrial automation providers. The global data‑centre and AI infrastructure market is projected to grow at 15–20% per year through 2050, with AI‑driven financial and logistics platforms growing at 12–18%. A family that controls even a small but specialised slice of this—say, a regional data‑centre cluster, a niche logistics‑optimisation stack, or a domain‑specific AI‑finance platform—can plausibly reach 30–50 million CHF in annual revenue by 2050, with 25–35% net margins if the business is tightly integrated with global clients and AI vendors.

To execute this, the family should aim to own or co‑own:

• Critical nodes in infrastructure (data‑centre capacity, private networks, edge‑computing hubs) worth 20–40 million CHF of CAPEX by 2050, funded from earlier AI‑driven cash flows.
  
• Specialist software and integration firms that sit between global AI platforms and local industries, commanding 10–15% of a 100–150 million CHF market segment by 2050.
  
• Governance and influence roles—board seats, policy‑adjacent advisory positions, and participation in technical‑standards bodies—so the family can help shape the rules that favour AI‑integrated capital and manage the managed majority.
  

In this role, the family treats the working class primarily as a flexible labour pool and a mass market for AI‑driven services. AI‑assisted platforms, gig‑work marketplaces, and algorithmic media ensure that the family’s businesses can keep labour costs low and margins high, while the family’s assets sit in the layer that controls data, infrastructure, and distribution.

An enclave builder strategy, in contrast, is about constructing semi‑autonomous domains—rural or ex‑urban bases, island‑like developments, or clustered urban enclaves—that are deeply integrated with AI but deliberately insulated from the volatility of global politics and finance. The goal is not to overthrow the core, but to create a stable, high‑tech micro‑society that can survive regardless of what happens to the managed majority. AI is used to run energy, water, agriculture, and internal logistics with minimal outside dependence, but the family keeps control over the code, hardware, and governance.

Enclave builders target:

• Large, self‑contained land holdings with integrated renewables, storage, and water‑management infrastructure, often 100–500 hectares in size, requiring 10–30 million CHF of CAPEX over 2040–2060.
  
• On‑site AI‑driven farms, light manufacturing, and service platforms that can supply basic needs for a few hundred to a few thousand people, with projected 8–12% annual growth in internal cash flow from 2040 to 2060 as technology improves.
  
• Hybrid governance structures that blend family ownership, cooperative‑style participation, and limited local autonomy, so the enclave is politically stable but economically self‑sufficient.
  

In this role, the working class is treated as a mix of local labour and nearby market. The enclave will need a small cadre of highly skilled workers and a larger layer of semi‑skilled or manual labour to run maintenance, security, and services, but the number of people needed is kept low by automation. The nearby external population can be a captive market for surplus food, energy, and services, bought at prices that keep them dependent and the enclave profitable. Local AI‑driven retail and service platforms can capture 10–20% of resident and nearby spending, generating 1–3 million CHF in annual revenue per 10,000 adjacent consumers by 2060.

By 2060, the family’s choice will be visible in its asset mix:

• Core integrators will have 60–70% of assets tied to AI‑centric infrastructure, data, and global platforms, with 20–30% in liquid, multi‑jurisdictional holdings.
  
• Enclave builders will have 40–60% of assets in land, local infrastructure, and on‑site production, with 20–30% in AI‑driven software and 20–30% in liquid reserves and global options.
  

Whichever path is chosen, the entrepreneurial family’s logic is the same: own the assets that AI makes more valuable, and treat the working class as the layer that is managed, not that manages. The difference is whether the family swims inside the core’s current or builds its own AI‑assisted oxbow, and the consequences for security, influence, and long‑term survival are profound.

10.4 2060–2100: Earth Bases, Off‑World Options, and Captive Markets

Between 2060 and 2100, the entrepreneurial family that has navigated the earlier phases of the AI‑driven collapse has a clear next move: secure reinforced earth bases, position for limited off‑world options, and lock in captive markets tied to the managed majority. The working class is by now a stable, low‑margin, but highly predictable input of labour and consumption, and the family’s role is to own the assets that mediate those flows at scale.

Earth bases should be treated as long‑term security infrastructure. The family should consolidate a small portfolio of high‑quality, hard‑to‑confiscate land and infrastructure bundles—several large rural or semi‑rural sites with integrated water, renewable energy, and basic processing capacity. Each site might require 10–25 million CHF of cumulative CAPEX between 2060 and 2100, depending on size and technology, but the return is resilience: these bases can buffer against political shocks, supply‑chain disruptions, and local unrest. In a world where AI‑driven global systems stay fragile at the human level, the family can use AI‑driven agriculture, light manufacturing, and local logistics to supply 50–80% of each base’s own needs, keeping the remaining gaps small and manageable. With 3–5 such bases, the family can expect 15–25 million CHF in consolidated annual internal cash flow from 2080 onward, primarily from food, energy, and processed goods used by the enclave and its immediate environs.

Off‑world options are more speculative but not irrelevant. By 2060, orbital and lunar habitats are already running at industrial‑scale financial levels, with annual budgets for major consortia often in the hundreds of millions of CHF. Mars settlements follow a similar trajectory, but with slower growth and much higher risk. The family should not try to build its own off‑world city; instead, it should position itself as a co‑investor, infrastructure provider, or service‑contractor to existing ventures. Direct equity stakes in space‑related firms, investments in advanced materials and robotics, and equity or royalty participation in closed‑loop life‑support and energy systems can all generate 5–15% of the family’s total portfolio returns by 2100, assuming the space sector grows at 10–12% per year. The cost of this exposure—5–10 million CHF in cumulative investment between 2060 and 2100—is modest relative to the potential upside, and the main benefit is optionality: a foothold in the only human domains that are, in principle, insulated from Earth‑level social collapse, even if those domains are small and tightly controlled.

Captive markets are where the family’s leverage is most explicit. The working class, now firmly entrenched as either labour or market, provides a large, predictable, and politically constrained customer base for AI‑driven consumption. The family should own or tightly align with AI‑driven retail, media, finance, and basic‑service platforms that sit between the global supply chain and the local consumer. The global AI‑driven consumer‑facing services market is projected to grow at 10–15% per year through 2100, driven by cheap goods, targeted media, and algorithmic credit. A family‑controlled or family‑aligned platform that captures even 1–2% of a 100–150 million CHF regional market can generate 1–3 million CHF in annual revenue by 2080, with 20–30% net margins if the business is run with AI‑driven personalisation and churn‑optimisation. Multiply that across several jurisdictions, and the family can expect 10–20 million CHF in annual platform income by 2100.

At the same time, the family should treat the local labour pool as a managed asset. Gig‑work platforms, low‑cost AI‑assisted service providers, and AI‑driven security and logistics networks can be used to keep labour costs low and flexible, while the family’s owned assets—warehouses, data centres, local hubs, and land—provide the physical backbone. The working class is not a partner in this arrangement; it is a resource whose labour and data feed the AI systems that the family controls, and whose consumption flows through the channels the family owns.

By 2100, the entrepreneurial family that has followed this path will be deeply embedded in the AI‑capital order. It will own earth bases that provide security and self‑sufficiency, partial exposure to off‑world ventures that offer long‑term optionality, and a portfolio of AI‑driven captive markets that extract value from the managed majority. The working class will be treated as a layer that is managed, not that manages; the family’s role will be to own the infrastructure, the data, and the platforms that keep the system running, and to sit in a position where it can survive regardless of the broader social and political turbulence that surrounds it.

