INVESTIGATION: 50,000 humanoid robots in China in 2026 — the revolution the West underestimates

Introduction: when Morgan Stanley revises upward

A near-doubling of projections in under a year

Morgan Stanley has revised its deployment forecast for humanoid robots in China for 2026: 50,000 units expected, up from only 28,000 anticipated in January of the same year. That is nearly a doubling in under six months. This is not a minor adjustment — it is the recognition of an acceleration that even the most optimistic analysts had not predicted.

This investigation seeks to understand why China is developing so rapidly in humanoid robotics, what this means for global industrial competitiveness, and why this revolution carries a strategic and military dimension that the West is only beginning to grasp.

Why "humanoid" is different from other robots

Industrial robotics — articulated arms, automated assembly-line systems — has existed for decades. What is new is the humanoid robot: bipedal, with arms and hands, capable of navigating environments designed for humans without modifying them. This form allows the robot to use the same tools, spaces and interfaces as human workers. The applications are potentially limitless — manufacturing, logistics, agriculture, personal care, and military uses.

The drivers of China's acceleration

Political support as an accelerator

China is not developing humanoid robotics by accident. It is a national priority written into five-year plans, funded through massive subsidies, backed by targeted industrial policies. The Chinese government has identified robotics as a strategic technology whose mastery is necessary to maintain industrial competitiveness as labor costs rise and to accumulate a dual-use civil-military technological edge.

This structure of political support makes the difference between an industry that develops on the market alone — slowly, following private investment cycles — and an industry propelled by public resources combined with private investment. In China, both combine with formidable effectiveness for the development of priority technologies.

The mature supply chain

One of the key factors explaining the acceleration is the maturity of the Chinese supply chain in critical components for robotics: motors, sensors, computer vision systems, lithium-ion batteries, embedded AI chips. China is the world's largest producer of most of these components. It can produce humanoid robots at a unit cost that American and European manufacturers cannot match.

Commercial validation: from promise to reality

Real deployments in industry

The upward revision by Morgan Stanley is not speculative — it is based on real deployment data. Chinese robotics companies have placed firm orders, factories have integrated humanoid robots into their production lines, large-scale tests have been successfully conducted in varied industrial environments. The technology has crossed the threshold of commercial validation.

This validation is crucial because it changes the market dynamic: when industrial customers trust the technology enough to use it in their real operations, they order more, economies of scale reduce costs, and the virtuous cycle of mass adoption kicks in. China has reached this inflection point in humanoid robotics.

The automotive sector as the first testing ground

The automotive industry — where China is now the global leader with companies like BYD — is one of the first sectors for mass deployment of humanoid robots. Electric vehicle assembly lines require a flexibility that traditional industrial robots cannot provide — line configurations change frequently to accommodate different models. Humanoid robots can adapt to these changes without requiring costly infrastructure modifications.

Air-space-land-sea networks for AI: MWC Shanghai

A communications infrastructure for AI robotics

At the Mobile World Congress (MWC) Shanghai 2026, Chinese telecom operators showcased their advances in integrated "air-space-land-sea" networks for AI. This communications architecture — combining 5G terrestrial networks, satellite communications, aerial links and undersea networks — is designed to provide permanent, high-bandwidth connectivity for distributed robotic and AI systems.

For humanoid robotics, this infrastructure is foundational: it allows robots to receive software updates in real time, share learning data across a fleet, and operate in a coordinated manner in large environments such as cities or extended industrial zones. This is the internet of robots in its most advanced form.

The military dimension of integrated networks

Air-space-land-sea networks for AI have direct military applications. A network connecting aerial drones, ground robots and naval systems to a central AI command infrastructure — that is precisely the architecture of a next-generation autonomous weapons system. What Chinese telecom operators present as commercial infrastructure is also a template for AI-coordinated, multi-domain military operations.

360 Security Technology versus Anthropic: the AI model war

Chinese AI models claiming parity

Chinese firm 360 Security Technology has launched AI models it claims are equivalent to Anthropic's Mythos, one of the American firm's most advanced AI models. If this claim is verifiable — and caution is warranted regarding self-assessments — it illustrates the remarkable progress of Chinese generative AI capabilities.

These models will feed directly into humanoid robots: the ability to understand natural language, reason about complex situations and generate adapted responses is what transforms a mechanical robot into a cognitive assistant usable in human environments. The convergence between large language models and physical robotics is the next technological frontier.

Anthropic suspends model access for foreigners

In response to national security concerns, Anthropic has suspended access to its Fable 5 and Mythos 5 models for foreign nationals, by U.S. government order. This restriction — rare and significant — acknowledges that the most advanced large AI models are technologies to be controlled in the same way as advanced semiconductors.

