Selling RMB 550 million like hotcakes, a 53-year-old female president takes her company public today: Her robots top the national sales list.

The alumni circle of Xi'an Jiaotong University is estimated to be in an uproar. Today, Zhi Tao, an alumnus of Xi'an Jiaotong University, officially rang the bell as Yunji, the company she founded, went public in Hong Kong with a market value of approximately 9.8 billion yuan. Zhi Tao is already 53 years old this year, which also shows that it's never too late to start a business, and even less so to go public.

Through the story of Yunji, let's take a look at what scarce opportunities there are for robots at present and in the future.

- 01 -

Anyone who has stayed in a hotel may be familiar with Yunji and have seen its service robots. When you order takeout, in the past you had to go downstairs to pick it up, but now Yunji's robots deliver it to you, doing the repetitive work that you don't want to do.

Although the public impression of robot entrepreneurship started from the Spring Festival Gala in 2025, in fact, the first wave of robot enthusiasm began around 2014 - even the well - known robot company Unitree Technology was founded in 2016.

Many people say that robots are just a gimmick, only capable of performing tasks like dancing and running, and it's difficult for them to solve real - world needs and for people to be willing to pay for them.

Actually, that's not the case. The commercialization of robots has been happening in the past 10 years, and there have always been people paying for them. Of course, robots are divided into many categories. Some categories have succeeded, some have failed, and some are still in the process.

For example, the service robots where Yunji operates have actually been successful, especially in the hotel scenario.

According to Frost & Sullivan, in 2024, Yunji Technology ranked first in the robot intelligent agent market in the hotel scenario with a 13.9% domestic share, exceeding the combined share of the second to fifth - ranked companies.

This is actually a bit puzzling: Why can robots succeed in the hotel scenario but not in others?

Which scenarios don't work well? For example, reception robots.

You should know that from 2014 to 2015, there was a group of entrepreneurs who didn't bet on robots in the hotel scenario but on reception (guide robots). But until now, no particularly large companies have emerged in this field.

Which scenarios work well?

First of all, scenarios related to "delivery" are more likely to achieve commercial success.

Yunji's products are essentially hotel delivery robots. What do they deliver? Takeout, guest - needed items (such as slippers and toothbrushes), emergency medicines, etc.

Although Yunji hasn't achieved profitability yet, the scale of its revenue has proven its necessity.

From 2022 to 2024, Yunji's revenues were 161 million yuan, 145 million yuan, and 245 million yuan respectively, with a cumulative total of 550 million yuan. During the same period, the company's net losses were 365 million yuan, 265 million yuan, and 185 million yuan respectively.

Another type of delivery robot that has achieved good commercialization is industrial delivery (heavy - load) robots.

There are many such examples. Seer Robotics, Mech-Mind Robotics, Standard Robots, etc., are all unicorn companies. They shuttle through various factories, liberating workers' hands and feet and handling all kinds of dirty, tiring, and difficult tasks.

Secondly, besides delivery, another type of robot that has achieved good commercialization is the sweeping robot.

Typical players include Dreame, Ecovacs, and Yunjing Intelligence.

In the second quarter of 2025, the top five global sweeping robot manufacturers were all Chinese companies, namely Roborock, Ecovacs, Dreame, Xiaomi, and Yunjing Intelligence. Together, these five companies accounted for nearly 70% of the global market share.

Another is educational robots. In the fiscal year 2024, the revenue from UBTECH's educational intelligent robots and solutions was approximately 363 million yuan, accounting for 27.8% of the total revenue.

- 02 -

We can find that in the Robot 1.0 era, the core characteristic of commercialization was to help humans do manual labor better. The more disliked the job by humans, the more room for robots to play.

For example, do you know about steel - making robots? In steel mills, there is a fixed job - the bottom - blown furnace cleaning worker. The bottom - blown furnace is a steel - making equipment that needs to be cleaned regularly (slag accumulates at the furnace mouth). If a human is sent to clean it, it would be really miserable.

You have to stand next to the bottom - blown furnace, and the temperature can reach over 50 degrees, which might make you faint. Even if workers wear heat - insulating work clothes and protective gloves, they will still feel extremely uncomfortable.

But the more such jobs there are, the more room for robots to play, especially those dangerous, uncomfortable, and jobs beyond human limits - these application scenarios are particularly suitable for the Robot 1.0 era (2015 - 2025).

Now, these things may seem easy to understand. But when you are in the great historical trend and face the huge unknown of the future, it's not that clear.

For a company like Yunji, which has firmly adhered to the hotel scenario for 10 years, it takes strong determination. First, you have to believe, and second, you have to persevere. Otherwise, as you go along, you'll start to think: Well, is it okay to switch to reception guide robots?

Well, the Robot 1.0 era is already in the past. After all, in all the scenarios we just listed, there are already giants: either these giants have gone public or are unicorns.

Moreover, today's technology is advancing rapidly, and the infrastructure is far from what it was 10 years ago. With the rapid progress of AI technology, it would be a bit of a waste to use today's technology to do what was done 10 years ago (such as delivery and sweeping).

Previously, Pencil News interviewed Yu Guangdong, the founding partner of Full - Speed Capital. He said that if there is still a team doing what was done in the 1.0 era today, no matter how favorable the conditions are, he won't invest.

So, in the next 10 years, in the Robot 2.0 entrepreneurship era, which specific sub - sectors are worth investing in? Yu Guangdong said that he is interested in robots with an AI brain.

What does it mean to have an AI brain? In the 1.0 era, we saw many robots painting walls, cleaning windows, painting ships, cleaning swimming pools, weeding, etc. If they don't have the attribute of "human wisdom", they can basically be defined as "brainless".

In the 2.0 era, "AI brain + robot execution" may be a big opportunity. As for whether these opportunities will continue in the delivery and sweeping scenarios of the 1.0 era, it's hard to judge.

What we can judge now is that the "AI brain" can already empower the scenarios of the 1.0 era.

For example, Yunji's products have already been implanted with some "AI brains", which can intelligently recognize the voice, text, image and other needs of guests, dynamically coordinate resources such as robots, employees, and equipment, and achieve automatic task assignment and full - link tracking.

Previously, Pencil News interviewed Dreame Technology. On the one hand, large - scale models have been implanted in their sweeping robots; in addition, they have also explored extended functions similar to those of a "cleaning aunt", such as folding clothes.

Perhaps from the exploration of Unitree Technology, we can try to summarize the new opportunities in Robot 2.0.

According to Reportify data, Unitree Technology's most core product at present is the quadruped robot, which accounts for about 65% of the company's revenue. And 80% of the quadruped robots are used for research, education, and the consumer field; 20% are used in the industrial field, such as inspection and fire - fighting.

From these scenarios, we can see that compared with the 1.0 era, the jobs of robots in the 2.0 era don't have a huge difference.

The reason behind this is that today's robot AI brains are still far from being developed. Even though the AI large - scale models have made rapid progress in the past 3 years, they were not invented specifically for robots. This "adaptation" process may take 5 years or even longer.

This long adaptation process also means more infinite possibilities.

This article does not constitute any investment advice.

This article is from the WeChat official account "Pencil News", author: Zhi Ba. It is published by 36Kr with authorization.