10.5 2060–2100: Diversified Bases in Human Needs and Infrastructure (Including AI)

An entrepreneurial family that wants to preserve some moral distance from the debasement of the working class will not rely solely on gambling, predatory finance, or addiction‑driven industries, even if those sectors offer strong, AI‑driven cash flows. Instead, it can anchor its strategy in base human needs and essential infrastructure, treating AI as a force that they must work with, not one they must directly exploit for moral harm. The family’s goal is not to “resist” AI, but to ensure that it thrives on AI‑driven growth while staying in sectors that are socially necessary or at least neutral.

Among the clear candidates are food, basic utilities, housing, healthcare and education infrastructure, and cultural staples such as alcohol and fermented drinks, which respond to deep, stable human drives rather than just fads. These are not “zero‑cost” industries and will not shrink to zero in profits, even as AI drops costs in many adjacent areas. On the contrary, the constraint is often land, regulation, quality, and taste, not the abstract cost of computation. AI can optimise farming, logistics, and supply chains, but it cannot replace soil, climate, tradition, or local trust overnight. The family should therefore treat food systems, beverage production, and secure housing as long‑term, AI‑assisted, but still profitable, legs of the portfolio.

For example, global food and beverage markets are projected to grow at 4–7% per year through 2100, driven by population, urbanisation, and the need for stable, predictable supply. A family that owns AI‑driven farms, distribution networks, and branded food or drink products—especially in resilient, climate‑adapted forms—can expect 10–15% annual growth in real cash flow from 2060 onward, assuming moderate price inflation and efficiency gains. The working class remains the core market, but the family’s role is to supply a necessity rather than to manipulate addiction or debt.

Alcohol and similar cultural staples are more morally ambiguous, but they are not inherently evil. People have used fermented and distilled products for thousands of years, often in socially and ritually embedded ways. The family could choose to stay out of mass‑addiction‑oriented products and instead focus on high‑quality, small‑scale, or culturally rooted beverage production—craft spirits, wines, beers, or herbal preparations—sold at comfortable margins to a stable, non‑impoverished customer base. This is not a “debasement” business in the sense of forcing dependence; it is an enhancement of a longstanding human ritual, delivered with higher quality and lower environmental cost.

Gambling and speculative finance are harder to reconcile with a desire to avoid debasement. These sectors will grow strongly under AI—algorithmic trading, AI‑driven betting engines, and highly targeted, real‑time gaming and prediction platforms can expand rapidly, perhaps 10–15% per year—but they also tend to concentrate harm among the working class. A morally cautious family might treat these as short‑term, high‑risk, or strictly limited exposure, if they participate at all, and avoid building the family’s identity or long‑term stability around them.

More acceptable from a moral‑prudence standpoint are utilities, energy, and digital infrastructure. These sectors are not “zero‑cost” and will not collapse in profitability, even as AI and automation cut operating expenses. AI will drive down marginal costs in data transmission and processing, but the underlying need for copper, fibre, electricity, cooling, and secure physical sites will not disappear. Mining of critical metals, data‑centre operations, and network‑related hardware and maintenance are all likely to continue growing at 8–12% per year through 2100, driven by AI itself, mobile ubiquitous computing, and off‑world ambitions. The family can own mining equity, infrastructure funds, or regional network‑and‑energy providers, earning steady, AI‑assisted cash flows without directly pushing others into demeaning forms of labour or gambling‑driven exhaustion.

Computing hardware and advanced materials also sit in a similar zone. The production of chips, advanced sensors, and high‑precision components is capital‑intensive, R&D heavy, and hard to automate fully; it is a domain where AI can help with design and testing, but the physical and material constraints still create durable profit margins. The global semiconductor and advanced‑materials market is expected to grow at 10–12% per year, and participation through manufacturing, fabrication, or supplies can yield 15–25% net margins for well‑run operations. The family does not have to “own” AI directly; it can own the physical and material layer that AI sits on top of.

Where the “everything goes to zero” logic bites hardest is in generic software and commoditised digital services that become indistinguishable and fungible. If AI can produce endless, nearly free copies of routine software, templates, and content, then the economic value of those layers will collapse, even as the value of data, interfaces, and distribution remains high. The family should therefore avoid building its core on such vanishing‑marginal‑cost activities and instead focus on data ownership, vertical integration, and physical‑digital bridges—energy, land, logistics, materials, and branded infrastructure—where AI reduces costs but does not eliminate scarcity.

In this framework, the entrepreneurial family accepts AI as an irreversible, fast‑moving force, but does not hitch its moral identity to the worst forms of human exploitation. It aligns with AI to grow food, energy, shelter, and essential infrastructure, and to participate in the physical and data‑related layers that underpin the system, but it keeps its distance from gambling‑driven addiction, hyper‑predatory finance, and the most manipulative forms of mass‑attention capture. The working class, treated as labour and customers, becomes a stable, predictable domain of demand, but the family’s profits come from serving real needs and building real assets, not from the systematic debasement of others.

10.6 Self‑Hosted AI as a Strategic Counterweight: Escape from AI‑Oligarch Rent

As AI‑driven automation, cloud‑scale platforms, and AI‑oligarch capital concentrate power and rent extraction, a parallel layer of self‑hosted, open‑source AI is emerging that can give entrepreneurial families a meaningful degree of sovereign leverage rather than total dependence on big‑cloud AI platforms. Tools like OpenClaw and its successors—open‑source, free, community‑supported AI agents that can reason, act, and integrate with external tools—show what is possible when the core intelligence stack is not owned by a handful of AI‑oligarch corporations, but by those who deploy it. This is not a hobby‑layer; it is a strategic option for the family that wants to avoid being fully rent‑exposed to the dominant AI‑platforms.

10.6.1 What self‑hosted AI gives you

Self‑hosted AI means running AI models and agents on infrastructure that the family controls or directly sponsors, rather than relying on per‑token, closed‑API services from the big AI‑cloud providers. This has three concrete effects:

• Ownership of data and workflows: The family’s sensitive data—legal, financial, medical, educational, estate‑related—does not have to leave the local stack. AI‑driven analysis happens inside the family’s own environment, reducing surveillance and extraction risk.
  
• Lower long‑term cost leverage: Instead of paying marginal‑per‑token rents to AI‑oligarch platforms, the family can amortise AI use over its own infrastructure and agents, especially as open‑source models grow in capability and efficiency.
  
• Resilience against platform‑risk: If global AI‑clouds face regulation, geopolitical fracture, sanctions, or sudden pricing shifts, a self‑hosted AI layer can still run the core workflows that the family depends on.
  

In practical terms, self‑hosted AI is a “shadow AI stack” that runs alongside the public AI‑platforms rather than replacing them outright. It is insurance against being fully dependent on a single, concentrated, politically‑exposed AI‑infrastructure.

10.6.2 OpenClaw and its successors as a case study

OpenClaw and its successors represent a new class of open‑source, agentic AI systems that can:

• Retrieve context from the web and documents,
  
• Plan, break down tasks, and take real‑world actions,
  
• Integrate with tools, APIs, and workflows,
  
• Learn from feedback and adapt to specific domains.
  

For entrepreneurial families, these systems are not just productivity tools; they are building blocks for:

• Custom research and due‑diligence agents that scan legal, regulatory, and market data for the family’s portfolio without exposing everything to a public AI‑cloud.
  
