The humanoid robot race is China’s to lose

The United States remains stronger in advanced models and commercial deployment. Gulf states are building positions through capital, compute and public procurement. Neither challenger can yet match China’s output volumes — but volume alone will not decide the winner.

China turns demonstrations into industrial capacity

China’s advantage is no longer limited to public demonstrations. It now rests on the ability to produce, finance and deploy humanoid robots at scale.

Reuters reported that Chinese national, provincial and local governments spent at least US$230 million on humanoids and related items in the first half of 2026. That compares with US$62 million in the same period of 2025 and just US$6 million in the corresponding months of 2024. The acceleration is striking. Authorities are also expanding real-world training sites to generate operational data at scale.

The technology remains at an early stage outside controlled settings. Many machines still struggle with complex manipulation and reliable operation in variable environments. That gap matters for investors.

The sector has moved beyond a pure technology bet. Manufacturing yield, component costs, training data and customer retention now matter as much as locomotion performance. China has used a familiar playbook. Public authorities support strategic firms, local governments compete for projects, and suppliers build capacity before demand fully matures.

This approach helped expand China’s electric-vehicle and solar industries. The comparison has limits. Humanoid robots are more complex than electric vehicles. They must perceive changing environments, plan movements and handle objects reliably. They also need durable batteries, actuators, sensors and software stacks that work together.

Still, China holds a major structural advantage. Its robotics firms operate close to large electronics, machinery and battery supply chains. That proximity lowers costs and shortens development cycles. The investment thesis has therefore shifted. The central question is no longer whether humanoids can perform impressive demonstrations. It is whether companies can convert those demonstrations into repeatable enterprise use.

Why China reached the front of the field

China’s policy support accelerated sharply during 2025 and 2026. Ministries and local authorities began linking humanoid robots with embodied artificial intelligence, advanced manufacturing and industrial automation.

On 3 June 2026, the general offices of China’s Ministry of Industry and Information Technology and the State-owned Assets Supervision and Administration Commission jointly issued a notice launching a 2026 real-scene training campaign for humanoid robots and embodied intelligence. The notice set goals of more than 100 high-value scenarios and deployment capability at the 10,000-unit scale by the end of 2026.

In April 2026, State Grid reportedly planned approximately 6.8 billion yuan in total investment for about 8,500 embodied-intelligence systems, including 5,000 quadruped inspection robots, 3,000 dual-arm inspection robots and 500 humanoid robots. These deployments may not offer immediate commercial returns. They support product refinement, safety testing and data collection. For early-stage hardware firms, that government-backed demand provides a valuable bridge to private customers.

China’s lead is therefore a scale advantage, not only a research advantage. The companies that convert that scale into reliable systems will capture the strongest returns. As one analyst framing the sector succinctly puts it: robotics valuations now depend on useful work, not impressive movement.

The data behind the humanoid robot race

Funding activity shows how quickly investor interest has expanded. Morgan Stanley expects the average price of a humanoid robot to decline by 15% in 2026, before recovering modestly thereafter. Lower prices support adoption. They also pressure manufacturers before software revenues become meaningful.

Xinhua reportedly quoted Gan Xiaobin, a senior Ministry of Industry and Information Technology official, as saying in July 2026 that China’s humanoid-robot production was expected to exceed 100,000 units in 2026. The forecast sits well ahead of confirmed global shipments from 2025. Investors should therefore distinguish between production capacity, announced output and paid commercial deployment. Those are three very different numbers.

Funding rounds across the sector are large relative to current commercial revenues. They show investors are backing manufacturing capacity and software development. They also raise the risk of excess capacity if enterprise demand develops slowly. The strongest candidates will show repeat orders, falling intervention rates and improving unit economics.

How the US is approaching commercialisation

The United States retains important strengths in artificial intelligence models, software and venture capital. Its companies have focused more heavily on advanced autonomy, industrial pilots and commercial partnerships.

Agility Robotics is a prominent example. Its commercial progress will depend on production reliability, customer contracts and the cost of servicing deployed machines. A public listing would give investors a direct view of a humanoid developer — but it would also expose the company to quarterly scrutiny before the market has established clear revenue benchmarks.

