Who Gets to Live in the AI Economy?

The New Household Economy, Part 2

Prepared: 19 June 2026
Editorial series: The New Household Economy
Style: research-led strategy editorial

The AI economy will be sold as intelligence.

It will be won by infrastructure.

That is the less glamorous truth. The countries and households that benefit most from AI will not simply be the ones with the cleverest slogans, the most dramatic launch events, or the loudest national strategies. They will be the ones with the boring things working underneath: electricity, broadband, schools, devices, compute, procurement, labour protections, competition policy, practical training, and employers who know how to redesign work without breaking the people inside it.

In other words, AI readiness is not a press release.

It is a household condition.

If a country has weak power, expensive connectivity, underfunded schools, fragile institutions, and few routes for workers to retrain, AI does not become a national miracle just because people can open a chatbot. It becomes another technology whose best version is captured by the already connected.

That is the second claim of The New Household Economy:

The AI divide will not only be between workers. It will be between households backed by functioning systems and households left to adapt alone.

The Country Divide Is Blunter Than The Software Demo

The IMF's AI Preparedness Index places advanced economies well ahead of emerging and low-income economies on average. UNCTAD warns that AI development is concentrated among a relatively small group of companies and countries. Stanford's AI Index shows rapid generative AI adoption, but with large differences by country and GDP per capita.

This does not mean rich countries simply win and poor countries simply lose.

The picture is more interesting.

Rich countries face faster disruption because more of their workers are in cognitive, office, professional, administrative, and service roles that AI can reshape. Poorer countries may face less immediate direct exposure because more work is physical, informal, agricultural, local, or service-based. Middle-income countries may adopt quickly, but many are exposed in global services, freelance work, business process outsourcing, tutoring, support, coding, content, and back-office tasks.

So the map is not safety versus danger.

It is timing versus capacity.

Advanced economies face earlier disruption but have more money, institutions, and infrastructure to manage it. Lower-income economies face less direct exposure in the short term but risk missing productivity gains. Middle-income economies sit in the sharpest place: enough digital talent to benefit, enough service exposure to be vulnerable, and often not enough institutional cushion to absorb the shock smoothly.

The United States: Capital, Compute, And The Middle-Class Squeeze

The United States has the strongest AI capital base, frontier labs, cloud infrastructure, enterprise adoption, and commercial scale. That gives American households enormous upside if they sit close to technology, finance, law, defence, healthcare, software, research, media, consulting, or entrepreneurship.

It also creates a harsher labour-market mirror.

The U.S. has many workers in highly exposed cognitive roles. It has expensive education, expensive healthcare, high housing costs in opportunity hubs, and a labour market where job loss can quickly become a household crisis. For senior professionals and founders, AI can multiply output. For junior white-collar workers, routine office staff, call-centre teams, low-end freelancers, and middle-income families with fixed costs, it can compress the ladder.

America may produce many of the tools that reshape work.

That does not guarantee every American household shares in the upside.

There is also the infrastructure bill. Data centres are becoming local political objects: electricity demand, water use, grid queues, tax incentives, land use, and community benefit deals. The household question is not only whether AI raises wages. It is also who pays for the power system that makes the AI economy possible.

Europe: Protection Is Not The Same As Power

Europe's strength is seriousness: regulation, public institutions, industrial depth, education, consumer protection, and a stronger instinct for social cushioning.

Its weakness is scale.

The UK has strong research, services exposure, finance, legal work, creative industries, and public-sector pressure. London and the South East may capture a disproportionate share of the gains, while routine office work, outsourced support, and junior professional ladders face pressure.

Germany has an industrial advantage. AI could matter deeply in manufacturing, engineering, logistics, robotics, quality control, maintenance, and enterprise software. The risk is not only job loss. It is competitiveness loss if firms move too slowly and industrial AI value migrates elsewhere.

France has talent, state capacity, and ambition, but like much of Europe, it faces the classic translation problem: turning research, regulation, and talent into companies and systems that scale globally.

The harsh truth for Europe is simple:

Labour protection can slow household damage. It cannot by itself guarantee value creation.

If Europe protects workers but fails to build and scale the systems where AI value is captured, it may preserve the social floor while losing part of the economic ceiling.