• Concierge‑style AI agents for education, health‑care coordination, and lifestyle management that operate inside the family’s own stack, using private data and private rules.
  
• Local‑stack AI assistants for estate management, energy‑grid optimisation, food‑supply logistics, and property‑maintenance, all running on infrastructure the family already owns or can sponsor.
  

Because OpenClaw‑type systems are open‑source, community‑supported, and free to run on self‑hosted infrastructure, they lower the barrier to building AI‑leveraged services that do not depend entirely on AI‑oligarch rent.

10.6.3 Economic and entrepreneurial opportunities

Self‑hosted AI is not just a defensive play; it is a new entrepreneurial layer. The family can position itself at several levels:

• Infrastructure and hosting: Invest in or own the compute and storage stacks that run self‑hosted AI agents for a network of aligned businesses, enclaves, or trusted partners. This creates a private AI‑cloud that is not exposed to the same regulatory and geopolitical shocks as the big public AI‑clouds.
  
• Agent‑tooling and niche SaaS: Back or build specialised AI‑agent tool‑kits that let small‑scale businesses, subsistence enclaves, or local‑level enterprises plug into AI‑assisted workflows without being fully extracted by big‑platform per‑token pricing.
  
• Open‑source roadmap sponsorship: Support or fund the evolution of open‑source AI‑agent ecosystems, giving the family insider‑level influence over the direction of those tools while keeping them outside the direct control of AI‑oligarch corporations.
  

For the family, this is a way to participate in AI‑driven value creation without being reduced to a pure consumer of AI‑platform rent. It is a “non‑oligarchic AI” layer that can coexist with, but not depend on, the big‑cloud platforms.

10.6.4 Strategic positioning for the family

Given the trajectory of AI‑driven consolidation, the family’s long‑run strategic interest is not to reject AI, but to own the rails on which AI runs, while cultivating alternative stacks that reduce dependence on the dominant platforms. Self‑hosted AI is a critical lever in that strategy.

To integrate this into the family’s broader AI‑leveraging plan:

• Treat self‑hosted AI as a complement, not a replacement: Use big‑cloud AI for public‑facing scale where appropriate, but run sensitive, high‑leverage, or mission‑critical workflows on self‑hosted stacks.
  
• Plan for multi‑stack resilience: Assume that at least one layer of AI‑driven decision‑making, research, and automation is running on self‑hosted, open‑source AI agents, so the family is not crippled if the dominant AI‑platforms fracture, spike in price, or fragment geopolitically.
  
• Use self‑hosted AI as a control point: In AI‑ubiquitous worlds, the family that owns its own AI‑stack owns the final say over how AI is used inside its own ecosystem—data, workflows, and rules.
  

In short, self‑hosted AI is not a marginal tech‑geek side project; it is a strategic counterweight to AI‑oligarch rent. For entrepreneurial families, cultivating sovereign AI leverage—through tools like OpenClaw and its successors—means holding one of the few genuine options for escape from the pure‑rent economy of big‑cloud AI platforms, while still harnessing the full power of AI to secure, reposition, and grow wealth across the 21st century.

11. Three Pathways to 2100: Unlikely, Most Likely, and Most Hopeful AI Futures

If the argument so far assumes a relatively smooth, multi‑decadal slide from 2026 to 2100, that may be too optimistic. The real risk is that AI does not just erode work: it collapses demand at shocking speed, producing a violent lull after the initial “cheap goods” boom. The pattern resembles what happened when China flooded the USA with cheap manufactured goods: in the short run, consumers enjoyed ultra‑low prices and more gadgets; in the long run, entire local industries hollowed out, skills vanished, and regional economies collapsed, leaving a large population with little income to buy all those cheap goods anymore. AI can repeat and magnify that dynamic 100‑fold or more because it hits cognition, not just manufacturing.

Below, three scenarios are laid out in rough quantitative terms, with emphasis on demand, labour, and income. These are stylised, not crystal‑ball forecasts, but they provide a framework to stress‑test the earlier timeline.

11.1 Scenario 1 (Unlikely): AI Is Shackled, Income Preserved, Transition Controlled

In this scenario, AI is deliberately slowed, heavily regulated, and integrated into a managed transition that preserves most income and employment. The “free‑capital” logic is interrupted by strong state and social‑contract policies: wage floors, re‑education guarantees, capital‑tax regimes, and AI‑specific redistribution schemes mean that productivity gains are shared rather than hoarded.

Key assumptions (2026–2100, stylised):

• Global labour‑market exposure to AI‑driven automation: 40–50% of tasks by 2040, rising to 60–70% by 2060.
  
• Annual job‑displacement shock: 1–2% of the global workforce per year “off‑boarded” from core roles, offset by 0.8–1.5% in new, AI‑adjacent roles.
  
• Aggregate wage share: stabilises at 55–60% of GDP (versus 60–65% pre‑2025), not falling to 40–45% as in freer‑capital regimes.
  
• Household consumption growth: 2–3% per year real, driven by AI‑driven price drops but sustained by income‑preservation schemes (AI‑dividends, job‑guarantee‑style buffers).
  

In monetary terms, imagine a global GDP of roughly 150 trillion CHF by 2040 (about 1.5x 2025 levels in real terms). In this controlled transition, an extra 15–20 trillion CHF of productivity is generated every 10 years, with perhaps 8–10 trillion CHF captured as corporate profit and 6–10 trillion CHF pushed back into wages, benefits, and public consumption. That keeps consumer demand buoyant even as AI slashes unit costs, preventing the “cheap goods but no buyers” crash.

The downside is political difficulty: this scenario requires a level of global coordination and national discipline that is historically rare. It also burns capital: AI‑tax rates, compulsory retraining funds, and public‑AI‑infrastructure investments could consume 1–2% of global GDP per year, or roughly 1.5–3 trillion CHF per decade. The result is a “managed high‑tech society” in which the AI‑driven elite are tamed, the middle class is preserved, and collapse is narrowly avoided—but this scenario is unlikely unless major shocks force a coordinated response.

11.2 Scenario 2 (Most Likely): AI Dominates, Jobs Evaporate, Societies Collapse

In this more realistic, “no‑guardrails” path, AI scales rapidly, capital is unchained, and labour is treated as an expendable cost. The initial effect is a consumption surge driven by ultra‑cheap goods and services; the second effect is a collapse of demand as millions of workers lose income and cannot afford even the cheap stuff.

Quantitative stylisation (2026–2040, sharp‑turn variant of the earlier timeline):

• AI adoption speed: 30–40% of tasks in advanced economies automatable by 2030, 50–60% by 2035, 70–80% by 2040. In the Global South, 20–30% of tasks automated by 2035; 40–50% by 2040.
  
• Job‑displacement shock: 3–5% of the global workforce thrown into immediate unemployment or under‑employment per year, with only 0.5–1.5% absorbed into net new AI‑adjacent roles. Within 15 years, 30–50% of existing jobs show “net loss” after replacement, not just disruption.
  
• Wage share of GDP: falls from 60–65% pre‑2025 to 40–45% by 2040 in core AI‑capable economies.
  

In financial terms, imagine a global GDP of 130–140 trillion CHF by 2040. AI could add 10–15 trillion CHF of productivity per 5 years, but most of that flows to capital, dividends, and AI‑related capital spending, not to labour. If wages fall 15–25% real terms over 15 years while the number of workers in “good” jobs shrinks, the result is a dramatic drop in effective demand.

