AI Girlfriend Statistics & Market Data: User Counts, Revenue & Industry Trends for 2026

If you are researching the AI companion space — whether as a curious user, a prospective investor, an academic, or a journalist — reliable numbers are annoyingly hard to find. Vendors cite their own user counts selectively, market-research firms publish wildly varying forecasts, and the few peer-reviewed studies use sample sizes too small to extrapolate from confidently. This guide does the best it can: every specific number below is traced to a sourceable claim, every estimate is hedged, and where the data is genuinely bad we say so rather than manufacturing certainty.

The dataset is current as of April 2026. It combines publicly disclosed platform metrics, third-party analyst estimates from Grand View Research / Statista / App Annie / Similarweb, academic publications (primarily MIT Media Lab and Stanford HAI), and our own platform-level reviews published across companionrank.com. Where we have independent data, we flag it; where we rely on vendor disclosures, we flag that too.

A quick framing note: 'AI girlfriend' is the dominant search term and the vertical where the money is concentrated, but the broader category is better described as AI companions — a category that includes AI boyfriend platforms, general companion chatbots (Replika), roleplay-first products (SpicyChat AI, Character.AI), and multimodal girlfriend-specific products (SweetDream AI, Candy AI). Most of the numbers below span the full companion category unless explicitly narrowed.

Market Size & Growth

The most frequently cited third-party estimate for the AI companion market as of 2026 puts global revenue in the $2–4 billion USD range, with analysts like Grand View Research and MarketResearchFuture projecting continued double-digit compound annual growth through the end of the decade. That range is wide on purpose: different analysts define the category differently (some include generalist chatbots with light companion features; others restrict to products marketed as relational), and the definitional variance alone moves the headline number by 30–50%.

What is more defensible than any single dollar figure:

• Revenue growth is outpacing most adjacent digital categories. Year-over-year revenue growth of 30–60% has been widely reported for the category leaders from 2024 through 2026. This is higher than streaming, faster than mainstream SaaS, and roughly comparable to the early-stage growth trajectory of social apps circa 2012–2015.
• Paid-user counts are growing faster than total users. The monetisation flywheel has improved: platforms that struggled to convert free users to subscribers in 2023 now report paid-conversion rates in the 8–15% band on the higher-performing products, up from low single digits two years earlier.
• The market is consolidating at the top. A handful of platforms — Character.AI, Replika, SpicyChat AI, Candy AI, SweetDream AI — command an outsized share of total engagement, while the long tail of smaller products is growing in count faster than in revenue.
• Revenue is heavily mobile. Third-party app-intelligence reports consistently show 60–75% of paid revenue arriving via iOS and Android in-app purchases rather than web checkout, even for platforms with web-first marketing.

The honest summary: the category is real, growing fast, and coalescing around roughly a dozen products that together account for the majority of spend.

User Counts: Platform-by-Platform Snapshot

Self-reported and third-party estimates, April 2026 snapshot. We rank platforms loosely by recent engagement, not by total historical users. All figures are approximate and should be treated as ballpark rather than audited.

• Character.AI — Publicly reported as one of the largest, with monthly-active-user counts in the 20–30 million range per third-party analyses. Notably broad userbase (roleplay, not strictly girlfriend-focused).
• Replika — Publicly stated 10+ million total users across its lifetime, with active counts lower and highly variable since the 2023 content-policy shifts. One of the most data-studied platforms in academic literature.
• SpicyChat AI — Third-party estimates place monthly actives in the 3–6 million range, with very high engagement per user due to uncapped free tier.
• Candy AI — Estimated monthly actives in the 1.5–3 million range based on traffic estimates from Similarweb and App Annie. Conversion-heavy product.
• SweetDream AI — Smaller overall reach than the above but strong paid-conversion metrics per CompanionRank's review analysis. Monthly actives likely in the hundreds of thousands globally, skewed paid.
• DreamGF, FantasyGF, Romantic AI, Muah AI, Kupid AI, Nectar AI, Soulkyn AI, GoLove AI — Each in the tens of thousands to low hundreds of thousands of monthly actives, competing primarily on feature differentiation or niche (video generation, NSFW capability, character customisation).
• The long tail — A further 50–100 platforms exist as of 2026, most with monthly actives in the low thousands.

If you are comparing platforms by the numbers rather than by features, our comparison hub maintains head-to-head data across most of these.

