SoftBank’s Masayoshi Son Sees 1 Billion Humanoid Robots by 2040 — and a $34 Trillion AI Economy

Son’s Grand Prediction: 1 Billion Humanoids and a $34 Trillion Economy

At SoftBank Group’s annual investor conference, founder and CEO Masayoshi Son presented a vision that would reshape the world economy. He projected that by 2040 the planet will host 1 billion humanoid robots equipped with advanced artificial intelligence, supplemented by 100 trillion autonomous AI agents capable of self-replication and evolution without continuous human commands. Son argued the shift would be fundamental, moving from a human-centred society to an economy dominated by intelligent machines. He estimated that AI would account for roughly 20% of global GDP — equivalent to more than $34 trillion — but realised only through annual investments of around $5 trillion in data centres, semiconductors and robotics.

What sets humanoid robots apart is their ability to operate in environments built for people. Unlike traditional industrial machines, they can climb stairs, use existing tools and switch tasks without costly facility redesigns. Son claimed each unit could replace the work of about ten people, promising a jump in productivity not seen since the Industrial Revolution — but compressed into little more than a decade. Companies such as Tesla with its Optimus project, US start-up Figure AI, and a host of Japanese and Chinese robotics firms are already racing to commercialise such machines.

The forecast was met with a mix of enthusiasm and caution, given Son’s history of both spectacular bets and painful misses. Nevertheless, analysts point out that regardless of the exact number of robots, the direction is unmistakable. Governments in the world’s largest economies are increasingly treating automation as a response to demographic decline and labour shortages, particularly in fast-ageing societies like Japan and South Korea.

The Forces Behind the Forecast — Winners, Losers, and the Global Inflection Point

The raw numbers Son offered are attention-grabbing, but the deeper story lies in how the pieces interact — and who stands to gain or lose if anything close to this vision comes to pass.

Why the Humanoid Form Factor Matters

The physical design is both a weakness and a decisive advantage. A human-shaped robot is not optimal for every task, yet it instantly inherits the entire accumulated infrastructure of human workplaces. That means deployment can scale rapidly without the need to redesign factories, warehouses or care facilities. This versatility opens automation in sectors long considered too complex, from construction and eldercare to short-haul logistics and hospitality — areas where labour shortages are already acute.

The Productivity Promise and the Jobs Shock

Son’s claim that one humanoid could do the work of ten people implies a productivity surge on a scale that has no peacetime precedent. Research from institutions such as McKinsey Global Institute already indicates that AI-driven automation could force hundreds of millions of workers to change occupations by 2030. The humanoid wave would hit repetitive and physical roles hardest: assembly-line jobs, warehouse picking, cleaning and basic service tasks. History shows that such transitions rarely distribute benefits evenly. Without large-scale adult retraining and robust social safety nets, structural unemployment and deeper income inequality are real risks.

The Energy and Capital Hurdle

The infrastructure required is staggering. Son’s scenario calls for data centres that would consume 3 terawatts of power — equivalent to today’s total global electricity generation. That demands building a vast number of new power plants, upgrading transmission grids and securing stable baseload sources such as nuclear or renewables. Annual spending of $5 trillion on AI infrastructure can be shouldered only by the largest economies and corporations, threatening to widen the technology gap between rich nations and the rest. Rising energy demand may also push up electricity prices for ordinary consumers and businesses, weighing on living costs and industrial competitiveness.

Global Winners and Losers

China appears best positioned: it has a colossal manufacturing base, strong state backing for technology, and an urgent demographic challenge that robots could address. The United States benefits from deep innovation and venture capital, with firms like Tesla and Figure AI pushing the frontier. Japan and South Korea bring decades of robotics expertise and ageing populations that make humanoids a necessity, not just an opportunity. Conversely, developing nations that compete on low-cost labour stand to lose their main advantage. Sectors such as textiles, electronics assembly and basic logistics across Southeast Asia and Latin America would face intense competitive pressure as advanced economies re-shore production with robots instead of people.

The Dual-Use Dilemma

Humanoid robots have an inherent dual-use nature. Designed for civilian tasks, they can navigate the human environment — opening doors, climbing stairs, handling tools — which also makes them capable of handling weapons and operating military vehicles. Companies like Boston Dynamics and Tesla are already watched by defence ministries worldwide. In conflict, humanoids could perform missions that currently require soldiers, reducing human casualties and operating in contaminated environments where people cannot go. This creates a new dimension in the global arms race, where civilian robotics breakthroughs directly feed military capability.

What Son’s Vision Means for Business Leaders and Policymakers

Even if Son’s trillion-agent vision is a decade off, the trajectories in capital spending, energy infrastructure and labour markets are being set now. Concrete steps for decision-makers include:

• For corporate strategists: Model scenarios in which a single humanoid substitutes for up to ten workers in manufacturing, warehousing or logistics. The claim is ambitious, but even partial realisation would make labour-intensive operations in rival low-cost jurisdictions uncompetitive. Re-evaluate automation roadmaps against Son’s unit-cost assumptions.
• For energy and utility firms: The projected 3 TW data centre demand — equal to today’s global electricity use — points to a generational infrastructure opportunity. Identify sites for gigawatt-scale facilities and begin permitting processes now, especially in regions likely to host AI hyperscalers. The $5 trillion annual spend figure signals that first-mover capacity will command premiums.
• For investors: Track tightly the capital flows into humanoid companies (Tesla/Optimus, Figure AI, Chinese robotics champions) and into the semiconductor and data centre supply chain. The ramp from prototypes to mass production remains the critical value-unlock window. Rising energy input costs may also make baseload power providers more attractive.
• For policymakers in ageing economies: Japan, South Korea and parts of Europe should accelerate regulatory sandboxes for humanoid deployment in eldercare and public services. Delaying frameworks could hand an irreversible lead to China, which is already coordinating state-backed robotics scale-up as a demographic remedy.