US firms benefit from a strong technology ecosystem. Cloud companies, chip designers and foundation-model developers can provide the computing and software needed for robot training. However, the US faces higher manufacturing costs than China. It may need overseas suppliers or new domestic factories to achieve competitive unit economics.

That creates a strategic divide. China is building a broad hardware base and large production runs. The US is concentrating on advanced models, integration and higher-value applications. The division may not remain permanent. Chinese firms are improving their software rapidly. US companies are likely to increase manufacturing investment. Both countries will compete across the full stack in time.

Commercial deployment will decide the outcome. Warehouses, factories, logistics sites and hazardous environments offer the clearest early markets. These settings provide controlled workflows and measurable labour savings. Household robots remain a harder proposition. Homes contain more unpredictable objects, more variable layouts and higher safety requirements. They also lack the clear return-on-investment measures found in industrial facilities.

Gulf states build strategic positions

That transition favours data centres, deployed computing and automation. It also creates opportunities for robotics firms that can connect AI with physical operations. The Gulf’s advantage lies in coordinated investment. Governments can direct procurement, land, energy and infrastructure towards strategic projects. Sovereign capital can support long development cycles that private markets find difficult to fund.

The constraint is local depth. Gulf states do not yet match China’s robotics supply chain or the United States’ software ecosystem. They will need partnerships with foreign firms and local workforce development to close that gap. The strongest Gulf strategy may involve becoming a deployment market and infrastructure hub. It need not produce every robot domestically to capture significant value from the sector’s growth.

What does the race mean for investors?

Investors should separate three layers of the industry. The first is the hardware layer. It includes actuators, motors, batteries, sensors, controllers and structural components. These suppliers may gain from wider adoption regardless of which robot brand wins market share.

The second is the intelligence layer. It includes training data, simulation, computer vision, control systems and foundation models. This layer could produce higher margins, but it faces intense competition from well-capitalised technology firms in multiple countries.

The third is the deployment layer. It includes fleet management, maintenance, integration and sector-specific software. This may become the most durable source of recurring revenue across the industry. No single region currently controls every layer. That creates room for cross-border partnerships and specialised suppliers to establish durable positions.

Governments gain productivity tools and strategic autonomy. They also face procurement risks if early systems deliver limited value. Manufacturers gain access to capital and testing environments. Yet they face warranty costs, safety obligations and uncertain production yields. Investors gain exposure to a potentially large automation market. They also face unclear exit paths, rapid price declines and weak near-term revenue visibility.

The most exposed firms are those with high fixed costs and limited customer concentration. A company may raise substantial funding yet fail to achieve profitable production. The strongest candidates will secure repeat orders, reduce human-operator intervention and report improving unit economics quarter by quarter.

Outlook: from spectacle to operating proof

Output growth alone will not establish industry leadership. The decisive test will involve paid hours worked, task completion rates, maintenance costs and customer renewals. China’s Ministry of Industry and Information Technology target of more than 100,000 humanoids in 2026 signals production ambition. It does not guarantee equivalent commercial demand.

The United States may advance more slowly in volume. Its firms could still capture significant value through higher-performance systems and software-led business models. The Gulf’s progress will depend on infrastructure execution. Regional data-centre expansion and public procurement could support large-scale pilots that accelerate regional adoption.

Three signals deserve close attention from investors. First, track the share of deployments outside government demonstrations. Second, monitor the percentage of tasks completed without human intervention. Third, compare robot prices with the labour costs they replace. A fourth signal will come from component suppliers. Falling actuator and battery costs would improve adoption broadly. Persistent component bottlenecks would delay mass deployment.

The sector also needs better corporate reporting. Companies should disclose units shipped, units active, revenue per robot and customer retention. Without those measures, valuation comparisons will remain fragile and speculative. The US retains a credible path through software and industrial integration. Gulf states can shape the market by funding infrastructure and buying working systems at scale.

The next phase will reward operators that convert capital into reliable machine hours — investors should watch commercial deployment rates, price curves and data-centre commitments as the humanoid robot race enters its most consequential stage.

Quick answers