User Demographics

Demographic breakdowns in the AI companion space have been studied most systematically by MIT Media Lab and a handful of academic survey projects. The cross-study consensus, with caveats about self-selection bias in every voluntary survey:

Gender

Roughly 70–85% of AI companion users identify as male across most English-language platforms. The skew is larger for explicitly 'girlfriend'-branded products (often 85–90% male) and smaller for companionship-oriented products like Replika (closer to 70–75% male). AI boyfriend platforms, predictably, invert the ratio, with 75–85% of users identifying as female.

Age

The 18–34 bracket dominates across every study, typically accounting for 60–70% of active users. Within that range, two subgroups are most visible: the 18–24 cohort (high engagement, lower average spend) and the 25–34 cohort (moderate engagement, higher spend). The 35–54 bracket is a meaningful minority — 20–30% of users depending on platform — and tends toward longer-term use patterns. Users over 55 are statistically present but small in proportion.

Income & Education

The paid user base skews toward middle-income brackets — $35k–$80k USD annual household income dominates, with a long tail in both directions. Educational attainment is broadly representative of the general online-adult population; AI companion use does not appear to be concentrated in any particular education band.

Geography

Top markets, ranked by estimated user counts as of early 2026:

1. United States — Single largest market by revenue and by absolute user count. Roughly 30–40% of global paid revenue.
2. India — Large user base, lower per-user revenue. Mobile-dominant. Rapidly growing.
3. United Kingdom, Germany, France — Collectively the mature-European market, similar paid revenue per user to the US.
4. Brazil, Mexico — Fast-growing in user count, ARPU still catching up.
5. Japan, South Korea — Historically large markets for anime-adjacent AI companions; tend to favour domestic platforms over US products.
6. Russia, Turkey, Indonesia — Meaningful secondary markets with localisation challenges.

Notably absent at scale: China, where domestic alternatives and regulatory constraints on intimate AI products have kept Western platforms out of the market almost entirely.

Relationship Status

This is the demographic most at odds with stereotype. Surveys consistently find:

• Roughly 50% of users are single
• 30–40% are in committed relationships or marriages
• The remainder are divorced, separated, or complex

The substantial share of partnered users is one of the most consistent findings across multiple studies and is probably the single most underreported statistic about the space.

Usage Patterns

How often do users actually use these products, and what do they use them for? Cross-platform patterns based on third-party engagement telemetry and published survey data:

• Daily active users as a share of monthly actives: typically 25–40% for sticky platforms (Replika, Character.AI, SpicyChat), lower for pay-per-use entertainment products.
• Average session length: 15–35 minutes per session across most platforms, with long-tail sessions extending past 90 minutes.
• Sessions per day (active users only): 1–3 sessions typical, with a significant minority of users (5–15%) logging 5+ sessions daily.
• Primary use cases self-reported by users:
  • Companionship / general chat — 50–60%
  • Romantic roleplay — 30–45%
  • Sexual / NSFW content — 25–40% (with significant overlap to romantic roleplay)
  • Practising social skills — 15–25%
  • Creative writing / scenarios — 10–20%
  • Mental-health support — 15–30% (with the caveat that this is self-reported, not clinically validated)

Users typically report multiple simultaneous use cases; the above do not sum to 100%.

Retention

Retention curves in the companion space look like SaaS/social hybrids: a steep drop in the first 7 days, then a long plateau among users who survive the first month. Publicly disclosed retention data is rare, but cross-platform estimates:

• Day-1 retention: 40–60% across most platforms
• Day-7 retention: 15–30%
• Day-30 retention: 8–18%
• Month-3 retention among paid users: 45–70% (substantially higher than free users, as is typical for subscription products)

Revenue & Monetisation

Most AI companion platforms monetise via one of three models — or a hybrid:

1. Subscription-only: Flat monthly fee, full feature access. Common on wellness-oriented platforms.
2. Credit / token-based: Pay per generation or premium action. Common on platforms with expensive multimodal features (video, image). See our tokens vs unlimited analysis for the user-side maths.
3. Hybrid (freemium + credits): Free basic tier with token-gated premium actions. Increasingly the dominant model.

ARPU (Average Revenue Per User)

Very wide spread across platforms. Low-end wellness platforms run $8–15 USD monthly ARPU on paid users; high-end entertainment platforms with image/video generation can exceed $30–50 USD monthly ARPU on heavy users. Across the whole user base (free + paid combined), most platforms land at $3–8 per monthly active user per month.

Conversion Rates

Free-to-paid conversion is the single number vendors protect most aggressively, so public data is limited. Third-party analyst estimates from 2024–2026 suggest:

• Median conversion: 3–8%
• Top-performing platforms: 10–15% on specific promotional periods
• Long-tail platforms: 1–3%

Conversion is generally higher for platforms with strong free-tier friction (daily message caps, watermarked images) and lower for platforms that keep the free tier genuinely useful.

