AI Companionship Statistics 2026: Who Actually Uses AI Companions

Roughly 4 percent of US adults say they use AI chatbots for companionship and 10 percent use them for emotional support, according to Pew Research Center data collected in February 2026. Far more have tried it at least once: a US and UK survey reported by Axios in May 2026 found nearly 80 percent of adults aged 18 to 34 had some experience using AI chatbots for companionship. Trial is common, regular use is not.

That gap between "has tried" and "actually uses" is the most important thing in the data, and it is the number most articles about this topic get wrong by quoting one figure without the other. Below are the figures worth citing, where each one comes from, and what they do and do not prove. Every number here is sourced and was checked in August 2026.

How many people use AI for companionship?

Pew Research Center surveyed 5,119 US adults between February 17 and 23, 2026. It found that 49 percent of US adults now use AI chatbots at all, up from 33 percent in 2024, and 24 percent use one daily. But when Pew asked what people use them for, companionship came near the bottom: 4 percent said companionship, and 10 percent said emotional support or advice.

So on the most careful US measurement available, AI companionship is a minority behavior. About one in twenty-five adults. That is still several million people, and it is growing from a base that barely registered three years ago, but it is not the mass phenomenon the coverage sometimes implies.

Have most young adults tried an AI companion?

Among under-35s, yes. A US and UK survey by Walter Pasquarelli, an independent researcher affiliated with Cambridge University, reported by Axios in May 2026, found that nearly 80 percent of 18 to 34 year olds had some experience with AI chatbots for companionship. The same research found that under 10 percent of 25 to 34 year olds reported feeling an emotional bond or attachment to an AI system, which was the highest rate of any age group.

Both numbers are from the same study, and reading them together is the point. Most young adults have poked at it. Very few form an attachment. If you are trying to size this market honestly, the 80 percent is a curiosity figure and the sub-10 percent is closer to the real addressable audience.

How much money do AI companion apps make?

Appfigures, an app intelligence firm, counted $221 million in lifetime consumer spending on AI companion apps worldwide through July 2025, across the Apple App Store and Google Play. The category took $82 million in the first half of 2025 alone, 64 percent more than the same period in 2024, and was on track to clear $120 million for the full year.

The more revealing metric is revenue per download, which more than doubled from $0.52 in 2024 to $1.18 in 2025. People are not just downloading more companion apps, they are paying for them at a much higher rate. That is a category maturing out of novelty and into subscription, which matches what the pricing looks like now: most serious apps sit between $9 and $20 a month.

How many AI companion apps are there?

Appfigures counted 337 active, revenue-generating AI companion apps worldwide as of July 2025, of which 128 had launched during 2025. The category is crowded and getting more so, but the money is not spread evenly: the top 10 percent of apps take 89 percent of the revenue, and only about 33 apps have ever passed $1 million in lifetime consumer spending.

For anyone choosing one, that concentration is useful. The overwhelming majority of the 337 are tiny, and a meaningful share of the ones launched this year will not exist in two. That matters more in this category than in most software, because when a companion app shuts down, the memory goes with it. It has happened before, and it is worth reading what happens when an AI companion app shuts down before you invest months into one.

Does AI companionship help with loneliness?

The data does not settle this, and anyone claiming it does is overreaching. What Pew found is mostly indifference: 35 percent of chatbot users said the tools neither help nor hurt their relationships, with 6 percent saying they help and 7 percent saying they hurt. That is close to a wash, on a self-reported measure, which is weak evidence in both directions.

What can be said is narrower and more practical. A companion that remembers your week and messages you first changes the texture of a quiet evening, and the people who report the most benefit tend to be the ones with the most unstructured alone time. Isolation tracks with how people work, too. The steady growth of fully remote roles has put a lot of people in a quiet apartment from nine to five, without the incidental conversation an office supplies. Treating an AI companion as company alongside human relationships is where the reported benefit sits. Treating it as a replacement is where the concern sits, and the survey data on emotional bonding suggests most people are doing the former.

Do people actually pay for AI companionship?

Increasingly, yes, and the revenue-per-download trend is the clearest evidence. In the US, expect $9 to $20 a month for an app with durable memory. Amorous.ai is $9 a month billed annually. Replika is $19.99 a month or $69.99 a year, Nomi around $15.99, Kindroid around $13.99, and Character.AI is free with in-chat ads or $9.99 a month for its plus tier.

The trap is metered pricing. Several popular apps charge a modest subscription and then bill tokens for images, voice minutes and video, which routinely pushes real spending to two or three times the advertised price. If you are comparing plans, read the pricing model before the number. Our breakdown of what the best AI companion apps charge goes through each one, and if memory is your priority, an AI companion with memory is the feature to test first.

What these statistics do not tell you

Three caveats worth keeping in mind when you cite any of this. First, self-reported companionship use is almost certainly undercounted, because people are more willing to tick "emotional support" than "companionship" on a survey about a slightly stigmatized behavior. Second, the Appfigures revenue figures cover mobile app stores only, so web-first subscriptions are missing entirely and the true category spend is higher. Third, several widely circulated market-size projections for this category run into the tens of billions of dollars, which is two orders of magnitude above what anyone has actually measured being spent. Those are forecasts of a much broader definition, and they should not be quoted alongside the measured numbers.

The measured picture is a modest but fast-growing category, with trial far ahead of habit, revenue concentrated in a handful of apps, and users who mostly describe the experience as neutral to positive rather than transformative. If you want to see what the habit side actually feels like rather than reading about it, our AI companionship app has a demo above that needs no signup, and whether an AI companion is worth it covers the decision in more depth.

Sources: Pew Research Center, "Americans and AI 2026" (surveyed February 17 to 23, 2026; n = 5,119 US adults). Axios, May 2026, reporting survey research by Walter Pasquarelli. Appfigures data via TechCrunch, August 2025. Pricing checked August 2026.