How the same AI and robots will split twelve economies, and what decides which way each one goes.

Artificial intelligence is spreading across the global economy, but there is little reason to assume that its economic consequences will be the same everywhere.

A factory in Japan and a business-services company in India can deploy similar AI systems and experience completely different outcomes.

The Japanese factory may need automation because it cannot recruit enough workers.

The Indian company may use the same technology to reduce the number of junior employees it needs.

In one case, AI helps preserve production.

In the other, it can weaken the career pathway through which young people enter the middle class.

This distinction is the starting point of our latest THRIVE IN CHAOS research.

We examine twelve economies with very different demographic structures, labour markets, capital resources and institutional arrangements.

Our central conclusion is that AI will tend to amplify the conditions that already exist rather than automatically make economies converge.

The most important question is therefore not which country adopts AI fastest.

It is which countries can turn higher productivity into productive employment, household incomes and sustainable demand.

Consider four developments from 2026.

In the United States, the share of nonfarm business output paid to workers fell to 52.8% in the second quarter, its lowest recorded level since the series began in 1947.

This occurred during a period of substantial investment in AI-related infrastructure and technology.

In China, unemployment among people aged 16–24 who were not in education reached 18.9% in August, matching the highest level recorded under the current statistical method.

At the same time, China maintained enormous industrial production capacity and a goods trade surplus of almost $1.2 trillion.

In Russia, unemployment was approximately 2%, reflecting an exceptionally tight labour market.

Businesses faced difficulty recruiting workers and increasing pressure on wages and costs.

In Japan, the number of foreign workers reached 2.57 million, a new record, as companies continued to struggle with domestic labour shortages.

These observations point toward four different economic problems.

The United States is highly productive, but the share of output reaching workers is under pressure.

China has enormous production capacity, but employment opportunities and household consumption have not expanded proportionately.

Russia has jobs but insufficient workers.

Japan has capital and advanced technology but a shrinking domestic workforce.

All four economies can use AI.

What differs is what they need AI to accomplish.

Our series uses the concept of the Economic Core to examine the relationship between production and economic sustainability.

The Economic Core consists of residents who simultaneously produce economic value, earn sufficient income, contribute to public finances and participate in domestic consumption.

Its capacity depends on several connected factors: the number of productive people, their skills, capital per worker, the productivity leverage provided by technology, the share of income reaching households and their net fiscal contribution.

This is an analytical framework rather than an official statistical measure.

Its purpose is to show why technological progress alone is insufficient.

AI can increase the amount of output produced by one worker.

But it does not automatically increase the number of people in productive employment.

It does not guarantee that wages will rise.

And it does not ensure that additional output will be purchased by domestic consumers.

Those outcomes depend on the economic structure surrounding the technology.

Japan, South Korea, Germany and the European Union form the first group.

Their central challenge is demographic ageing and labour scarcity.

For these economies, automation has a relatively straightforward economic purpose: allowing fewer workers to produce more.

Japan is the clearest example.

People aged 65 and over account for almost 30% of its population.

The economy increasingly depends on maintaining productivity despite a declining working-age population.

Automation can help factories, healthcare providers, logistics companies and service businesses continue operating with fewer employees.

South Korea faces similar pressures.

Its working-age population is projected to fall from 36.6 million in 2023 to 27.2 million in 2044.

It already has the highest industrial robot density in the world, according to the International Federation of Robotics figures used in our research.

Germany has strong industrial institutions and a history of using automation alongside skilled labour, but it also faces industrial employment losses and high energy costs.

The opportunity across this group is to raise productivity per worker.

The constraint is domestic demand.

Robots can manufacture goods, but they do not become consumers.

An economy with fewer residents can remain prosperous, but maintaining aggregate demand becomes more challenging.

This means that the long-term success of automation depends partly on whether productivity gains raise household incomes enough to offset demographic decline.

China, Vietnam and Brazil form the second group.

These economies face a difficult combination.

They need to increase productivity rapidly while their demographic advantage is weakening.

China already possesses extensive manufacturing infrastructure, advanced industrial automation and globally competitive industries.

But its household consumption accounts for only about 40% of GDP.

That creates a structural imbalance.

If AI increases industrial output while domestic purchasing power remains relatively weak, the economy becomes more dependent on foreign buyers.

This can intensify trade tensions and competition with producers elsewhere.

China's central challenge is therefore not simply to produce more.

It is to convert production capacity into stronger household demand.

Vietnam faces a different version of the problem.