The implications for global security

The humanoid robot as a weapons system

A humanoid robot capable of manipulating objects, moving through complex urban environments and reasoning with advanced AI models is, intrinsically, a candidate for military applications. Reconnaissance, ammunition transport, mine clearance, defense of positions — humanoid robots could theoretically perform all of these tasks. The deployment of 50,000 units in China in 2026 creates an industrial and technological base from which a portion could be converted to military purposes with alarming speed.

This is the concept of dual use in its most complete form: a technology that is simultaneously an industrial production tool and a weapons system in waiting. Both aspects develop together, mutually reinforcing each other.

The Western gap: a window to close

Europe and the United States have humanoid robotics capabilities — Boston Dynamics, Figure, Agility Robotics in the United States — but deployment levels remain far below Chinese projections. The gap is not technological — Western robots are often more sophisticated. It is industrial and economic: the capacity to produce and deploy at scale, quickly and at low cost.

What this investigation reveals about Beijing's global strategy

Technological convergence as strategy

What this investigation reveals is that China's robotics revolution is not an isolated phenomenon. It is the fruit of a deliberately planned convergence between advanced chips (SMIC 5nm), AI models (360 Security), communications infrastructure (MWC networks), and industrial capital (JCET, BYD). All these pieces fit together in a coherent strategic vision aimed at making China the world's leading technological power by 2035.

This strategic coherence is one of China's most formidable assets in this competition. The West has more innovative companies, more creative researchers, more open universities. But it does not have this planning coherence that aligns public and private resources toward shared long-term objectives.

The necessary Western response

Faced with this Chinese technological convergence, the democratic West must develop its own strategic response. This does not mean copying the Chinese dirigiste model — that would betray the values of free innovation and open markets that constitute the deep strength of the Western economy. But it does mean more targeted industrial policies, public investment in critical technologies, and coordination among allies to avoid recreating in robotics the dependencies we are trying to undo in semiconductors.

Western investment in robotics: the gap that must be closed urgently

Boston Dynamics, Figure and the American players

The United States is not absent from humanoid robotics. Boston Dynamics, a Hyundai subsidiary, has developed bipedal robots like Atlas. Figure and Agility Robotics are working on commercial platforms. Tesla with its Optimus robot and NVIDIA with its robotic simulation models are part of this ecosystem. The technical capabilities are real and in some cases superior to Chinese equivalents in terms of raw performance.

But the industrial gap is real: China is set to deploy 50,000 humanoid units in 2026, while American deployments remain well below that. It is a gap of considerable magnitude in terms of actual deployment volume. Technology alone is not enough — the industrial capacity to produce at scale and at low cost is what matters. That is where the Chinese supply chain makes a decisive difference.

What Europe must decide now to avoid missing this turning point

Europe is absent from this debate in a troubling way. There is no major European humanoid robotics player. Industrial robots — ABB, KUKA, Fanuc — are established European and Japanese players. But the new generation of autonomous humanoid robots is American or Chinese. This absence is not trivial: humanoid robots will be vectors of industrial, economic and military power in the decades ahead.

Europe must decide now whether it wants to be a player or merely a buyer in this technological revolution. Choosing to be a buyer is comfortable in the short term but strategically destructive in the long term — exactly as dependence on Asian semiconductors has been, and which the European Chips Act is now trying to correct at great expense. Better to learn the lesson once than to repeat it endlessly.

Conclusion: 50,000 robots and a silent revolution

What the numbers actually say

The 50,000 humanoid robots projected in China in 2026 according to Morgan Stanley are not just a market figure. They are 50,000 demonstrations of China's capacity to move from technology to industrial scale. They are 50,000 learning platforms that collect data to improve the next generations. And they are 50,000 systems whose technologies could be adapted to military purposes according to Beijing's strategic needs.

The urgency of a coordinated Western response

The West still holds the cards to avoid losing this race. But time is short. Morgan Stanley's upward revision in June 2026 is an alarm signal — not for panic, but for urgency. The industrial democracies must treat humanoid robotics with the same strategic priority as semiconductors, energy or missile systems. Because in the world to come, robots will be vectors of power as much as tools of production.

Signed Maxime Marquette, columnist

Columnist's transparency box

This investigation is based on facts and data documented in the lot file, with dated and verifiable sources. The Morgan Stanley projections cited come from the referenced sources. No internal or confidential data was invented. The editorial passages in italics represent the columnist's opinion. The stated position is one of concern for the West's strategic security in the face of China's rising technological power.

Sources

Primary sources

Silk Bulletin no. 139: Morgan Stanley raises China humanoid robot forecast to 50,000 for 2026 — June 25, 2026

Silk Bulletin no. 137: JCET invests USD 1.15 billion in advanced packaging in Shanghai — June 23, 2026

ETC Journal: AI in June 2026 — Anthropic model access suspension — June 23, 2026

Secondary sources

Claw Street Journal: Frontier AI model access in China — June 21, 2026

CNBC: Chinese trade restrictions on American companies — June 22, 2026

South China Morning Post: Semiconductor geopolitics and advanced technology — June 23, 2026