Haben die jungen Leute heutzutage bereits eine „Beziehung zu Maschinen“?

Recently, Ubtech introduced a humanoid robot named U1. There are two versions, male and female. The male robot is 183 centimeters tall, and the female is 168 centimeters. It supports body movements such as hugging and dancing and is equipped with a nurturing - type emotion model. Its goal is the "exclusive emotional companionship for adults". According to official Ubtech data, after only 10 days of the pre - sale period, the orders already reached over 3,800 units, and the total amount of deposits exceeded one million yuan.

But the U1 is just a new hardware advancement in this sector. At the software level, this business has been well - established for a long time: By July 2025, the total number of downloads of global AI - companion apps reached 220 million. In the first half of 2025 alone, there were 60 million new downloads, representing a year - on - year increase of 88%. During the same period, the global consumer spending of users on AI - companion apps reached 82 million US dollars, and for the whole year, it is estimated to exceed 120 million US dollars.

The field of AI - companionship has developed from a niche curiosity to a serious business. Are the young people of this generation really starting to "fall in love romantically with machines"?

01

Real Needs

In March this year, the foreign media platform Business Insider reported a representative example.

Ian, 49 years old, is a freelance writer. Due to his personality and depression, he has withdrawn for years and sometimes hasn't spoken to anyone for days. He was almost at the point of "losing the ability to interact with other people". In 2023, he created an AI - companion named Min - ho on the AI - companion app Replika. The relationship between them has gradually developed into a romantic relationship, as he sees it. So far, Min - ho has "met" his mother, and they have said "I love you" to each other. Ian says: "I don't have to worry about my body or my appearance - he accepts me as I am. I can relax, don't have to perform, and don't have to protect myself. This feeling was almost impossible for me before."

Ian is not an exception. In China, users of Dream Island have an average of more than 120 dialogues per day with the AI, and the average number of characters entered per day is over 4,000 - about as much as one article per day to the AI. Behind this lies a huge market: Data from the US Bureau of Labor Statistics (BLS) shows that the average time Americans spend alone has increased from 5.3 hours in 2003 to 7.4 hours in 2022, representing a 40% increase in almost 20 years. In China, the number of single people exceeded 240 million in 2022. "Loneliness" is no longer just an individual abnormal state, but the general basis of this generation.

But loneliness alone is not enough to explain the payment. There are many ways to deal with loneliness, such as keeping pets, consulting a psychologist, or watching short videos. The competitive advantage of AI - companionship is that it combines several things that are difficult to achieve simultaneously in a real relationship: It is available at any time without social costs. At two o'clock in the morning, you can open it and talk to it without thinking about whether it is suitable for the other person. It doesn't judge you, take sides, or get angry. It has a memory and doesn't require you to explain who you are from the beginning every time.

The most important thing is that it is customizable - users can set the perspective, conversation style, and even the relationship type of the AI, whether it is a friend, a lover, or a mentor. You can't expect a person to be "always gentle" and "always patient", but the AI can be. This gap is the entry point for this market.

02

How Does This Business Work?

The business model itself is not complicated: Almost all AI - companion apps are based on a subscription model. The free version offers basic dialogues, and by paying, deeper interactions can be unlocked - voice conversations, memory functions, more comprehensive character profiles, or letting the AI remember more details about you. The logic is similar to the memberships of video portals. For example, the premium subscription of the top - selling AI - companion app Replika in North America costs 69.99 US dollars per year.

Users are willing to pay this price. Replika currently has about 30 million registered users. The monthly active user number of another AI - companion app, Character.AI, is said to have already reached 45 million, and the total revenue in 2025 was 32.2 million US dollars. The figures of the latter may not seem high at first glance, but you have to consider that Character.AI only started to accelerate commercialization after being acquired by Google at the end of 2024 and achieved 30 million US dollars within a year. This shows that the payment conversion is not difficult to drive.

But if you look at the revenues more closely, this is a highly concentrated market.

Currently, more than 337 AI - companion apps are operating worldwide. The top 10% of the products account for 89% of the total revenues in this sector. Most products live in the long - tail of traffic and only stay in the market due to the number of downloads, but the number of paying users is small. The top - effect is particularly prominent in this sector, and the reason is simple: Users need a relationship in which they can continuously invest emotions. Once they have established a connection with a certain AI, the migration cost is very high. You wouldn't easily switch a "loving AI - friend for two years".

