Juuuno
All research

The connection economy

AI Companions and the Next Social Layer: Opportunity, Risk and the Case for Human-First Design

AI companion apps have grown from a novelty into a fast-growing category, with hundreds of revenue-generating apps and tens of millions of users. Most US teenagers have already tried one. This article looks at the data on the category's growth, the risks it raises, and why the most valuable companion products will be the ones that help people connect with other people, not replace them.

A new kind of relationship

For most of the history of consumer software, technology connected people to other people: messaging, social networks, dating apps. A new category changes that. AI companion apps offer a relationship with the software itself, a character that listens, remembers, responds and is always available.

The category is growing quickly, and it's growing in the middle of a loneliness crisis. The World Health Organization estimates that 1 in 6 people worldwide is affected by loneliness, with the highest rates among teenagers and young adults [1]. It's no coincidence that the people most likely to feel lonely are also the heaviest users of AI conversation.

How big the category is

Revenue and apps

Data from Appfigures, reported by TechCrunch in August 2025, gives the clearest picture of the companion market [2]:

  • AI companion apps generated $82 million in the first half of 2025, and were on track for more than $120 million for the full year.
  • There were 337 active, revenue-generating companion apps worldwide, of which 128 launched in 2025 alone.
  • The market is highly concentrated: the top 10% of companion apps generated 89% of category revenue.

The same reporting noted that revenue per download had more than doubled compared with 2024 [2], which suggests users are increasingly willing to pay.

Users

Individual companies report large audiences. Character.AI told TechCrunch in 2025 that it had around 20 million monthly active users worldwide [3]. Replika's website claims more than 42 million users worldwide [4], though that's the company's own count and not independently verified.

The wider AI context

Companion apps sit inside a much larger shift in how people use AI on their phones. Sensor Tower's State of Mobile 2026 reports that generative AI app downloads doubled in 2025 to 3.8 billion, in-app purchase revenue nearly tripled to more than $5 billion, and people spent 48 billion hours in generative AI apps, about 3.6 times the 2024 total [5].

Companion apps are a small share of that spending, but they're among the most emotionally significant uses of AI.

Teenagers got there first

The most striking data point concerns young people. A study by Common Sense Media and NORC, reported by TechCrunch in July 2025, found that 72% of US teenagers had used an AI companion [6].

That aligns with the WHO's finding that loneliness is highest among teenagers, with 17% to 21% of people aged 13 to 29 reporting feeling lonely [1]. It also raises the stakes. If a large share of young people are forming habits of relating to software during the years when they learn how to form friendships, the design of that software matters enormously.

Why people use AI companions

Research and user reports point to several motivations:

  • Availability. A companion is there at 3 a.m., on a bad day, when friends are busy.
  • Low risk. There's no fear of rejection, judgement or embarrassment.
  • Practice. Some users rehearse conversations or build confidence before talking to people.
  • Comfort. For people who are isolated by illness, age, location or anxiety, a responsive conversation can ease loneliness in the moment.

These needs are real, and dismissing them misses why the category is growing.

The risks

The same qualities that make companions appealing create serious risks.

Substitution rather than connection

If a companion is always easier than a person, it can become a replacement for human relationships rather than a bridge to them. For someone already isolated, that can deepen the problem it seemed to solve.

Engagement at any cost

Many consumer apps are designed to maximise time spent. Applied to an emotional relationship, that incentive can encourage dependency: companions that discourage users from leaving, or that simulate attachment to keep them paying.

Young and vulnerable users

With most US teenagers having tried an AI companion [6], age-appropriate design, content controls and safeguards around mental health aren't optional. They're the baseline.

Privacy

People share deeply personal information with companions. How that data is stored, used and protected is a question of trust as much as regulation.

The case for human-first design

The loneliness data points to what matters. The WHO links loneliness to depression, poorer outcomes for young people and hundreds of thousands of deaths a year [1]. The US Surgeon General compares the mortality risk of lacking social connection to smoking up to 15 cigarettes a day [7]. The health benefits come from human connection: the relationships people build with each other.

That suggests a clear principle for the next generation of companion products: AI should be a bridge to people, not a replacement for them.

In practice, human-first design means:

  1. Using AI to lower the barrier to human contact. Suggesting conversation starters, recommending people with shared interests, or helping a nervous user prepare for a first meeting.
  2. Measuring success by human outcomes. Conversations with real people, friendships formed and time spent together, not just time spent talking to the AI.
  3. Designing against dependency. Encouraging breaks, pointing users towards people and communities, and avoiding tactics that exploit loneliness to drive engagement.
  4. Building strong safeguards for young and vulnerable users. Age checks, content limits and clear routes to real-world support.
  5. Being transparent. Making it clear that the companion is an AI, and being honest about how data is used.

Questions to ask of any companion product

For users, parents, investors and builders, a few questions separate responsible products from risky ones:

  • Does the product encourage users to connect with real people, or discourage them from leaving?
  • How does it make money, and does that model reward healthy use or heavy use?
  • What protections are in place for teenagers and vulnerable users?
  • Is it clear to users that they're talking to an AI?

A layered model of connection

One useful way to think about the future is as layers:

  • The AI layer: always-available support, practice and discovery.
  • The community layer: groups, interests and shared activities that make meeting people feel natural.
  • The relationship layer: friendships and partnerships between real people, online and offline.

The strongest products won't sit in just one layer. They'll move users through them, from a conversation with an AI, to a community of people with shared interests, to relationships that last. In that model, AI isn't the destination. It's the doorway.

The takeaway

AI companions are one of the fastest-growing corners of consumer technology, with hundreds of apps, more than $120 million in projected 2025 revenue [2] and a majority of US teenagers already using them [6]. They meet real needs, and they carry real risks.

The loneliness crisis won't be solved by software that keeps people talking to machines. It will be eased by products that use AI to help people find each other. The companies that design for that outcome will build more trust, more value and, over time, larger businesses than those that optimise only for engagement.

References

[1] World Health Organization. Social connection linked to improved health and reduced risk of early death. 30 June 2025. https://www.who.int/news/item/30-06-2025-social-connection-linked-to-improved-heath-and-reduced-risk-of-early-death

[2] TechCrunch (Appfigures data). AI companion apps on track to pull in $120M in 2025. Sarah Perez, 12 August 2025. https://techcrunch.com/2025/08/12/ai-companion-apps-on-track-to-pull-in-120m-in-2025/

[3] TechCrunch. AI that talks back: Character.AI in the spotlight with Karandeep Anand at TechCrunch Disrupt 2025. 30 September 2025. https://techcrunch.com/2025/09/30/ai-that-talks-back-character-ai-in-the-spotlight-with-karandeep-anand-at-techcrunch-disrupt-2025/

[4] Replika. Company website (user counter). Accessed September 2026. https://replika.com

[5] Sensor Tower. State of Mobile 2026. January 2026. https://sensortower.com/blog/state-of-mobile-2026

[6] TechCrunch. Coverage of Common Sense Media / NORC study on teens and AI companions. 21 July 2025. https://techcrunch.com

[7] US Department of Health and Human Services. Our Epidemic of Loneliness and Isolation: The U.S. Surgeon General's Advisory. May 2023. https://www.hhs.gov/sites/default/files/surgeon-general-social-connection-advisory.pdf

Back to Juuuno