Robots Atlas>ROBOTS ATLAS
Robotics & Hardware

Japan vs. China: who will win the humanoid robot race?

Japan vs. China: who will win the humanoid robot race?

Japan invented humanoid robots — WABOT-1 at Waseda University in 1973, followed by Honda's ASIMO series. Yet at this year's Humanoids Summit in Tokyo, Chinese humanoids outnumbered Japanese ones three to one. Some Japanese firms even used Chinese robots in their own technology demonstrations. IEEE Spectrum analyzed the causes of this shift and possible paths forward.

Key takeaways

  • At the 2026 Humanoids Summit in Tokyo, Chinese robots outnumbered Japanese 3:1
  • Japan's industrial robot density dropped from world No. 1 (1994–2009) to 5th place in 2024
  • South Korea reached 1,220 robots per 10,000 workers, Japan — 446
  • China installed 54% of all industrial robots worldwide in 2024 and holds approx. 2 million operational units
  • AIRoA (Japanese nonprofit) has collected 80,000 hours of data for VLA models as a response to the technological lag

Japan as a guest at its own summit

The atmosphere at the Humanoids Summit in Tokyo was telling. On stage appeared Geminoid HI-6 — the sixth-generation android twin of Osaka University professor Hiroshi Ishiguro, equipped with a large language model trained on his own writings. It was one of the few Japanese humanoids present at the event.

Japanese company Omakase Robotics demonstrated a modified Unitree G1 — a Chinese robot with additional hardware added by Japanese engineers. Its CTO Shuichi Nagao put it plainly: "In China, the government is pushing humanoid development. They didn't have this industry 20 years ago. The people driving it are in their 20s and 30s. It's a completely different mentality. Big players in Japan are still looking for use cases for humanoids. In China, they're already in mass production and reducing costs."

GMO AI & Robotics, a Japanese subsidiary of GMO Internet Group, is testing G1 robots in partnership with Japan Airlines for loading and unloading cargo containers at Haneda airport.

Numbers that speak for themselves

For decades, Japan held the top position globally in industrial robot density — the number of multipurpose robots per 10,000 manufacturing workers. It led from at least 1994 to 2009, then fell to second in 2014, third in 2019, and fifth in 2024.

South Korea reached a density of 1,220 robots per 10,000 workers versus Japan's 446. China installed 54% of all industrial robots worldwide in 2024 and maintains approximately 2 million operational units — about 4.5 times more than Japan.

According to McKinsey's Ani Kelkar, "the recovery has not yet happened as Japan missed the rapid acceleration in AI for robotics and is now playing catch-up."

How Japan can respond

Kelkar points to a $100 billion opportunity for Japan in general-purpose robotics. The country has hardware experience and high-quality components, but must shift strategy around AI, software, and robotics platforms.

Tetsuya Ogata, director of the Institute for AI and Robotics at Waseda University and chair of AI Robot Association (AIRoA), presented at the summit how the organization is building a shared data infrastructure for foundation models. AIRoA has collected 80,000 hours of data from remote operation of mobile manipulators and claims it is the largest dataset of its kind. Vision-Language-Action (VLA) models have been built and verified on this data.

The world of AI is inherently a game of scale. Japan's absolute prerequisite is to secure a competitive baseline of scale — in data, computing resources, and talent. But what I consider most critical is a mindset shift: rather than trying to hoard scale within a single nation or company, we must grow stronger by collaborating with a diverse ecosystem.

Tetsuya Ogata, AIRoA / Waseda University

Specifically, AIRoA proposes a shared "collaborative domain" for data — an open, pre-competitive infrastructure on which companies can build their own applications. An initiative open to global players would, according to Ogata, allow Japan to become a "third pole" alongside the US and China.

A historical warning

Japan launched the world's first mobile internet platform (i-mode) in 1999. Being first did not make it a smartphone leader — it became a component supplier to others. Will humanoids follow the same path? Japan's historical advantages — ASIMO, WABOT, a culture of precision hardware manufacturing — represent real capital. Unitree's G1 costs $16,000 and is commercially available. ASIMO, which defined the previous era, was retired in 2022, the same year ChatGPT launched.

Why this matters

The humanoid race is taking on a geopolitical dimension. The United States is debating the GUARD Act, which would restrict imports of Chinese robots. Japan, as the historical leader in industrial robotics, faces a choice: enter the race with a new data-driven, collaborative strategy, or remain a precision component supplier for Chinese and American ecosystems.

The critical industry question is whether Japan's hardware advantage — motors, gearboxes, encoders, actuators — will be sufficient in an era where the value in humanoids is increasingly created at the software and training data layer. Ogata is right: this is a game of scale, and scale cannot be built in isolation. The AIRoA initiative with 80,000 hours of data is a step in the right direction, but still modest compared to Chinese datasets backed by thousands of deployed robots.

What's next?

  • AIRoA has begun collecting data for dual-arm mobile manipulation models — results will inform VLA quality available to Japanese manufacturers in 2026–2027
  • The Japanese government and companies such as Toyota are exploring regulatory changes enabling bipedal humanoid deployments (currently blocked by safety rules) — deregulation timeline unconfirmed
  • Japan Airlines is piloting G1 robots for cargo handling at Haneda — pilot results expected within the year

Sources

Share this article

Related articles