Consider a stylised medium‑income country exporting services (e.g., the Philippines):

• 2025 baseline: 1.5% of GDP in BPO and IT services, 10–15% of formal urban employment in service‑export hubs.
  
• 2026–2035 AI impact: AI‑driven automation cuts 30–50% of routine service‑export tasks, even as AI‑assisted “premium” roles grow modestly.
  
• Net result: 20–30% decline in formal service‑export jobs over 10 years, with wages for remaining workers depressed by 10–20%.
  
• Consumption impact: a 5–10% drop in urban middle‑class disposable income, even though AI‑driven imports of gadgets and digital goods are 15–20% cheaper. The “cheap‑goods” boom cannot offset the income shock; the domestic market contracts.
  

Globally, the pattern is similar. The first 10–15 years of AI‑led automation look like a consumption bonanza: electronics, media, transport, and basic services get cheaper, and many people upgrade phones, subscriptions, and experiences. But as AI accelerates, large segments of the population lose stable income, and the “price‑level effect” of cheaper goods is swallowed by the “income‑level effect” of job loss. The point of maximum demand destruction comes when AI‑driven productivity peaks and labour demand troughs, likely in the 2035–2045 window.

In this scenario, the timeline from 2026–2030 to 2035–2045 is compressed. Instead of a slow 70‑year drift, you get:

• 2026–2030: AI agents eat 20–30% of routine white‑collar tasks, with 5–10% of jobs visibly cut.
  
• 2030–2035: AI expands into creative, technical, and customer‑service domains; 15–25% of jobs show net loss, and wage growth stalls or reverses.
  
• 2035–2040: AI‑driven automation hits 40–60% of the global workforce in advanced and semi‑advanced economies; real wages fall 10–20% in many regions, and large service‑export economies suffer 20–30% contractions in formal‑employment sectors.
  
• 2040–2045: demand destruction accelerates; “cheap goods but no buyers” lull sets in, with AI‑driven price levels at or below 2025, but aggregate effective demand 15–25% lower than it would have been without AI‑driven labour shock.
  

In this “most likely” case, the 2040s and 2050s are not a continuation of growth; they are a period of negotiated austerity, social unrest, and the consolidation of AI‑stabilised oligarchies and rejection zones. The “slow” 2100‑oriented timeline is superseded by a sharper, 2030–2050 crisis phase, after which the world reorganises around the post‑AI order.

11.3 Scenario 3 (Most Hopeful): A Split World, with AI‑Utopias and Subsistence Refuges

The third, more hopeful but still non‑utopian, scenario assumes a world where AI is not globally shackled, but some regions choose to go all‑in, while others retreat or impose strict limits. The result is a multi‑speed planet: a few AI‑utopias that become incredibly productive and stable, mixed‑AI countries caught in angst and conflict, and subsistence or low‑tech enclaves.

Quantified stylisation:

• AI‑utopia bloc: 10–15% of the global population, 25–30% of global GDP, with 80–90% of tasks AI‑driven by 2040 and 90–95% by 2060. Wage share falls to 35–40%, but AI‑dividends, basic‑income schemes, and high asset‑ownership among elites keep consumption buoyant. Real GDP growth 3–4% per year, driven by AI‑enabled innovation and off‑world extensions.
  
• Mixed‑AI countries: 40–50% of population, 40–50% of GDP, with 50–70% of tasks AI‑driven by 2040, 70–85% by 2060. Wage share 40–50%; political volatility high as AI drives job loss while social‑safety nets are patchy. Real GDP growth 1–2% per year, with frequent recessions.
  
• Subsistence and low‑tech enclaves: 35–40% of population, 25–30% of GDP, with 10–20% of tasks AI‑driven by 2040, 20–35% by 2060. Wage share 50–60%, but output per capita low. Real GDP growth near 0–1%, with periodic contractions. These are often rural or semi‑urban regions, plus intentional low‑tech communities.
  

In this scenario, the low‑AI‑adapted populations migrate outward from AI‑utopias, but not all of them can be absorbed into mixed‑AI or subsistence zones. The AI‑utopia bloc becomes a highly selective, AI‑centric elite enclave, with cities optimised for AI‑driven living and off‑world ambitions. The mixed‑AI countries are in constant political churn, with AI‑driven inequality fuelling conflict, while subsistence zones grow as people retreat to low‑tech lifestyles, subsistence farming, and local economies. The “end of normal work” is not a global collapse, but a disaggregated one: some regions live the AI‑dreams, others the AI‑nightmares, and others the quiet, low‑tech half‑life.

11.4 Wrapping up

In summary, the earlier timeline may be too gentle. The dramatic drop in demand caused by AI‑driven job loss could compress the shock phase into 2030–2045, with a sharp lull in consumption and social instability as the “cheap‑goods boom” gives way to the “no‑income boom”. The most likely future is one of AI‑dominated collapse for many, with only a handful of regions managing to stabilise into AI‑utopias or mixed‑AI states. The entrepreneurial family’s role is to position itself in the AI‑utopia or mixed‑AI blocs, owning the assets that AI enhances, while recognising that the majority of humanity will be managed, not empowered.

12. The Human Consequences: From Isolation and Despair to Demographic Collapse

The previous sections map the economic and political architecture of the AI‑driven world, but they leave out what that architecture does to the human psyche, the body, and the basic drive to reproduce. The “end of work” is not only a change in jobs; it is a transformation of everyday life at the most intimate level. In the absence of meaning, connection, and stable relationships, the result is not just poverty; it is a slow erosion of the human project itself. The working class, treated as labour or market, becomes a managed majority who are tracked, nudged, and exploited without being nourished. In that vacuum, isolation, depression, and the collapse of the sexual and reproductive drive take hold, and the human species enters a long‑term demographic suicide that cannot be fixed by policy, moralising, or “free‑time utopianism”.

12.1 Isolation and the erosion of community

AI‑driven automation and precarious work dismantle the social infrastructure of work. Off‑site offices, platform‑mediated gigs, and remote monitoring strip away the cafeterias, after‑work hangouts, and shared routines that once linked people. The worker does not talk to a boss; they talk to an algorithm. The manager does not supervise a team; they watch dashboards. The result is a society of people in the same city but living in separate, screen‑mediated bubbles, with hundreds of “friends” and no one to turn to in a crisis. Locally rooted institutions—churches, clubs, unions, neighbourhood groups—wither, as people move for gigs, migrate for work, or fall into poverty. The AI‑driven economy does not replace these institutions; it hollows them out. The working class is left in a social vacuum, where the only constant is the algorithmic presence of platforms that demand work, track behaviour, and extract data.

12.2 Depression, despair, and the loss of meaning

The AI‑driven economy is optimised for efficiency, not meaning. Work is stripped of narrative, ritual, and status; the worker is no longer a “craftsman” or “caregiver” but a “task‑executor” or “data‑provider”. The AI oversees everything, is opaque, and indifferent. The worker is rewarded not for skill or empathy, but for speed, accuracy, and compliance. That lack of meaning is corrosive. Humans derive identity from work, relationships, and contribution; under AI, those sources are eroded. The worker cannot take pride in their job, build relationships, or feel a sense of impact. The AI‑driven platforms compound the problem: they nudge, prod, and manipulate the worker into working more, spending more, and staying online, creating a loop of stimulation and fatigue. The result is a wave of depression, anxiety, and despair that is not a side effect; it is a core feature of the system.