Its favourable demographic period is expected to end around 2036.

The country has approximately a decade to strengthen its position in higher-value manufacturing, engineering and domestically owned enterprises.

If it succeeds, it could demonstrate how a middle-income economy can use technology to raise incomes before ageing becomes a stronger constraint.

Brazil has a large domestic market and considerable natural resources, but its productivity growth remains a concern.

For all three countries, the next decade is an opportunity that will not remain open indefinitely.

India, Egypt and Iraq face the opposite demographic problem.

They have large young populations that need productive employment.

Historically, this could provide a powerful development advantage.

A growing workforce supported industrial expansion, domestic consumption and rising tax revenues.

But AI changes the relationship between population growth and employment.

India is particularly exposed because its development model has relied heavily on information technology and business services.

These industries have provided career opportunities for millions of educated young people.

AI can automate many routine professional tasks, potentially reducing demand for entry-level workers.

But it can also make employees substantially more productive and allow companies to serve far more customers.

The outcome depends on which strategy firms pursue.

Egypt illustrates the scale of the employment challenge.

Approximately 1.3 million young people enter its labour market annually, while roughly 500,000 jobs are created.

A technology strategy focused only on replacing workers would widen that gap.

Iraq faces even more fundamental constraints, including infrastructure, electricity, water, education and institutional capacity.

AI cannot substitute for these foundations.

For young economies, the most valuable application of AI may be enabling more people to perform productive work rather than reducing the number of people required.

That distinction will help determine whether demographic growth becomes an economic advantage or a source of instability.

The United States occupies a unique position.

It has leading AI companies, research institutions, semiconductor design capabilities, cloud infrastructure and deep capital markets.

It is well placed to benefit from the next generation of technology.

But there is a significant distribution question.

The share of nonfarm business output paid to workers fell to 52.8% in Q2 2026.

A single quarter cannot establish the long-term consequences of AI, and the labour share is influenced by the broader business cycle.

Nevertheless, the longer-term decline in labour's share of output raises an important issue.

If AI allows companies to produce substantially more with fewer workers, how much of the resulting value will reach households?

One possibility is that productivity gains create new occupations, raise wages and support stronger consumption.

Another is that gains become increasingly concentrated among capital owners while median household incomes lag.

Both outcomes are technologically possible.

The difference lies in labour-market conditions, business organisation, competition and institutions.

America's AI future will therefore be determined partly by economic decisions that are not themselves technological.

Russia is a distinct case.

It faces an acute shortage of workers, which should create strong incentives for automation.

But successful automation requires access to equipment, engineering capabilities, industrial software, advanced components and financing.

Restrictions on parts of that technology supply chain complicate the response.

Where businesses can obtain suitable systems, automation may help offset labour shortages.

Where they cannot, rising wages may increase production costs without equivalent productivity improvements.

This creates the risk of persistent cost pressure rather than the more favourable combination of labour scarcity and productivity growth observed in some advanced ageing economies.

Across the twelve economies, two forces are particularly important.

The first is the balance between available workers and productive jobs.

When workers are scarce, businesses have strong incentives to use technology to increase the productivity of existing employees.

When workers are abundant, automation can become a way to reduce recruitment and labour costs.

These are tendencies rather than universal rules, but they shape how businesses evaluate technology.

The second is income distribution.

A company may become more productive without paying higher wages or expanding employment.

At the national level, this can create an imbalance between productive capacity and domestic purchasing power.

If many companies follow the same path, aggregate production can grow faster than household demand.

The result may be greater dependence on exports, increased competition for foreign markets and weaker economic resilience.

These two mechanisms help explain why AI adoption alone is an inadequate measure of economic success.

Our research identifies three broad scenarios.

Base scenario — 55%: Divergence by adaptation.

Economies continue following different paths according to their starting conditions.

Ageing countries use automation to compensate for labour shortages.

China maintains high productivity but makes limited progress toward stronger household consumption.

India captures some AI-related opportunities while youth unemployment remains elevated.

Vietnam continues developing but faces growing demographic constraints.

Stress scenario — 30%: Gains to capital, demand stalls.

Productivity increases, but a growing share of the benefits accrues to capital owners.

Household demand weakens relative to production.

Export competition intensifies, and young workers in several economies face increasing difficulty entering professional employment.

Extreme scenario — 15%: AI investment bust.

AI-related investment fails to deliver expected returns.

A sharp correction in technology-related asset prices affects household wealth, business investment and employment.