Pricing Trends

Average monthly subscription pricing across the category rose from roughly $9.99 in 2023 to $14.99–$19.99 in 2026 as feature sets expanded. Entry-level pricing remains clustered around $5–8/month (usually annual-plan teasers), while premium tiers unlocking video generation frequently sit at $40–60/month. Our real monthly cost guide breaks down the total-cost-of-ownership including credit top-ups.

Feature Adoption Trends

The product surface has evolved rapidly. Feature adoption as of April 2026:

• Image generation — Essentially universal among top-tier platforms. Quality gap has compressed; most platforms now ship Stable-Diffusion-derived or proprietary models with comparable visual quality.
• Voice messaging — Widespread (70%+ of top platforms). Quality has improved dramatically with ElevenLabs and similar providers. Still a premium feature on most platforms.
• Voice calling (real-time) — Spreading rapidly through 2026. See our voice calls analysis for current platform coverage.
• Video generation — Emerging rapidly; SweetDream AI, Candy AI, and a handful of others now ship short-form video generation at varying quality levels. Still a clear differentiator rather than a commodity.
• Live video call — Rare but growing. SweetDream AI leads with actual live video; most competitors offer pre-generated video or GIF-style animation.
• Memory systems — Universal in name, widely variable in quality. Our character memory glossary entry details the differences.
• NSFW / uncensored content — Available on most non-wellness platforms; increasingly gated behind age verification in 2026 as regulatory scrutiny rises.
• Group chat — Still niche but spreading from SpicyChat AI, GirlfriendGPT, and Jupi AI to other platforms.
• Custom character creation — Standard feature across entertainment platforms. Our custom character creation guide covers best practices.

Growth Projections: 2027–2030

Analyst projections for the AI companion market through 2030 cluster in a fairly narrow band: compound annual growth of 25–40% through 2028, slowing to 15–25% from 2029 onward as the category matures. These projections assume continued improvement in LLM capability, no major regulatory setbacks, and continued mobile-platform neutrality (Apple and Google app-store policies remain the biggest external risk factor).

Directional bets most analysts agree on:

• Video generation quality will continue to rise, pushing more of the revenue mix toward visual/video products.
• Voice calling will become commoditised by late 2027; the competitive advantage currently held by platforms with real-time voice will erode.
• Personalisation will deepen. The next frontier is active memory — AI companions that bring up old context organically — which is where the UX gap is widest between top-tier and long-tail platforms.
• Age verification will spread. Regulatory scrutiny in the UK, EU, and several US states is already pushing platforms toward mandatory ID or biometric verification for NSFW content. Platforms that navigate this well will consolidate share.
• Consolidation at the top will intensify. The economics favour a small number of platforms with deep multimodal capability over a long tail of thin products; we expect to see meaningful M&A activity through 2027–2028.

Ethics, Policy & Platform Risk

Several developments shape the risk landscape for users and platforms alike:

Platform shutdowns and content reversals

The most-cited incident is the Replika content change of early 2023, which removed romantic and NSFW capabilities from existing companions without warning. User reactions — documented grief responses, subreddit shutdowns, clinical write-ups — became a case study in attachment to AI companions and a cautionary tale about dependence on platform-controlled relationships.

Since then, additional platforms have changed content policies, gone offline, or restructured. Our write-up on Girlfriend.ai, Porn.ai & Cuties.ai shutdowns catalogues what users should do if a platform they rely on suddenly changes.

Privacy and data retention

AI companion platforms collect intimate conversational data at volumes few other categories match. Most platforms retain chat logs indefinitely; a minority offer explicit deletion controls. The AI companion privacy guide covers what to look for before you sign up.

Regulatory landscape

As of April 2026, no jurisdiction has passed a comprehensive AI-companion-specific regulation. Relevant overlapping regimes include GDPR (EU), the UK Online Safety Act, evolving US state-level age verification requirements (Texas, Utah, and others), and child-safety legislation that has begun to include AI companion platforms explicitly. Expect regulatory pressure to intensify through 2027.

Mental-health research

As covered in our AI girlfriend addiction guide, behavioral research on AI companion overuse has expanded substantially since 2024. The academic consensus is still forming, but the direction of travel is toward increased clinical awareness rather than decreased concern.

