The market can't suppress its urge to crunch the numbers, but meals are still eaten one bite at a time.

Yesterday noon, rumors suddenly circulated that Tesla had placed an order for linear actuators for its Optimus humanoid robots with Sanhua Intelligent Controls, valued at approximately 5 billion yuan, with plans for delivery in the first quarter of 2026. Industry estimates suggest this order volume is sufficient to produce at least 180,000 Optimus units.

Influenced by these rumors, Sanhua Intelligent Controls' A-share stock surged straight to the daily limit up in the afternoon, with total trading volume approaching 12 billion yuan.

However, in an announcement released in the evening, Sanhua Intelligent Controls stated: After verification, the rumor is untrue; the company also has no major matters that should have been disclosed but were not; and on October 15, 2025, the company did not accept interviews from any media outlets.

Sanhua's direct clarification brought this mix-up to a close.

5 billion hit the market's impatient heart

Actually, over the past two months, discussions within the industry regarding whether robots can truly be mass-produced and when they will achieve large-scale deployment have become increasingly intense.

The current state of the robot industry is completely different from expectations at the end of last year or early this year. Before the end of last year, we rarely saw humanoid robots performing dances or demonstrating household functions; people held a relatively curious attitude toward humanoid robots at that time. Therefore, videos showcasing these robots' stunts often received positive feedback in the market, with some robot stocks achieving three- to five-fold increases within just a few months. Even in the primary market, many robots had already reached valuations of hundreds of billions of yuan.

However, after seeing various robot demos for half a year, people have likely grown somewhat fatigued by them. Although some robots can now enter factories to work and have secured some orders, this has not yet fully justified the inflated valuations. The market's expectations for robots have shifted to when they can achieve large-scale mass production and deployment—essentially, when they will start generating revenue.

In a recent report on Tesla AI, Musk predicted that Tesla's robots would reach an annual production of one million units before 2030. Additionally, the latest Figure 03 robot has set a goal of producing 100,000 units over four years. Major companies are providing mass production guidance, and many in the industry are claiming that the 'year of robot mass production' has arrived.

Against this backdrop, news of a '5 billion yuan order' undoubtedly hit precisely at the market's psychological expectation for the first shot fired in mass production and deployment. The market inevitably began calculating the core supply chain components such as actuators, joints, and dexterous hands. If true, this would be a landmark event in the development of the robot supply chain. However, clearly, this is not realistic at present.

It is undeniable that humanoid robots have significant room for future deployment. But the question is: have robots really reached a stage where they can accept a massive 5 billion yuan order without regard for technological maturity?

Hold on, the road for robots is still long

The progress of robots in the past two years has been rapid, with AI being the most crucial driver. We can now see robots performing tasks that were previously unlikely, such as dancing, boxing, and doing household chores, thanks to embodied models. However, the biggest challenge in bringing robots to market today is precisely that embodied models are not keeping up.

Robots' performance in motion is already quite impressive; at the recent World Humanoid Robot Games, they demonstrated running endurance comparable to humans. Yet, in practical applications, robots suffer from slow reaction speeds, particularly lacking robust autonomous work capabilities, which limits the scope of household chores they can perform. Moreover, in industrial settings, robots have not yet reached a level where they can fully replace human labor.

Wang Xingxing stated earlier at the WRC (World Robot Conference): "The current development of embodied intelligence is completely lagging behind, which is the key factor limiting the large-scale application of robots, especially humanoid robots."

Therefore, increasing investment in the model side is what the global robotics industry has been doing recently. Even in terms of data, the entire robotics sector lacks a perfect solution, with everyone facing challenges regarding how to collect high-quality data, establish data quality standards, improve model utilization of data, and decide between real-machine versus synthetic data.

Collecting data and training models are extremely costly technical hurdles.

Figure, which recently announced a $1 billion financing round, has launched the Go-Big project to build the world's largest pre-training dataset for humanoid robots. This could raise the ceiling of its proprietary VLA model, Helix, but few manufacturers currently have the capability to do so.

Currently, both models and data are in relatively early stages, and they are far from mature enough for large-scale application of robots.

So, is the hardware ready? No, it is not yet; at least it is far from reaching the stage where major companies would place orders worth 5 billion yuan.

The current robot industry needs to implement practical applications to continuously improve and adjust software and hardware details through these applications. This approach is sound, after all, robots cannot remain solely in laboratories.

Therefore, during this process, some robots have already entered factories ahead of schedule and secured partial orders. However, based on recent market performance, most single orders for robot bodies in actual applications are around 100 million yuan.

From a supply chain management perspective, it is hardly realistic for body manufacturers to take the risk of placing component orders worth billions of yuan. Even with Tesla having set mass production targets, the specific details of its third-generation robot have not yet been announced, so how likely is it that they would directly place large component orders?

Current robot hardware is far from converging; many solutions are being continuously improved following repeated practical applications. During this process, if certain hardware components are eliminated due to verification failures in applications, wouldn't previously placed large component orders become burdens?

So, if you look at it from the perspective of supply chain management, in the current context of the robotics industry, a 5 billion yuan order naturally lends itself to an intuitive assessment.

Take It One Bite at a Time

Humanoid robots are the future, and this is beyond doubt. However, we are still only at the early stage of the robotics industry's development. Therefore, there is no need to rush the calculations; taking it slower is healthier.

Ten years ago, when we thought about robots, we mostly saw them on television. Now that robots can truly appear before us, people naturally have strong expectations for their mass production, implementation, and large-scale application. This has allowed unrealistic order news to take advantage of the situation.

Currently, one should not expect the mass production of robots to happen overnight. The industry must proceed step by step, seeking the direction of hardware-software convergence within a closed loop of continuous application verification and improvement. Such a path of industrial development will be healthier.

Perhaps we need to get used to the fact that future robot orders will be characterized by relatively small amounts but higher frequency—advancing through verification. Over time, you will realize that we have already taken a significant step forward.

This is what 'learning by doing' means.