India: The Opportunity And The Threat Are The Same Shape

India is one of the most important countries in the AI household story.

It has a young population, deep technical talent, English-language advantage, digital public infrastructure, a huge services base, and fast adoption. AI could upgrade education, small business, healthcare navigation, language access, government services, exports, and entrepreneurship at extraordinary scale.

But India's risk sits inside its strength.

Millions of households have climbed through IT services, business process outsourcing, tutoring, coding, content, support, admin, and global back-office work. These are precisely the kinds of language and workflow tasks that AI can reprice.

This does not mean India's service economy disappears. It means the bargain changes.

The old model sold labour, scale, and process.

The stronger future model sells AI-enabled expertise, domain specialisation, productised services, local language systems, trust, implementation, and human judgement at scale.

India can win this transition. But it has to move from labour arbitrage to intelligence orchestration before the market forces the move on households without enough savings.

Brazil, The Philippines, Indonesia, And Mexico: The Service-Ladder Test

Several middle-income economies sit in a similar strategic category. They have young populations, growing adoption, digital services, remote work, outsourcing exposure, and large urban middle classes trying to climb.

Brazil has scale, creativity, digital energy, and a large services sector. Mexico is tied deeply into U.S. manufacturing and cross-border operations. The Philippines has major BPO exposure and English-language service strength. Indonesia has scale, youth, mobile adoption, and domestic market potential.

These countries can benefit from AI in education, small business, agriculture, logistics, healthcare, manufacturing, public services, and entrepreneurship.

They can also discover that the global market no longer pays the same premium for routine remote service work.

The strategic question is whether these economies can help households move up the value chain fast enough. From call handling to customer operations. From admin to workflow management. From content production to brand and market strategy. From basic coding to implementation and product judgement. From labour supply to trusted delivery.

That transition will not happen by motivational language.

It needs training systems, employer incentives, affordable tools, English and local-language AI fluency, SME adoption, and pathways for workers to prove higher-value capability.

Africa: Leapfrogging Still Needs Electricity

Africa's direct generative AI exposure is lower in many countries because more work is informal, physical, agricultural, local, or service-based.

Lower exposure is not the same as safety.

The risk is exclusion from the productivity gains while still facing AI-enhanced global competition. A young designer, writer, tutor, coder, support worker, or entrepreneur in Lagos, Nairobi, Johannesburg, Accra, or Kigali may be able to reach global markets with AI. But the advantage weakens if electricity is unreliable, data is expensive, devices are poor, schools are under-resourced, and payment or trust infrastructure is fragile.

Nigeria has youth, English-language advantage, entrepreneurship, and cultural export power, but infrastructure constraints are severe. South Africa has stronger corporate infrastructure in finance, mining, telecoms, and professional services, but inequality will concentrate gains unless the transition is deliberate. Kenya has a strong digital-services and mobile-money foundation, but informal workers need usable tools in real contexts, not abstract national ambition.

The harsh truth:

No country leapfrogs electricity, connectivity, education, and institutions for long.

AI can accelerate a capable system. It cannot permanently substitute for one.

Small Rich States: The Advantage Of Coordination

Singapore and the UAE show another path: small, wealthy, infrastructure-rich, policy-driven, and able to coordinate quickly.

These countries can treat AI as a national productivity system. They can connect government services, enterprise adoption, education, infrastructure, regulation, and talent policy more tightly than large fragmented economies.

That creates real household upside for citizens and high-skill residents.

But it also raises a distribution question. If migrant workers, lower-wage service workers, and less powerful residents do not share in the gains, the AI economy becomes a polished surface over a familiar hierarchy.

National AI success should not be measured only by adoption dashboards. It should be measured by whether ordinary households gain time, income, access, education, and resilience.

Energy Becomes A Household Issue

AI feels weightless when it appears in a browser.

It is not weightless.

It lives in data centres, chips, cooling systems, transmission lines, water systems, planning decisions, and power contracts. The IEA expects data-centre electricity demand to rise sharply this decade, with AI-focused facilities growing quickly.

This makes compute a local political issue.

