Robots’ muscles and brains: Will China have the lead?
While Wall Street questions its AI bet, the contest between China and the US looks very different in robotics
THE retreat in US technology shares after the midterms has revived a familiar question. Are expectations for artificial intelligence running ahead of returns?
The sell-off spread to Chinese and Hong Kong shares as investors reassessed the AI trade. In its initial public offering filing, Anthropic argued that AI could transform the economy on a scale greater than electricity or the Internet.
The bills for chips, power and data centres are due now, and investors want to know how long they will wait for that transformation.
That debate is mostly about software and the infrastructure behind it. Much less attention goes to the part of AI that works in factories, warehouses and ports. There, China and the US are far more evenly matched.
The physical AI arena
Our Essec Digital Disruption Matrix 2026 compares countries on research output and on unicorns, startups valued at US$1 billion or more.
In generative and agentic AI, the US leads on both. Robotics, however, looks different. China publishes about twice as much robotics research as the US.
In descriptive AI, the pattern recognition that helps machines make sense of what they see and sense, China publishes about 1.7 times as much as America.
A useful robot needs both: the mechanics to act and the intelligence to interpret its surroundings. China is working on the muscles and the brains at the same time.
Publication counts measure volume more than quality, so research leadership is only part of the picture. Companies are the other part.
The US still leads, with just under half (48.9 per cent) of the robotics unicorns we track, and Chinese firms hold close to two in five (38.3 per cent). Between them, the two countries account for almost nine in 10.
China has built a serious company base alongside its research output. Each country leads on a different measure, and the contest in physical AI remains open.
Where the payoff may come sooner
New technology rarely pays off straight away. A company has to train people, change how work gets done and find out which uses are worth keeping.
Measured productivity often dips before it rises, a pattern economists call the “productivity J-curve”. The dip tends to be shorter where the process is already measured and people already know the equipment. On this measure, China starts with an advantage.
China has the world’s largest market for industrial robots. Its factories employ engineers who know how to fit a machine into a production line and who have years of experience judging whether it earns its keep.
A factory that adds AI to a robot it already runs is improving a process it already measures, with people who already maintain it. Replacing an existing robot with a better one can deliver quick gains without directly displacing a worker.
An office introducing AI agents faces a harder path. The process is often less clearly measured, and the agents take over work that people do today. The question of what happens to those jobs can slow adoption.
The US bet rests heavily on this white-collar route, which could pay off across a wide range of industries if it works. China’s strength in robotics places more of its bet on the factory floor, where gains are easier to measure and may arrive sooner.
Many businesses have yet to factor in these considerations. In our survey, fewer than half of professionals (43 per cent) expected robotics to affect their industry now or soon. For descriptive AI, the proportion was higher (91 per cent).
The global conversation has centred on chatbots and agents. Yet, the timing of the payoff depends on the equipment, skills and processes a business already has in place.
Singapore in between
For Singapore, this is a reason to look past the daily swings in technology stocks.
Many local businesses run factories, warehouses and port operations, where physical AI may deliver gains they can measure sooner than office automation.
At Tuas Port, driverless carriers already move containers between the wharf and the yard. The economies those containers connect are making different bets on how AI will create value.
The US leads on models and agents, while China combines machines with the intelligence to run them.
Singapore trades with the two giants and can learn from both. The question is who can connect intelligence to machines, integrate those systems into industry and turn technical progress into reliable economic value. On our evidence, that race is still open.
Both writers are from Essec Business School. Jan Ondrus is professor of information systems and Jeremy Beaufils is executive director of the Essec Digital Disruption Chair.
TRENDING NOW
Ex-Nvidia CEO Jensen Huang carried a secret for more than a year. It led to his downfall
South-east Asia has lost Nvidia’s best salesman
Demand, tech capabilities position Apac and AI as Skyscanner’s next growth frontiers: CEO Bryan Batista
Jensen Huang discusses comeback as Nvidia CEO with allies: report