12.3 The collapse of sex and sexual drive

Sex in the 20th century was a social and emotional act as much as a biological one; in the AI‑driven era it becomes increasingly mediated by technology, data, and distraction. The working class is already isolated, exhausted, and over‑stimulated. The worker who spends the day in front of a screen, in a gig, or under AI‑supervision is not in a state for intimacy or bonding. Dating apps and social platforms are designed to keep swiping, matching, and chatting, not to sustain deep relationships. The sexual drive depends on connection, trust, and intimacy; in the AI‑driven world those are in short supply. The result is fewer relationships, less sex, and a growing sense that sex is not pleasure but scarcity. The worker is not in a position to be sexual; they are in a position to be compliant.

12.4 No f#cking, No babies, No humans: The demographic suicide

Human reproduction is not a mechanical function; it is a social fact. People have babies when they are in stable relationships, feel secure, and expect a future. In the AI‑driven world, those conditions dissolve. The working class lives in isolation, precariousness, and despair. They are not in families; they are in gigs. They are not raising children; they are surviving. They are not investing in the future; they are paying bills. The result is a collapse of fertility: the total fertility rate (TFR) drops below replacement, children become rare, and the demographic line of the species begins to shrink. That shrinkage is not a temporary dip; it is a long‑term structural trend once AI remakes work, relationships, and identity. The AI‑driven economy is not a life machine; it is a machine that consumes human labour, attention, and ultimately the next generation.

12.5 The plain‑language conclusion: why this leads to a calculable end date

In a closed, modern society, once TFR stays permanently below replacement, the human population is on a mathematically predictable path to functional extinction. The only way to stop that path is to push TFR back above about 2.1 for several generations; if AI‑driven isolation, depression, and precarity push TFR down and keep it there, the “end of humanity” in many regions becomes a date you can write down, not a vague someday.

Assume a stylised AI‑core bloc of 1.5 billion adults in 2025, with no migration, constant 25‑year generations, and TFR fixed for the long term:

• At TFR = 1.5, each generation is 75% the size of the last.
  Rough trajectory:
  
  • 2025: 1.5 billion adults.
    
  • 2100: about 0.6 billion adults.
    
  • 2150: about 350 million adults.
    
  • 2200: about 200 million adults.
    
  • 2250: about 150 million adults.
    
    By 2400–2450, the bloc is effectively too small to sustain complex civilisation without external labour or AI‑driven systems that themselves depend on human capital. 2450 – only 10 million humans. This is a working‑class driven demographic collapse: low‑fertility, isolation, and stress hollow out the reproduction base.
    
• At TFR = 1.0, each generation is 50% the size of the last.
  
  • 2025: 1.5 billion adults.
    
  • 2100: about 190 million adults.
    
  • 2150: about 47 million adults.
    
    By 2250–2300, the bloc is already a tiny rump (by 2300 only 750,000 humans remain); the line of reproduction is effectively broken, and the human‑scale economy collapses with it.
    
• At TFR = 0.5, each generation is 25% the size of the last.
  
  • 2025: 1.5 billion adults.
    
  • 2100: about 20 million adults.
    
  • 2125: about 5 million adults.
    
    By 2200–2250, the population is so small that sustaining anything recognisable as a functioning AI‑driven economy or even local infrastructure is impossible. Less than 150,000 humans by 2250. In this scenario, the “end of humanity” in the bloc arrives in early‑ to mid‑22nd century.

12.6 Tabulating the Results – Demographic collapse “end‑decade” by TFR (1.5‑billion bloc)

Here is a reference table all starting from 1.5 billion adults in 2025, with a 25‑year generational step and the “end‑decade” defined as when the bloc falls below 1 million adults (or, in the TFR = 0 case, when the last generation dies out).

TFR	Approx. decade when population falls below 1 million adults	Human‑line meaning
2.1	Never	Population roughly stable; no demographic collapse.
1.5	2575	By late 26th century the bloc is too small to sustain human‑scale civilisation.
1.0	2300	By late 23rd century the human line is effectively gone in this bloc.
0.5	2150	By mid‑22nd century the bloc is below 1M adults; human‑scale reproduction ends.
0.0	2100	No babies born; last generation dies out; the human line in the bloc terminates.

In each case the drop in demand is the final blow: fewer people mean fewer consumers, fewer borrowers, fewer renters, fewer workers to fund AI‑driven infrastructures. The AI‑driven economy was built on growth; when headcount and demand contract at TFR‑driven speed, the system does not adapt; it collapses with the species. The end of humanity is not a side effect; it is the arithmetic consequence of TFR staying too low for too long.

13. Human‑Scale Sectors That Endure: Non‑AI Activities in an AI‑Dominated World

The AI‑driven world will not be “all AI business.” Alongside AI‑centric platforms, data‑centres, and algorithmic markets, large domains of economic and social life will remain stubbornly non‑AI‑centric, or will only use AI as a tool at the margins. These are the “human‑scale” sectors: survival infrastructure, care, craft, local subsistence, and conflict‑driven politics. Over the 2026–2100 period, their role will shift in clear, measurable ways. Some will contract as AI hollows out demand and labour, others will expand as AI‑driven societies become more fragile and dependent on physical, embodied, and local systems. For entrepreneurial families, understanding this map is essential: it shows where value will migrate when the illusion of AI‑totality breaks down.

13.1 Survival infrastructure: Food, water, energy, housing, and local repair

Survival infrastructure is the backbone of real‑world resilience. AI can optimise routing, pricing, and maintenance, but the core value is the physical asset itself and the labour that keeps it running.

• 2026–2050
  
  • AI‑driven automation and remote‑management reduce the labour intensity of large‑scale energy, grid, and logistics by 20–40% in high‑tech regions, while pushing wages for on‑the‑ground technicians and managers toward stagnation or slight decline.
    
  • Ownership of critical infrastructure (utilities, energy grids, cloud‑linked logistics hubs) grows 15–25% annually in AI‑enhanced, capital‑heavy segments.
    
  • In semi‑peripheral regions, local housing, water, and basic repair networks begin to gain relative importance as AI‑driven export‑service sectors face pressure; local‑infrastructure share of GDP rises 2–5 percentage points cumulatively.
    
• 2050–2075
  
  • In AI‑dominated cores, employment in core‑infrastructure sectors flattens or falls 5–10% as AI‑centric remote control and predictive maintenance spread, while AI‑linked margin capture (platforms, data, remote control) grows 40–60% cumulatively.
    
  • In low‑tech and subsistence‑leaning regions, local food, water, and housing contracts only 5–10% in absolute terms if global demand falls 10–20%, but their relative share of regional GDP climbs to 25–35%, as AI‑driven manufacturing and finance shrink.
    
  • Large‑scale AI‑centric projects (mega data‑centres, long‑haul logistics automation) rise, but increasingly depend on a thin layer of human engineers, emergency‑response crews, and security who cannot be fully replaced.
    
• 2075–2100
  
  • If TFR‑driven population decline pushes core regions toward rump‑size (~10–20 million adults from a former 1.5‑billion bloc), local survival infrastructure contracts 40–60% in absolute volume but remains the largest residual economy in that bloc, representing 40–60% of remaining GDP.
    
  • In smaller, fragmented enclaves, the “AI‑service overhang” collapses, and the core economy is survival‑infrastructure‑centric: housing, energy, water, and basic transport dominate the remaining balance sheet.
    

