Robots Have Become “Appliance Prices”

$4,370. Approximately 650,000 yen.

For this amount, you can purchase a bipedal humanoid robot on AliExpress. It’s the “R1” from the Chinese company Unitree.

650,000 yen is about the cost of a commercial large refrigerator or a used light van. Just a few years ago, humanoid robots were research equipment costing tens of millions to hundreds of millions of yen. Now, they can be added to an online shopping cart.

What does this “price collapse” signify? And is this truly an opportunity for local small businesses? Let’s take a closer look.

If We Only Consider Price, Robots Are Already Competitive with Labor Costs

First, let’s organize the numbers.

As of 2024, the average hourly wage for part-time workers in Japan is about 1,200 yen. Working 8 hours a day for 20 days a month results in a monthly income of approximately 192,000 yen. When adding the employer’s share of social insurance, the actual cost becomes about 220,000 to 240,000 yen per month, totaling around 2.64 to 2.88 million yen annually.

The Unitree R1 costs about 650,000 yen. Assuming a lifespan of 3 years and an estimated annual maintenance cost of 100,000 yen, the annual cost would be around 320,000 yen. This is less than one-eighth of the annual cost of a part-time worker.

From a purely numerical standpoint, robots are overwhelmingly cheaper.

However, jumping to “let’s implement this right away” is premature. The real question is, “What can this $4,370 robot do right now?”

Current Capabilities: “Being Able to Move” and “Being Useful” Are Different Matters

Looking at the specifications of the Unitree R1, it is said to be capable of bipedal locomotion, object grasping, and executing basic movements. Unitree has a proven track record with its quadrupedal robot “Go2,” lending some credibility to the quality of its hardware.

However, let’s be honest. Current humanoid robots are not at a level where they can be deployed in practical settings.

What is required in small business environments is a series of tasks such as “picking parts from shelves, inspecting them, and packing them into boxes.” This requires situational judgment, adjusting force, and handling exceptions. Expecting a $4,370 robot to perform these tasks is unrealistic at this point.

So, does this mean we’re looking at a situation that’s “still far off”?

Not at all. What we should focus on is not the “price” but the “rate of price decline.”

The Real Focus Should Be on the “Speed of Decline”

Reflecting on the price trends of robots reveals a structural change.

• 2013: Boston Dynamics’ “Atlas” — Research model, estimated price 100-200 million yen
• 2020: Unitree’s “A1” quadrupedal robot — Approximately $10,000 (about 150,000 yen)
• 2023: Unitree’s “Go2” quadrupedal robot — $1,600 (about 24,000 yen)
• 2025: Unitree’s “R1” humanoid — $4,370 (about 65,000 yen)

The price of quadrupedal robots has decreased to one-sixth over three years. If humanoids follow the same curve, they could enter the $1,000 to $2,000 range in 2-3 years. That’s 150,000 to 300,000 yen, equivalent to the cost of a few used computers.

Moreover, there are players accelerating this price decline. Tesla.

The True Impact of Tesla’s Optimus

Tesla is advancing the mass production of its humanoid robot “Optimus.” Elon Musk has stated that the company aims to start full-scale operations internally by 2025 and plans to sell externally starting in 2026. The target price is set between $20,000 to $30,000 (about 3 to 4.5 million yen), with future aspirations to bring it below $10,000.

The impact of Tesla is not just about the performance of the robot itself. It’s about the very structure of “applying automotive mass production technology to robotics.”

Tesla produces nearly 2 million cars annually. The design philosophy of its production lines—cost reduction through mass production, over-the-air software updates, and AI processing with proprietary chips—can be directly applied to robotics.

In other words, the moment Tesla starts mass-producing Optimus, the “price ceiling” for the entire robotics industry will drop significantly. Chinese manufacturers like Unitree will respond with even lower price points, accelerating price competition.

This mirrors what happened in the smartphone industry. The iPhone created the market, Chinese manufacturers shattered prices, and ultimately smartphones became something anyone around the world could own.

The Evolution of Software Will Transform “Incapable” into “Capable”

Even if hardware prices drop, if the robot is “incapable,” it’s meaningless. Here, the evolution of robot control AI becomes crucial.

Recently, attention has been drawn to technologies that generate vast amounts of operational data for robots in simulated environments. Methods like ComSim and AffordSim allow for the learning of thousands of object manipulation patterns in virtual spaces, which can then be transferred to real machines.

What does this mean? The “training costs” for robots will dramatically decrease.

Traditionally, teaching robots new tasks required hundreds of trial-and-error attempts with the actual machine, consuming both time and costs. With the advancement of simulation technology, this process can be completed in virtual spaces. Moreover, training data for one robot can be distributed to all robots.

This is akin to how ChatGPT has reduced the “training costs of language.” Once trained, all users benefit. The same is about to happen for robots.

What Should Small Businesses Do “Now”?

After reading this, you might think, “So, should we just wait for 2-3 years?”

That’s half correct and half wrong.

While it’s necessary to wait for the practical application of humanoid robots, it’s also crucial to become an organization that can utilize them when they arrive, and this can start right now.

Specifically, here’s what to do:

1. Break Down Your Operations

Divide your company’s operations into “tasks requiring judgment” and “tasks with established procedures.” Robots will initially take on the latter. Companies that cannot break down their tasks will not be able to implement robots, even if they become cheaper.

2. Eliminate “Dependency on Individuals”

Tasks that can only be performed by “that one person” cannot be transferred to robots. Standardization of manuals and procedures is essential. This may seem mundane, but it is a prerequisite for robot implementation.

3. Experiment with “Partial Automation” First

Even without waiting for humanoids, there are automation tools available now, such as collaborative robots (cobots), RPA, and AI image recognition. Collaborative robots from Universal Robots cost around 3 to 5 million yen, and used ones can be obtained for around 1 million yen. This will help accumulate experience in “robots and humans working together.”

The True Meaning of “Solving Labor Shortages”

Finally, I want to pose a question.

What does it really mean to “solve labor shortages with robots”?

It’s not simply about “robots working in place of humans.” It’s about “handing over tasks that don’t need to be done by humans to robots, allowing people to focus on tasks that only they can perform.”

What’s serious for small businesses in rural areas is that due to a lack of manpower for simple tasks, managers and skilled employees cannot allocate time to their core responsibilities—sales, product development, and customer relations.

If a $4,370 robot can take over just 3 hours of simple tasks per day, it would free up 60 hours of the manager’s time each month. If that time leads to acquiring just one new customer, the cost of the robot would be recouped in no time.

The collapse of robot prices is not just about “access to cheap labor.” It’s about “changing how managers and employees use their time.”

The $4,370 humanoid is, at this point, closer to a “moving demonstration unit.” However, what this price indicates is the fact that robots have become “something you can buy.” In 2-3 years, they will become “something you can use.”

When that day comes, the gap between companies that are prepared and those that are not will widen to an irreparable extent.

What needs to be done now is not to buy robots. It’s to organize the tasks that can be handed over when robots arrive.