Financing becomes more difficult, particularly for emerging economies.

These probabilities are analytical estimates, not statistically calibrated forecasts.

They provide a framework for monitoring developments and revising conclusions.

We have recorded four specific forecasts.

2027 — United States

There is a 70% probability that the US nonfarm business labour share will remain below 54% in Q2 2027.

2029 — China

There is a 65% probability that unemployment among people aged 16–24, excluding students, will be at least 15% in August 2029.

2030 — India

There is a 40% probability that unemployment among people aged 15–29 will fall to 9% or below for calendar year 2030.

2035 — Vietnam

There is a 50% probability that GDP per capita will reach at least $10,000 in current US dollars.

The forecasts have defined thresholds, verification sources and resolution dates.

Their purpose is to make the analysis testable rather than relying on broad predictions that cannot be evaluated.

The first step is to understand the labour-market conditions of the economy in which you work.

Are workers with your skills scarce?

Is your profession attracting more entrants than available positions?

Are wages rising alongside productivity?

These questions matter more than general statements about AI replacing or creating jobs.

Over the next 6–24 months, individuals should develop capabilities that complement AI rather than compete directly with its routine output.

These include domain expertise, client relationships, quality control, technical supervision and the ability to verify AI-generated work.

For those considering international relocation, it is important to compare countries not only by salary levels but also by labour scarcity, employment access and the relationship between productivity and wages.

A high nominal salary in a saturated labour market may provide less long-term security than a lower initial salary in a country where your skills are genuinely scarce.

Businesses need to adapt their AI strategy to the economic conditions of each market.

In labour-scarce economies, automation should focus on improving the capabilities of existing workers.

In labour-abundant economies, companies should evaluate whether AI can help them expand services, reach new customers and create higher-value work.

Across all markets, AI adoption should be measured against more than headcount reduction.

Important indicators include revenue per employee, service quality, customer retention, workforce capability and the creation of new demand.

Companies operating internationally should also recognise that their customers, suppliers and employees may belong to different economic worlds.

A single global strategy may fail because the structural problems of those markets are fundamentally different.

Capital allocation requires distinguishing technological capacity from sustainable economic demand.

The economies adopting AI fastest will not necessarily deliver the strongest long-term outcomes.

More durable opportunities may emerge where productivity improvements translate into household incomes and consumption.

Ageing economies create structural demand for automation, healthcare and services for older populations.

Young economies create opportunities in education, vocational training, labour-intensive manufacturing and services that enable first-time earners to participate in the formal economy.

Middle-income economies present both significant opportunities and risks as they attempt to raise productivity before demographic conditions become less favourable.

Concentration in AI-related assets should also be evaluated against the possibility that investment returns disappoint.

The central question is whether technological investment ultimately creates sufficient economic value to support the expectations already reflected in asset prices.

Several developments would challenge our central thesis.

If the US labour share recovered above 56% for a sustained period without a recession, concerns about the concentration of productivity gains would weaken.

If youth unemployment declined substantially in India while service exports expanded, it would suggest that AI was helping create productive opportunities rather than primarily removing entry-level jobs.

If China's household consumption share rose by four or more percentage points of GDP within five years, the imbalance between production and domestic demand would become less severe.

And if fertility rates in East Asia increased substantially and remained higher, the longer-term demographic outlook would improve, even though the workforce of the 2030s would be largely unaffected.

These are not secondary details.

They are the conditions under which our conclusions should be revised.

Artificial intelligence is not moving the world toward one economic future.

It is interacting with demographic, institutional and financial conditions that differ substantially between countries.

Where workers are scarce, AI can help maintain production and raise output per person.

Where productive employment is scarce, it can make entry into the labour market more difficult.

Where household incomes are weak, higher production can deepen dependence on external demand.

And where the benefits of productivity are concentrated among capital owners, technological progress may coexist with weaker broad-based purchasing power.

The most important question is therefore not how quickly a country can install robots or build data centres.

It is how effectively that country can combine technology, capital and human capabilities to create a productive, well-paid population.

The economies that succeed will be those that turn AI from a source of greater output into a foundation for stronger economic participation.

THRIVE IN CHAOS

Series I — The Shrinking Core

Article 3 of 4: Same Machines, Different Futures

Read the full research: thriveinchaos.ai

Signal → Meaning → Action → Stability

Next in the series: 2030, 2035, 2040 — forecasts and a playbook for the shrinking Economic Core.

This publication provides strategic analysis, not individual financial or investment advice.