For families, the lesson is clear: own the bricks, not just the algorithms. AI‑enhanced infrastructure is valuable, but the physical backbone it rides on is the hard‑to‑replace asset.

13.2 Care and emotional labour: Health, education, elder‑care, and therapy

Care work is deeply human‑centric. AI can assist diagnostics, scheduling, and record‑keeping, but the core value lies in presence, trust, and embodied empathy. Over the next 74 years, this domain will not shrink; it will reconfigure.

• 2026–2050
  
  • AI‑driven diagnostics and administrative automation reduce the cognitive workload in health and education by 20–30%, but direct‑care roles (nurses, teachers, therapists, caregivers) grow 5–10% as societies age and mental‑health stress deepens.
    
  • AI‑driven “care‑adjacent” roles (platforms, remote monitoring, diagnostics) grow 20–30% cumulatively, but the human‑centric core of caregiving remains the core revenue and social value.
    
• 2050–2075
  
  • As TFR stays low and life expectancy remains high, the share of population over 65 rises from ~20% today to ~30–40% in many cores, doubling or tripling relative demand for direct care even as AI handles more paperwork and triage.
    
  • In high‑AI‑penetration regions, this means direct‑care employment grows 15–25%, while AI‑care‑adjacent roles grow 30–40%.
    
  • In low‑tech regions, local‑care systems (community clinics, informal elder‑care, small‑scale schools) grow 10–20% in relative importance, becoming the primary social‑safety net once AI‑centric welfare and finance hollow out.
    
• 2075–2100
  
  • In regions that have fallen toward rump‑size, local care systems shrink in absolute terms by 30–50% with population, but their share of remaining GDP and employment rises to 30–50%.
    
  • These sectors become the last remaining “human‑centric” bulwark: the core of what AI‑driven enclaves pay for, even as the producers of that care live increasingly in the managed precariat.
    

For families, the implication is simple: care‑centric assets are long‑duration options. They will not vanish with AI; they will become more critical as the world ages and fragments.

13.3 Local, low‑tech, and subsistence economies

As AI hollows out middle‑income work and global demand falls, large swaths of the world will fall back into local, low‑tech, or subsistence‑leaning economies. These are not “AI‑businesses”; they are survival‑centric, often under‑digitised economies that rely on land, skills, and local trust.

• 2026–2050
  
  • AI‑driven financialisation and platform‑mediated trade shrink the global value of low‑tech exports by 10–20% as demand in AI‑core regions stagnates, but local subsistence systems grow in relative importance within those regions.
    
  • The share of economic activity that is local, cash‑based, and non‑AI‑platform‑linked rises by 0–5 percentage points in many semi‑peripheral regions, as people shift toward self‑sufficiency.
    
• 2050–2075
  
  • If AI‑driven austerity and TFR‑driven decline deepen, the local‑low‑tech share of GDP in many regions grows 5–15 points, reaching 30–40% of regional activity.
    
  • In countries that actively de‑globalise or are expelled from AI‑centric trade networks, this share can rise 15–25 points, with local farming, repair, barter, and community‑services forming the new core economy.
    
• 2075–2100
  
  • In regions that have effectively disconnected from the AI‑core, local subsistence‑linked activity contracts 20–40% in absolute terms due to population decline, but grows in relative weight to 40–60% of remaining economic activity.
    
  • These regions become “resilience enclaves”: low‑tech, high‑trust, and AI‑light, but still fully human‑centric.
    

For families, this means owning land and local‑scale productive systems is a hedge against AI‑driven fragility. These domains are not glamorous, but they are sticky and hard to erase.

13.4 Craft, art, and embodied skill

Craft, artisanal production, music, and embodied skill will remain human‑centric. AI can generate designs, ideas, drafts, and synthetic performances, but the premium will sit on authenticity, finish, and the act of creation itself. Over time, this sector will not shrink; it will fragment and re‑stratify.

• 2026–2050
  
  • AI‑generated content and design undercut 10–20% of low‑ to mid‑tier creative work, pushing many freelancers into under‑paid, platform‑driven gigs.
    
  • High‑end craft, bespoke architecture, bespoke food, and live performance remain resilient or grow 0–5%, as the “human‑made, non‑AI‑generated” premium begins to price in.
    
• 2050–2075
  
  • As AI‑driven goods flood markets at near‑zero marginal cost, the value of human‑centric craft grows relative to mass‑commercial creative work.
    
  • High‑end craft and live‑performance sectors grow 10–20% in real‑value terms, even as the broader entertainment and marketing‑services sector shrinks 15–25% with AI‑driven deflation.
    
  • In AI‑core enclaves, “human‑authored” art, music, and bespoke experiences become a status good for the AI‑oligarchy, while the rest of the world consumes AI‑generated fodder.
    
• 2075–2100
  
  • In rump‑size regions, human‑centric craft and art contract 30–50% in absolute size with population, but represent 20–30% of remaining cultural and discretionary spending.
    
  • These activities become the last discretionary outlet for meaning, identity, and distinction in an AI‑hollowed world.
    

For families, the message is: invest in human‑centric craft and authenticity, not just AI‑generated content. These are the residual “luxuries” that survive deflation and collapse.

13.5 Conflict, governance, and power beyond AI

Governance, security, and conflict will remain human‑centric domains. AI can monitor, predict, and nudge, but the core decisions—who owns what, who governs whom, who decides about war or migration—will be made by human actors and institutions. Over 2026–2100, this sector will not shrink; it will grow in relative weight as AI‑driven inequalities and fragilities mount.

• 2026–2050
  
  • AI‑driven surveillance, cyber‑warfare, and predictive policing grow 20–30% annually in many states, while the number of soldiers, police, judges, and politicians stay constant or rise 0–5%.
    
  • Conflict‑governance spending grows 5–10% in real terms, as AI‑centric tools account for 30–40% of that growth.
    
• 2050–2075
  
  • As AI‑driven inequality and demand collapse breed unrest, conflict‑linked spending (security, policing, paramilitary, border‑control) grows 10–20% in real terms.
    
  • AI‑centric share of conflict‑governance spending rises to 40–50%, but the human‑centric side (enforcement, negotiation, rule‑making) stays at 50–60%.
    
  • In many regions, governance becomes more “hard‑shell”: high‑tech AI‑centric tools inside a thick layer of human‑centric enforcement and control.
    
• 2075–2100
  
  • In rump‑size regions, conflict‑governance activity contracts 30–50% in absolute size with population, but its share of public spending rises to 20–30%.
    
  • Fewer people, but each still needing rules, enforcement, and protection; governance and security become the last remaining “core state functions.”
    

For families, the takeaway is: governance and security are the last bulwark against AI‑driven chaos. Owning adjacent real‑world assets, or at least understanding the incentives of those who control them, is a long‑term necessity.

13.6 Strategic implications for entrepreneurial families

The future will be AI‑heavy, not AI‑total. The “non‑AI” activities won’t vanish; they will adapt, contract, or expand according to how AI‑driven demand and population evolve. The key is to see them as:

• Stabilising anchors: Survival infrastructure, care, and local subsistence are the backstops when AI‑driven consumption and finance wobble.
  
• Expansion zones under stress: Conflict, governance, and certain forms of emotional and craft‑centric work grow relative importance as AI hollows out the middle.
  
• Non‑AI‑centric but AI‑leveraged: These sectors often use AI as a tool, not as the core business model.
  

