Source: Caixin.
Author: Zhang Yangyang.

"There are already two Optimus robots in use at the Fremont plant in California, and there will be more than 1,000, and even thousands of Optimus robots working for Tesla starting next year."

This is the latest data revealed by Tesla CEO Elon Musk about the application of Optimus (Prime) humanoid robots by Tesla. He believes that humanoid robots will become a consumer product that everyone can have in the future.

"Who doesn't want to see C-3PO (a character in Star Wars)? That would be great. I think everyone would want one. So I think the ratio of humanoid robots to humans may be at least 2:1, maybe even 1:1. So this means that the number of humanoid robots in the future will reach about 10-30 billion, or maybe 20-30 billion or even 300 billion."

At the annual shareholder meeting, with the story of humanoid robots, Musk also drew the largest pie in Tesla's history: Optimus alone could bring a market cap of $20 trillion, while the market cap of autonomous driving cars could reach $5-10 trillion. It is possible for Tesla to achieve a valuation 10 times that of the most valuable company today.

In the humanoid robot race, Musk is like an emotional barometer. Every time he reveals information about Optimus Prime, he can boost market sentiment. However, when Musk once again imagines the grand vision of Optimus Prime, the A-share humanoid robot sector sees a rare leading decline.

In terms of creating economic value and serving the high-quality development of mankind, humanoid robots are creating an irreplaceable product myth, which has boosted the market. Landing is the value standard for measuring all products, but in landing, Tesla's Optimus Prime and domestic humanoid robots are taking two completely different paths.

Song of Ice and Fire in humanoid robots entering the factory

Compared with Musk's confidence in the landing of Optimus Prime, the current landing of domestic humanoid robots is still a challenge, which is in sharp contrast to the industry's full expectations.

"To be honest, the pressure is very high." When asked about the impact of the current situation in the industry on his work, Wang Chun, a delivery engineer at Usu Technology, described his mood to a reporter from the Star Market Daily.

Wang Chun's pressure comes from the fact that humanoid robots are catalyzed to overheat, leading to high market expectations. If the actual performance of robots is greatly different from everyone's expectations, it may disappoint consumers and affect industry development.

The last time the Star Market Daily reporter met Wang Chun, he was in a meeting with NIO's technical staff, which lasted more than an hour. He described the meeting as a fierce confrontation and said,"I'm trying to align their understanding of the plump ideal and the bony reality of robots."

In the view of Lu Hanchen, director of the High-Tech Robot Industry Research Institute, the core points that cannot be avoided in the current landing path of humanoid robots are technology, cost, and scene. The reason why Musk can boast that there will be thousands of robots working at the Tesla factory by 2025 is that he has no obstacles in acquiring and training scene data in Tesla's factories. However, the current situation in China is that there are no robots in the scenes, or the robots are not capable enough. They have robots but no scenes.

"It is not necessary to be too pessimistic about China." Lu Hanchen believes that it still takes time, as long as there is a successful sample, it can usher in a breakthrough in landing.

Building an ecosystem is the key.

The automotive industry is currently one of the most potentially landed areas for humanoid robots. Since this year, market information about humanoid robots entering the factory has also been released. Previously, UBTECH's humanoid robots had entered the factories of both NIO and Dongfeng Liuzhou Motor.

Wang Chun told the reporter of the Star Market Daily that the first universal humanoid robot Unitree H1 and a four-legged machine dog made by Usu Technology have entered the final assembly shop of NIO's factory and are combined to automate the material picking and delivery work on the production line. Currently, both devices are in the debugging and training phase.

The automotive industry has high expectations for robots and hopes they can increase production efficiency and service experience. Wang Chun told the Science and Technology Innovation Board Daily that vehicles and robots share similarities in intelligence, such as autonomous driving and human-machine interaction. However, current humanoid robot technology is still in its infancy, largely facing issues with hardware maturity, stability and balance, making it difficult for robots to operate autonomously in complex situations. The software background, integration, and application of humanoid robots also need further improvement.

"The application of robots requires a complete solution, including the coordination between hardware and software and the support of the entire ecosystem," said Wang Chun.

Building an ecosystem for humanoid robots is already a consensus. At the recent 2024 China Humanoid Robot Developers Conference, Fourier Intelligence CEO Gu Jie said that the development of humanoid robots must keep up with upstream and downstream cooperation. In the hardware platform, Fourier has already opened up some capabilities, and there will be a large number of application scenarios in the next step that can be developed based on Fourier robots, hoping to build an ecosystem in the future.

In terms of the application of the automotive industry, Wang Chun mentioned that just like the development of drones which gave rise to various accessory manufacturers and scene experiment bases, a series of software and hardware integrators, service providers and application developers need to be developed around the robot ontology, forming a complete ecosystem. Automotive manufacturers can play an important role in this process because of their deep technical accumulation in vehicle interfaces and autonomous driving technology, and can quickly transplant these technologies to the robot platform.

Editor/Jeffy