Why AI Could Shrink the Global Workforce by 30–40% by 2045

The core idea is simple: AI dramatically increases productivity per worker, meaning fewer workers are required to produce the same economic output.

This has happened before in specific sectors. What makes AI unique is that it could affect nearly every sector simultaneously.

1. The Agricultural Example: A Preview of What Could Happen

Agriculture provides a historical example of how technology can drastically reduce labor demand.

United States agricultural employment

1800 — ~80% of workforce 1900 — ~40% 1950 — ~12% Today — ~1–2%

Despite the collapse in agricultural employment:

• food production increased dramatically
• food prices fell
• living standards improved

Technology such as tractors, fertilizers, and automation allowed a tiny workforce to feed the entire country.

AI could trigger similar transformations across many industries at once.

2. AI Targets the Largest Employment Sectors

The global workforce is concentrated in a few major sectors.

Approximate distribution of global employment:

Agriculture — ~27% Manufacturing — ~23% Services — ~50%

AI affects manufacturing and services heavily, which together represent roughly 70% of global employment.

This means the potential labor impact of AI could be extremely large.

3. Knowledge Work Is the Largest White-Collar Category

Within the service economy, knowledge workers make up a massive share of jobs.

Examples include:

• accountants
• lawyers
• analysts
• software developers
• marketing professionals
• HR staff
• administrative roles
• customer support
• researchers

Generative AI can already assist or partially automate many of these functions.

Early studies suggest AI could automate 25–60% of tasks in many knowledge jobs.

This does not necessarily eliminate the job immediately, but it reduces the number of workers required.

For example, a team that previously required 10 analysts may only need 4–5 analysts with AI assistance.

4. AI + Robotics Will Eventually Automate Physical Labor Too

AI today is strongest in digital work, but robotics is advancing rapidly.

Areas likely to see automation include:

• warehouse logistics
• trucking and transportation
• retail checkout
• construction planning
• manufacturing assembly

As AI becomes integrated with robotics, physical jobs will increasingly be automated.

This could affect hundreds of millions of workers globally.

5. The Productivity Explosion Effect

Historically, productivity increases gradually.

AI could produce unprecedented productivity growth.

Example scenario:

Traditional employee — 1× output Employee with AI copilots — 3–5× output Autonomous AI systems — up to 10× output

If productivity increases fivefold, companies may need far fewer employees to produce the same output.

If productivity increases tenfold, workforce demand could shrink dramatically.

6. The Startup Efficiency Signal

A sign of this trend is already visible in startup ecosystems.

Many AI startups today operate with extremely small teams.

Examples emerging in Silicon Valley:

A startup generating $10M in annual revenue may only have 10–15 employees.

A startup generating $50M in revenue may have only 20–40 employees.

Historically, companies with similar revenue often required hundreds of employees.

If this pattern spreads across industries, the economy may require far fewer workers overall.

7. The Speed of AI Adoption Is Much Faster Than Past Revolutions

Another factor that makes AI different from previous technological revolutions is the speed at which it spreads.

Historically, new technologies took decades to reach widespread adoption.

Approximate global adoption timelines:

Steam engine — about 80 years Electricity — about 60 years Automobiles — about 50 years Internet — about 25 years Smartphones — about 15 years AI tools — about 3–5 years

Generative AI tools reached hundreds of millions of users within months.

The reason is that AI is software, not physical infrastructure.

Previous revolutions required building factories, installing power grids, and manufacturing machines. Those processes naturally slowed adoption.

AI spreads very differently.

Because AI runs in the cloud, companies can integrate it into products and workflows almost instantly.

A new AI capability can reach millions of workers through software updates within weeks.

This means the economic impact of AI could unfold much faster than previous technological revolutions.

If companies integrate AI rapidly into workflows, labor displacement could occur faster than new industries and jobs have time to form.

Economic Consequences of a Smaller Workforce

If the global economy eventually requires 30–40% fewer workers, it would reshape many economic systems.

1. Impact on Salaries

The impact on wages could be uneven.

High-skill AI talent

People who design or build AI systems could see very high salaries.

These include:

• AI researchers
• robotics engineers
• chip designers
• data infrastructure engineers

Demand for these roles may far exceed supply.

Mid-skill knowledge workers

Many mid-skill white-collar jobs could see:

• slower wage growth
• increased competition
• fewer available positions

This could shrink the upper-middle class in developed economies.

Low-skill service workers

Some service jobs may remain difficult to automate, such as caregiving or hospitality.

However, wages could remain relatively low because the labor supply may increase.

2. Impact on Housing Markets

Housing demand is heavily tied to income stability and employment.

If AI reduces the number of high-paying jobs in certain regions, housing markets could change significantly.

Cities heavily dependent on tech employment — such as Seattle, San Francisco, or Austin — could experience:

• slower price growth
• greater volatility
• potential price corrections during tech layoffs

This is particularly relevant in regions where technology jobs drive housing demand.

Population concentration shifts

AI could also enable remote work and automation, reducing the need to live near economic centers.

This might lead to:

• weaker demand in expensive tech hubs
• stronger demand in lower-cost regions

3. Impact on Economic Growth

At first glance, fewer workers sounds negative for economic growth.

However, productivity matters more than workforce size.

If AI dramatically increases productivity, total economic output could still grow rapidly.

For example:

Today — 100 workers × productivity 1 → economic output 100

AI era — 70 workers × productivity 3 → economic output 210

Even with fewer workers, the economy could more than double in size.

4. Wealth Concentration Risks

The biggest economic risk may be wealth concentration.

AI productivity gains may flow primarily to:

• AI companies
• technology investors
• countries with advanced AI infrastructure

This could widen the gap between:

• technology-leading countries and others
• capital owners and workers

Which White-Collar Jobs May Shrink by 2035

AI is likely to reduce demand in professions that involve routine cognitive work or information processing.

Some of the most exposed professions include:

Legal document review and paralegal work Basic accounting and bookkeeping Customer support roles Market research analysts Junior software development roles Content writing and marketing copywriting Administrative and clerical positions

These roles involve tasks that AI systems can increasingly perform quickly and at very low cost.

This does not necessarily mean these professions disappear entirely, but fewer workers may be required in each field.

Which Professions May Grow

At the same time, new types of jobs are likely to emerge.

Professions that may grow include:

AI engineering and research Robotics engineering AI safety and alignment specialists Data infrastructure and compute engineering Cybersecurity Human-AI collaboration roles Healthcare and eldercare services Education and AI-assisted learning

Jobs that combine technical expertise, creativity, and human interaction may remain the most resilient.

Possible Policy Responses

Governments may need new economic policies to adapt.

Ideas being discussed include:

• universal basic income
• shorter workweeks
• AI taxes on automation
• large investments in education and reskilling

The political feasibility of these policies remains uncertain.

The Big Historical Question

Every technological revolution has raised the same fear:

“Will machines replace humans?”

Historically, the answer has always been no.

Instead, technology created new industries and new forms of work.

But AI could be different because it can perform general cognitive tasks, not just specialized physical ones.

The Most Likely Outcome

Most economists currently believe the most likely scenario is:

• AI eliminates many tasks
• jobs evolve rather than disappear
• productivity increases dramatically

However, the transition period could involve significant labor market disruption over the next 10–20 years.

Key Takeaway

AI will probably not eliminate work entirely — but it may dramatically reduce the number of workers needed in many industries.

If that happens, the effects could reshape:

• labor markets
• housing demand
• salaries
• global economic power

in ways we have not seen since the Industrial Revolution.