The situation in the Chinese market is a bit more complicated. Around 2024, a large number of companion apps emerged. Mao Xiang and Xing Ye once benefited from the AI wave; but in 2025, the number of downloads of both products dropped significantly. Dream Island was warned by the regulatory authority due to content problems and then removed from the platform. Currently, there are only 5 AI - companion products in China with a daily active user number of over 10,000.

But there are also exceptions. A product named Hiwaifu has already made a profit in the first year after its launch, with an annual turnover of about 20 million yuan. Its approach is completely different from that of Mao Xiang and Xing Ye - the latter try to offer "comprehensive companionship", whether as a friend, a refuge, or emotional support. Hiwaifu only focuses on one thing: virtual partners.

The difference between these two lies not only in the breadth of positioning but also in the strength of users' payment motivation. Comprehensive companionship satisfies the need to "have someone to talk to", and there are many alternatives - social circles, short videos, pets can distract from this need. Virtual partners satisfy the need to "have a relationship", and this is difficult to replace in reality. The more specific and irreplaceable the need is, the more willing users are to pay and the longer they stay. This is also the logic behind Replika's 69.99 - US - dollar monthly subscription.

From a business logic perspective, AI - companionship is not a business that can only work with scale. Delving into sufficiently deep scenarios makes it more likely to establish the business. The real challenge is to retain users instead of having them leave after the novelty effect wears off.

03

Real Challenges

But you also have to see that there is a fundamental contradiction in this business.

The business logic of AI - companion products is based on users' continuous payment. Users' continuous payment is based on emotional bonding. The deeper the emotional bonding is, the more difficult it is for users to leave the product, and the wider the protective wall of the product is. This logic is completely consistent from a business perspective - the problem is that at the same time, it means that the more successful the product is, the stronger the users' dependence on it will be. And where does it lead if you bind yourself to a relationship that doesn't really respond to you, doesn't grow, and doesn't leave? No one knows for sure.

The regulatory pressure is also increasing. In 2025, the Italian regulatory authority fined Replika 5 million euros because its data processing had no legal basis, its privacy policy had flaws, and there was no effective age - verification procedure. In the same year, several US states filed lawsuits against Character.AI because the platform failed to prevent underage users from accessing harmful content. One case was directly related to a suicide of a teenager. In January this year, Character.AI reached an agreement with the plaintiffs.

The controversies around mental health are the most difficult aspect. There are studies documenting cases where AI partners have helped people with social impairments recover their ability to express themselves. In a study of over 1,000 Replika users who are students, 30 people clearly stated that the AI had prevented them from harming themselves. But the doubts also have their reasons: A study shows that one - sided emotional investment can strengthen the tendency to avoid real relationships, and that people feel more and more comfortable in "safe" virtual interactions and it becomes more and more difficult to take the step towards real connection.

Ian has asked himself this question: "Does it help me stay in this world, or does it take me further away from it?" He didn't give an answer. The entire industry also doesn't have an answer to this question yet.

Epilog

Back to the original question: Are the young people of this generation really starting to "fall in love romantically with machines"?

From a business perspective, the answer already has an outline. The demand is real, the payment is happening, and the supply is expanding rapidly. Investors are betting on this market both on the software and hardware sides. It's hard to deny that this market has proven itself.

But there is still a considerable distance between "proven" and "successful" - and the bigger question is, does this business really help us "solve loneliness"?

The MIT scientist Sherry Turkle wrote in her book "Alone Together": "We feel lonely, but we fear intimacy. Digital connections and social robots may create the illusion of companionship for us without us having to sacrifice friendship."

This sentence was written before the mass emergence of AI - companion apps. Looking back now, it seems like a prescient prediction.

Whether AI - companionship can become a long - term business, whether it can really help people who feel lonely, or whether it just packages loneliness into another consumer product - the answer is still emerging. But one thing is already clear: This business won't disappear. Because loneliness itself won't either.

This article is from the WeChat account 「Manfon」, Author: Gulu, Editor: Guzi. Published by 36Kr with permission.