For entrepreneurial families, the strategic imperatives are:

• Own the human‑centric, non‑AI‑replaceable foundations: land, energy, water, housing, local care systems, and local‑scale infrastructure.
  
• Avoid over‑betting on AI‑only platforms and attention‑driven bubbles, which are fragile to demand and AI‑driven deflation.
  
• Position for resilience over 2026–2100: as AI hollows out normal work and drags global demand, these non‑AI‑centric domains will not disappear; they will reconfigure, and the families that own the real, human‑scale bricks and mortar will be the ones that survive the reshaking.
  

In an AI‑driven world, the question is not “Will AI dominate everything?” but “What will AI still depend on?” The answer is simple: humans, land, and real‑world infrastructure. That is where the long‑term value lies.

14. Conclusion: Living Through the End of Normal Work (Without Illusions)

By 2100, the world will have decisively passed beyond the 20th‑century idea of “normal work” as a stable, shared, and predictable anchor for life. For much of the last century, labour sat at the centre of identity, income, and social order: people expected to work in a recognisable role, to accumulate some degree of security, and to see their children live slightly better than they did. In the AI‑driven era, that linkage is ruptured for large segments of humanity. Work is no longer a reliable contract between effort and reward; it is a fragmented, precarious input into AI‑centric systems, designed more to optimise extraction than to sustain the people who perform it. Human life is no longer organised around a universal expectation of work with dignity; it is organised around managed scarcity, algorithmic nudges, and varying degrees of detachment from real power. The worker is no longer a “stakeholder” in the system; they are an asset to be scheduled, monitored, and monetised.

The split of humanity is not a metaphor; it is a material, numerically trackable reality. In the most likely, uncontrolled scenario, AI dominates, jobs evaporate, and societies collapse into austerity, managed precariat, and AI‑stabilised oligarchies. Wages for the working class are compressed by AI‑driven automation and global competition, while the owners of AI‑centric capital, data, and platforms capture an ever‑larger share of the surplus. Demand for labour‑intensive services and manufactured goods falls, and the social contract that once linked work, housing, and family formation is frayed to breaking point. In the more hopeful, split‑world scenario, AI‑utopia blocs live in high‑productivity, AI‑driven enclaves, mixed‑AI countries endure a mix of angst and conflict, and subsistence or low‑tech regions fall back into self‑sufficient, low‑demand economies. Even in the unlikely, tightly controlled scenario, where AI deployment is heavily regulated and social‑policy buffers are strong, the pressure on labour, fertility, and human‑scale demand remains severe, even if the pain is slower, more staggered, and partially insulated in certain enclaves.

Across all three trajectories, AI reshapes demand, hollows out middle‑tier jobs, and concentrates control in a small set of owners, platforms, and states. The AI‑driven economy is not a benevolent upgrade; it is a highly efficient, deeply unequal, and politically entrenched system whose form depends on how tightly AI is shackled, and where the shocks of job loss and demand collapse hit hardest. In the most likely, minimally‑regulated path, the drop in demand is sharper and earlier, as AI‑centric firms ruthlessly cut labour costs and as working‑class households respond to stagnant or falling incomes by cutting spending. In the more controlled or split scenarios, the decline is stretched out, patched, and partially buffered by social transfers, protectionism, or regional insulation, but the underlying trend remains the same: fewer people, fewer buyers, and fewer workers over time.

High‑tech cores and off‑world‑adjacent cities, where AI‑utopia tendencies play strongest, concentrate AI‑driven capital, data, and governance. These enclaves use AI to automate everything from R&D to logistics, from security to financial engineering, and to optimise the extraction of surplus from the broader system. The managed majority and permanent precariat—across AI‑dominated and mixed‑AI regimes—live in systems designed to keep them consuming, producing, and complying, but without real control over the architecture that shapes their lives. Semi‑peripheral regions supply resources, low‑cost labour, and niche services, while rejection zones and subsistence enclaves grow as both escape and collapse, particularly in the more chaotic or extreme AI‑driven paths. The working class is everywhere treated as either labour or market, tracked, nudged, and financed by platforms that extract value long before it reaches the worker. The only variation is tempo and severity, not direction.

There is no single, positive or utopian frame for this outcome. The AI‑driven economy is a machine that, left to its own logic, tends toward deeper inequality, slower demand growth, and concentrated power. The family’s survival depends on accepting that all three scenarios imply a world where AI reshapes demand, collapses many jobs, and concentrates control, even if the speed and geography of the pain differ. The working class is not a problem to be solved; it is a resource to be managed. The AI‑driven world is not a future to be embraced wholesale; it is a reality to be navigated, steered, and, where possible, reshaped.

AI will continue to grow, to compress labour costs, and to reshape the global order, and the only way to thrive is to own the assets that AI makes more valuable, regardless of which path the world follows. The moral distance lies not in avoiding AI altogether, but in choosing which sectors to engage with: food, energy, housing, and essential infrastructure, where the decline of demand is slower and the value of real‑world assets is higher, rather than gambling on AI‑centric bubbles, predatory finance, or hyper‑manipulative attention economies, which are fragile to both regulation and AI‑driven deflation. The end of normal work, under any of these scenarios, does not mean the end of human life, but it does mean the end of the old social contract. The family must therefore plan for a world where the working class is a managed, semi‑peripheral force, and where the elite and their AI‑centric enclaves hold the real power, whatever form the AI‑driven future takes.

Crucially, under AI‑driven conditions of isolation, despair, and collapsing demand, fertility tends downward. If TFR stays below replacement for decades, the human project in many regions becomes numerically terminal. As the calculations in this report show, a TFR of 1.5 may push the working‑class‑driven core bloc toward a human‑scale collapse by the late 27th century; at TFR = 1.0 the end‑decade shifts into the late 23rd century, as the bloc falls below 1 million adults around 2275–2300; at TFR = 0.5 it arrives in the mid‑22nd century, dropping below 1 million adults in the 2150–2175 window; and at TFR = 0.0, with no babies born, the last generation dies out by 2100–2125. The drop in purchasing demand accelerates this collapse: fewer people mean fewer buyers, fewer borrowers, fewer workers, and fewer children, until the system that AI was built to serve vanishes from underneath it. The AI‑driven economy is not a solution to the problem of humans; it is a machine that consumes them, and the working class is its fuel.

The AI‑driven future is not a clean separation between utopia and dystopia. It is a long‑run narrowing of human scope, where the number of working people, the number of buyers, and the number of children all fall in a compounding, measurable way. The family’s role is to adapt, to own, and to survive, even as the rest of humanity navigates the slow, mathematically predictable erosion of the human project that its own design has made possible. The question is not “Will AI win?” but “What will AI still depend on?” The answer is simple: humans, land, and real‑world infrastructure. These are the hard‑to‑replace, non‑AI‑centric foundations that remain critical even as the world hollows out. That is where the long‑term value lies, and that is where the family must anchor itself—not in the AI‑driven Mirage, but in the real‑world foundations that AI cannot fully replace.

15. Reference List: Deepening the AI‑Collapse Picture

This is a curated list of books, essays, and media that map the trajectory sketched in this report. They are not “AI solution” texts; they are stories and analyses that show how AI‑driven surveillance, capital concentration, and loss of work are already being dramatised and theorised in fiction and non‑fiction.

