When humanoid robots frequently "go viral" on social media, the public's anticipation for them to enter households has reached its peak. However, the forecasts from industry leaders are more calm and pragmatic.

At the 2026 China Online Media Forum, Unitree Technology founder and CEO Wang Xingxing delivered a keynote speech titled "When Robots Go Viral." He pointed out that although technological progress is rapid, the true "GPT moment" for embodied intelligence may still take two to three more years.

What is the "GPT moment" of embodied intelligence?

In his speech, Wang Xingxing set clear quantitative standards for this milestone:

Adaptability in unfamiliar environments: When you bring a robot into a completely unfamiliar environment (such as an office or home you've never been to).

Task completion rate: The robot can autonomously complete 80% to 90% of complex tasks with only simple voice commands.

Natural interaction: The robot is no longer a machine rigidly executing programs but an intelligent entity that can understand ambiguous instructions and possess physical common sense.

Technology roadmap: Major breakthroughs in the next two years

Although reaching the ultimate goal will take time, Wang Xingxing remains extremely optimistic about the recent technological development:

The eve of the breakthrough: He expects that in 2026 or 2027, the field of embodied intelligence will experience a significant technological leap.

Mass production acceleration: In fact, Unitree Technology has set a target, expecting its robot shipments to reach around 10,000 to 20,000 units this year, showing that the industrialization process is accelerating.

Extreme cold challenge: Previously, Unitree's humanoid robot successfully completed the world's first autonomous walking challenge in extreme cold weather, walking 130,000 steps at -47.4°C, proving the high reliability of the hardware.

Industry ambition: Building a general-purpose humanoid robot foundation model

As a leading company in the robot industry, Unitree Technology is accelerating the construction of its technical moat. The company plans to release a "general-purpose humanoid robot embodied base model" within the next three years, aiming to equip robots with a "universal brain" that can understand the physical world through software algorithm breakthroughs. Wang Xingxing once joked that creating a truly applicable large model for robots would be as challenging and meaningful as winning the Nobel Prize.

Conclusion: From "going viral" to becoming a "life partner"