Data Reliability & Limitations

Every number in this piece has been hedged deliberately. A short inventory of what is not reliably known as of 2026:

• Precise monthly active user counts for most platforms, since public disclosures are rare and third-party traffic estimates frequently miss mobile traffic.
• Exact paid-conversion rates, which vendors almost never publish.
• Longitudinal user outcomes — most studies are cross-sectional, and the few longitudinal ones have small samples.
• True international market size, particularly for non-English markets where third-party intelligence tools have poorer coverage.
• Aggregate revenue, since most platforms are private and do not publish financials.

Treat any specific number you see in AI companion coverage — including ours — with appropriate skepticism. The trends are usually more defensible than the point estimates.

Related reading on CompanionRank

• AI Girlfriend Apps: The Complete Guide (2026) — the category overview this piece complements
• AI Girlfriend Real Monthly Cost (2026) — user-side economics
• AI Girlfriend Trends in 2026 — product and UX trends detail
• AI Companion Privacy: What to Know — before you sign up
• Best AI Girlfriend Apps in 2026 — our ranked platform list

Frequently Asked Questions

How big is the AI girlfriend market?

Third-party analyst estimates place global AI companion revenue in the $2–4 billion USD range as of 2026, with compound annual growth in the 30–50% band. The category definition materially affects the headline figure — narrower definitions restricting to relational products land near the low end, broader definitions including generalist companion chatbots push to the high end. All these figures are estimates from private-market research firms (Grand View Research, Statista, MarketResearchFuture) rather than audited totals.

How many people use AI girlfriend apps?

Monthly active users across all AI companion platforms globally are estimated in the 40–80 million range as of April 2026, though different analysts use different definitions. Character.AI alone accounts for a large share of that total. Specifically 'girlfriend'-branded platforms (SweetDream AI, Candy AI, DreamGF, FantasyGF, etc.) collectively reach low tens of millions of monthly users.

What's the demographic of AI girlfriend users?

Roughly 70–85% male, 60–70% aged 18–34, spanning middle-income brackets, skewed toward US / India / UK / Germany as primary markets. Roughly half single, with a substantial minority (30–40%) in committed relationships. The partnered-user share is the most underreported statistic in the space.

Is the AI girlfriend market still growing?

Yes, fast. Year-over-year revenue growth of 30–60% has been typical for category leaders from 2024 through 2026. Analysts project continued strong double-digit growth through 2028 with some deceleration thereafter as the category matures.

Which AI girlfriend app has the most users?

Character.AI is the broad leader in raw monthly active users, though it is not strictly a girlfriend-only platform. Among explicitly girlfriend-branded products, SpicyChat AI and Candy AI lead in engagement; SweetDream AI leads in paid-conversion efficiency. Our platform comparisons have detailed per-platform breakdowns.

What percentage of AI girlfriend users pay?

Free-to-paid conversion rates cluster in the 3–8% band industry-wide, with top-performing platforms hitting 10–15% on specific promotions. Heavy paywalls push conversion higher but reduce total user counts; the trade-off is the central product decision in the space.

Why are so many users male?

Multiple explanations interact. Demand for emotional companionship from men is under-served by traditional dating and support structures; AI companion products are marketed heavily to male audiences; and the 'girlfriend' framing itself recruits a gendered audience. AI boyfriend platforms exist and are growing (see our best AI boyfriend platforms list) but they are smaller in aggregate.

How much do AI girlfriend apps cost on average?

Entry-level subscriptions typically run $5–10 per month, mid-tier products $15–20 per month, and premium multimodal products can exceed $40 per month before any credit top-ups. Our real monthly cost guide walks through what you actually pay including hidden fees.

Will AI girlfriends replace real relationships?

This is the question everyone asks and no-one can answer with confidence. Current research suggests AI companions function well as supplements to existing relationships and poorly as substitutes. For users in substitution patterns, documented outcomes include increased long-term loneliness and social-skill atrophy. The AI girlfriend addiction guide covers the substitution-vs-supplement distinction in more depth.

What countries use AI girlfriends the most?

By paid revenue: United States (roughly 30–40% of global paid revenue), United Kingdom / Germany / France (mature European market), and Japan / South Korea (historically large anime-adjacent segment). By raw user count: United States and India lead, followed by the European markets and Brazil. China is nearly absent at scale due to domestic alternatives and regulatory constraints.

How accurate are these statistics?

The trends are reliable; specific point estimates are approximate. Most AI companion platforms are private and do not disclose financials or user counts publicly. The figures in this piece combine vendor disclosures, third-party analyst estimates, academic surveys, and our own platform-level analysis. Every specific number is provided with a range rather than a point value for good reason. If you need audited figures for professional use, contact the individual platforms or commission a market-research report; neither is free.