1. 1984, by George Orwell

Often cited but rarely fully absorbed, Orwell’s classic is not just about state control; it is a treatise on how information, language, and labour are organised to serve a small, omniscient ruling layer. The “telescreen” and the Ministry of Truth are eerie prefigures of algorithmic surveillance, curated information feeds, and AI‑driven narrative control. The working class (“proles”) are treated as a managed, half‑ignored mass, anticipated, not integrated.

2. We, by Yevgeny Zamyatin

An earlier dystopian model that directly feeds into Orwell’s 1984, Zamyatin’s “We” imagines a hyper‑rational, glass‑walled society in which individuality and spontaneous work are systematically erased in favour of state‑organised productivity and conformity. The book’s logic of planning and control foreshadows AI‑driven centralised scheduling, performance tracking, and the psychologics of obedience that will accompany any AI‑stabilised oligarchy.

3. Brave New World, by Aldous Huxley

Unlike the grim exterior of 1984, Huxley’s vision is one of distraction, engineered contentment, and conditioning. The drug “soma” and the engineered caste system prefigure the way AI‑driven entertainment, cheap goods, and targeted rewards can keep a managed majority pacified while the elite control production and information. The “Fordian” industrial model is a prototype for AI‑driven optimisation of pleasure and consumption.

4. The Culture Series, by Iain M. Banks

A more ambivalent, post‑scarcity vision in which advanced AIs and humanoids live in a galaxy‑spanning society where most work is automated, and need to work has largely vanished. The series explores how AI governors and drones manage entire habitats, economies, and wars, while a small cadre of humans and AIs make strategic choices. It is useful as a counterpoint: a “liberal AI utopia” that still concentrates power in a tight circle, much like the “core integrator” scenario laid out here.

5. Neuromancer, by William Gibson

The cyberpunk classic that presaged the internet, corporate dominance, and data‑driven futures. The novel’s sentient AIs (“Wintermute”) manipulate humans and global networks to achieve their own ends, while characters live in a world of offshore data havens, global capital, and hyper‑connected yet fragile labour. It is a literary map of the “free‑capital” and cyber‑surveillance world that AI‑driven automation will harden, not invent.

6. The Matrix (film and related media, by the Wachowskis)

Though popular culture, The Matrix is a core reference point for the “AI‑simulation” and control mindset. The film imagines a world where human bodies are used as energy, and minds are trapped in an algorithmically managed illusion of normal life while the machines rule. The “desert of the real” and the “desk‑job illusion” resonate with the working class trapped in AI‑managed domains that feel real but are fundamentally alienated.

7. The Youtube video: “The AI Revolution: Our Immune System” (video id: x4ZY25OU4Ys)

This video explores the way AI is reshaping the immune system and biotechnology, but it also serves as a microcosm of the larger AI dynamic: a powerful, capital‑intensive technology is integrated into core infrastructure, promising leap‑year progress but at the cost of dependency on opaque, proprietary systems. The way medical data and AI‑driven decision‑making are centralised mirrors the trajectory of AI‑driven capital in the wider economy.

8. AI in Fiction: 10 to Read in 2025 (MBZUAI book list)

This article curates a set of AI‑centric novels, including classics like 2001: A Space Odyssey and I, Robot, and contemporary works such as Machinehood and Murderbot Diaries. These texts personify AI agents, capture the psychological impact of automation, and dramatise the loss of agency when machines make decisions about humans. They are not predictions, but they are diagnostic: they show how AI‑driven control and alienation are already part of the cultural imagination.

9. AI‑Utopias and AI‑Dystopias – Wikipedia and AI Fiction Surveys

Articles such as “Artificial intelligence in fiction” on Wikipedia and curated lists like “AI in fiction: 10 to read in 2025” and “AI‑dystopia novels” provide a broad survey of themes: AI rebellion, surveillance, post‑scarcity, and control. They help the reader see how the “AI‑driven elite”, “managed majority”, and “subsis‌tence/refusal” zones have already been dramatised in literature and film, often with more clarity than policy debates.

10. “How Science Fiction Dystopianism Shapes the Debate over AI …” (Discourse Magazine)

This essay critically examines how fictional AI tropes—HAL 9000, Skynet, the Borg—shape public and policy thinking about AI, often exaggerating apocalyptic scenarios while underplaying the more mundane, but no less corrosive, reality of gradual, capital‑driven job‑loss and control. It is useful for the discerning reader who wants to separate drama from structure, but still learn from the emotional clarity of the dystopian lens.

For an entrepreneurial family reading this report, these works are not only entertainment; they are symbolic maps that anticipate the AI‑driven world. They show how work will be redefined, how power will be concentrated, and how the working class will be managed. The timeline in this report is compressed, but the themes are not new; they are emerging from a long‑standing dialectic between technology, control, and power. The reference list above provides a way to deepen that understanding, beyond the technical and into the moral and psychological dimensions of the AI‑driven future.

16. Contact MECi for More

If you are an entrepreneurial family, multi‑generational investor, or strategic decision‑maker who wants to go beyond the analysis in this report and translate these projections into concrete positioning, asset‑allocation, and multi‑decadal planning, MECi Group International can help you design and execute a coherent strategy across the 2026–2100 horizon.

16.1 What MECi can do for your family

• Scenario‑oriented planning: Work with you to map your family’s assets, dependencies, and exposure to AI‑driven automation, demand collapse, and demographic risk, and build a 10‑, 30‑, and 75‑year planning framework aligned with the trajectories in this report.
  
• Asset‑repositioning support: Help identify and structure shifts toward AI‑enhanced infrastructure, essential real‑world assets (food, energy, water, housing, healthcare logistics), and human‑scale, non‑AI‑replaceable sectors, while reducing exposure to AI‑fragile, demand‑collapse‑prone industries.
  
• Demographic‑demand stress‑testing: Incorporate low‑TFR and AI‑driven isolation scenarios into your family’s balance‑sheet planning, so you can see, numerically, which holdings survive shrinking human‑scale demand and which are at risk.
  
• Access to curated networks: Connect your family with specialists in energy, real‑estate, infrastructure, governance, and care‑centric systems, so you can move from insight to action across different geographies and regulatory environments.
  

16.2 How to get in touch

This report is an open analysis, but the implementation that follows it is highly customised. If you want to:

• Ask deeper questions about the calculated “end‑decade” thresholds for human‑scale collapse at different TFRs,
  
• Run a tailored projection for your family’s assets across the 2026–2100 horizon, or
  
• Explore structured ways to own and anchor in the real‑world foundations that AI still depends on,
  

reach out via the MECi Group International contact channels listed on the organisation’s official website and LinkTree, where you can request a private briefing, a confidential working‑session, or a multi‑generational strategy workshop.

MECi members work on a project‑by‑project, family‑centric basis, with an emphasis on discretion, long‑term thinking, and AI‑aware realism rather than utopian optimism. The world you are entering will not be simple, but it can be navigated—if your family is willing to see it clearly, own differently, and position early.

Contact us for more here.

Here’s a PDF version of this report:

About the Author

Jeremiah Josey has a long history of business in the Middle East. His ventures include MECi Group International, a Swiss based business mentorship organisation, The Thorium Network, focusing on low cost energy systems using Blockchain and AI, and IPRI Tech, a company commercialising novel technology for extracting gold from difficult minerals.

Post links

MECi Group website

Jeremiah Josey’s website

MECi Linkedin

Tags

FutureOfWork #WealthPreservation #AIEconomy