Communities will ask sharper questions: Who gets the jobs? Who gets the tax revenue? Who pays for grid upgrades? Who absorbs water stress? Who benefits from cheap power contracts? Who carries the planning disruption? Why does the neighbourhood carry the infrastructure while the model owner captures the margin?

That is why energy belongs in a household economy series.

If AI infrastructure raises local electricity pressure, redirects public investment, affects water use, or changes tax incentives, households are involved whether they use the tools or not.

The AI economy is not just a cloud. It has a postcode.

Schools Are The Real AI Infrastructure

The deepest national divide may be educational.

Not whether schools allow chatbots. That is a small argument pretending to be the whole debate.

The real question is whether education systems can teach children how to think in a world where machine-generated language is abundant.

Students need writing, maths, statistics, memory, reading stamina, attention, practical problem-solving, debate, ethics, source judgement, and the ability to work through difficulty. AI can support all of this. It can also weaken all of it if used carelessly.

The household split will be visible.

Some children will use AI as a thinking partner: to compare explanations, test drafts, explore examples, practise languages, debug misunderstandings, and receive patient feedback.

Others will use AI as a shortcut machine.

The first group becomes more capable. The second becomes more dependent.

That is not a technology problem alone. It is an adult guidance problem. Families, teachers, schools, and governments need a practical standard for AI literacy that protects thinking rather than merely policing cheating.

The Employer Is Now A Household Institution

Employers have always affected households. In the AI economy, their role becomes even larger because they decide whether productivity becomes resilience or extraction.

They decide whether saved time becomes training, better service, shorter hours, career progression, higher pay, or just more output.

They decide whether junior work disappears or apprenticeship is rebuilt.

They decide whether admin workers are retrained or slowly squeezed.

They decide whether frontline workers are given tools or simply measured by systems.

They decide whether AI is used to improve customer service or trap people inside cheaper self-service mazes.

The blunt test remains:

If AI saves ten hours a week, who gets the ten hours?

This is not a soft question. It is the commercial and social question of workplace AI.

What Governments Should Actually Do

Governments need a less theatrical AI policy.

The priority is not only national champions, safety summits, and impressive strategy documents. Those may matter, but they do not answer the household question.

The household question is: can ordinary people absorb the transition and benefit from the upside?

That requires practical policy:

  • Affordable broadband, devices, and public digital access.
  • Energy planning that makes data-centre growth transparent and fairly allocated.
  • Schools that teach AI literacy without sacrificing foundational skill.
  • Vocational and adult training tied to real jobs, not vague future-work language.
  • Wage insurance, portable benefits, and support for self-employed workers.
  • Competition policy that prevents excessive concentration of model, cloud, and data power.
  • Public procurement that helps local firms build capability.
  • Human escalation in public services for vulnerable citizens.
  • Employer incentives to rebuild apprenticeship rather than hollow it out.

The winning national strategy will look almost boring from the outside.

That is the point.

In the AI economy, boring state capacity becomes a luxury good.

The Bottom Line

The countries that benefit most from AI will not necessarily be the ones with the loudest ambition.

They will be the ones that turn intelligence into household resilience.

Rich countries have the capital and infrastructure, but they also have highly exposed labour markets and expensive household obligations. Poorer countries have less direct exposure in some sectors, but risk missing the productivity dividend. Middle-income countries have enormous upside and serious vulnerability because their service ladders are close to the tasks AI can reprice.

For households, the future divide will not be simply "has AI" versus "does not have AI."

It will be:

  • households with guidance versus households with access only;
  • workers who command systems versus workers measured by systems;
  • countries with infrastructure versus countries with slogans;
  • employers who rebuild work versus employers who merely cut cost;
  • communities that share in AI infrastructure value versus communities that host the burden.

AI is not destiny.

But it is pressure.

And pressure reveals what a household, company, and country had in reserve.

The new household economy will be decided by that reserve: money, time, skill, trust, infrastructure, institutions, and the ability to move before the market moves beneath you.

That is why the most powerful AI policy may not sound like AI policy at all.

It may sound like electricity. Schools. Apprenticeship. Broadband. Competition. Public services. Worker transition. Honest employers.

The future will call it intelligence.

Households will experience it as margin.

Source Notes