Character.AI Revenue: what is known, and the consumer-AI lesson in it

Character.AI has never published a revenue figure. The best-supported public number is a third-party estimate — about $30 million annualised as of July 2025, from research firm Sacra — against a user base the company itself put at more than 20 million monthly in October 2024. That gap is the story: not how big the revenue is, but why it is so small next to the usage.

Contents

• ·What is Character.AI's revenue?
• ·Which numbers are actually public, and how solid is each?
• ·Why does 20 million monthly users produce so little money?
• ·What did Google's $2.7 billion actually buy?
• ·What did the under-18 ban cost?
• ·What do our indexed companion and chat cases say?
• ·What should you do if you are building one?
• ·Frequently asked questions

What is Character.AI's revenue?

There is no official answer. The strongest public figure is Sacra's estimate of roughly $30 million annualised revenue in July 2025, alongside a company projection Sacra relays of $50 million by the end of 2025. Character.AI has confirmed neither. Treat both as research, not accounting.

Three grades run through this page. Company-stated: an executive or official post said it. Reported by press: a named outlet published it. Third-party estimate: an outside firm modelled it. Only the first two can be held against anyone — and Character.AI has made almost none about money.

Which numbers are actually public, and how solid is each?

Six figures are worth carrying: two company-stated, two press-reported, two third-party estimates. Nothing in the public record establishes recognised revenue, subscriber count, revenue per user or gross margin — so no honest page can hand you a 2026 revenue number.

Not public: paying subscribers, ad revenue split, inference cost per user, churn after the teen ban, or audited revenue. A page quoting a precise 2026 figure modelled it and rounded up.

Why does 20 million monthly users produce so little money?

Because the company said out loud that it was not trying. In October 2024 interim CEO Dominic Perella told TechCrunch the company was not concentrating on revenue generation and was focused on growth instead, with the $9.99 c.ai+ plan named as one of the only consistent sources of money.

The product shape sets the trap. Companion chat has the worst cost curve in consumer software: the more someone loves it, the more inference you buy for them, and the heaviest users are teenagers and students — the cohort least able to pay. Subscription is the only clean lever, and consumer chat converts in low single digits. Advertising, the fallback, interrupts the one thing the user came for.

So Character.AI built one of the internet's largest consumer-AI audiences and monetised it at the rate of a mid-sized SaaS company. That is the default for free consumer chat, and why our revenue-reality hub keeps finding small paid apps out-earning enormous free ones.

What did Google's $2.7 billion actually buy?

In August 2024 Google took a non-exclusive licence to Character.AI's technology and rehired co-founders Noam Shazeer and Daniel De Freitas. The $2.7 billion figure was reported by press, not published by either company, and the US Department of Justice later examined whether the structure functioned as a merger that avoided review.

Most pages misread it: the figure prices a research team and a licence, not a business. The most valuable asset walked back to Google; the revenue line stayed behind. The deal argues against "huge usage equals huge value": the buyer paid for the people and declined the product.

What did the under-18 ban cost?

On 29 October 2025 Character.AI announced it would remove open-ended chat for users under 18, stepping down to a zero-hour limit by 25 November 2025. CEO Karandeep Anand said he expected churn and acknowledged the change could affect the company's bottom line. No result has been published since.

This is the part builders skip and shouldn't. Character.AI lost its biggest cohort to a rule, not a competitor — and made the call itself, under wrongful-death lawsuits and regulatory pressure. If you sell emotional companionship, policy sets part of your market — and moves faster than your roadmap.

What do our indexed companion and chat cases say?

Our index contains no businesses built on Character.AI — it has no developer surface anyone in the library turned into revenue, which is itself a finding. What we do have is five companion, chat and personality apps with their own evidence — consistently smaller, cheaper to run, better monetised.

Index grades: ✅ third-party verified (confirmed outside the business), 🗣 founder-reported (the operator's own claim), 📎 creator-relayed (a channel repeating a figure it never audited), 🔮 unproven.

• ·[Chai](/projects/chai) — $30M/year, 21 people, $450M valuation. ✅ Verified. The closest comparator: Character.AI's estimated revenue, produced by 21 staff. We file it as a cautionary tale for its content, not its economics.
• ·[Bible Buddy](/projects/bible-buddy-ai-christian-therapist) — $10K–$20K MRR, then sold in a four-app package that went for $500K combined across 1M+ users. 🗣 Founder-reported. A Bible Q&A box that asks a question back — a tool that became a companion.
• ·[Toxic Traits](/projects/toxic-traits-chat-screenshot-personality-app) — same operator, same shape: upload a chat screenshot, get a personality read. $10K–$20K MRR before the package sale. 🗣 Founder-reported.
• ·[Rizz GPT](/projects/rizz-gpt-screenshot-coach-viral-app) *(members)* — ₩3 billion across three apps in 18 months, from a 23-year-old who could not code. 📎 Creator-relayed, so weigh it accordingly. The mechanic is the point: screenshot in, advice out, paid immediately.
• ·[K-Test](/projects/ktest-mbti-personality-quiz-platform) — under ₩3 billion cumulative from personality quizzes served as static pages, monetised entirely by display ads. 🗣 Founder-reported. The one case where advertising worked, because nobody was mid-conversation.

What should you do if you are building one?

Charge on day one, and pick a cohort that can pay and will not get you regulated out of your own market. The pattern that works in our index is a narrow, nameable job — a Christian counsellor, a texting coach, a personality read — priced immediately, not a character playground monetised later.

Four positions we defend:

1. 1.Free-then-monetise is the wrong default for chat. Every engaged minute is a bill. Character.AI grew for two years before pricing seriously; Bible Buddy and Rizz GPT charged first.
2. 2.Treat inference as cost of goods, not an R&D line. Chai does Character.AI-scale revenue with 21 people. Headcount and GPU spend are the whole margin conversation.
3. 3.A wedge beats a universe. "Any character you want" has no pricing logic. "Tell me what to text her back" has one.
4. 4.Never benchmark against a valuation. The $2.7 billion is a talent price; the $1 billion is a 2023 funding event. Neither says what the product earns.

More of this — every figure graded, every case linked — is in our revenue-reality hub, alongside vibe-coded apps and apps built with ChatGPT.

Frequently asked questions

How much money does Character.AI make?

No official figure exists. The strongest public estimate is roughly $30 million annualised as of July 2025, from research firm Sacra, which also relays a company projection of $50 million by end-2025. Character.AI has confirmed neither, and has never published recognised revenue, subscriber counts or margins.

Did Google buy Character.AI for $2.7 billion?

No. Google took a non-exclusive licence to Character.AI's technology in August 2024 and rehired co-founders Noam Shazeer and Daniel De Freitas. The $2.7 billion price was reported by press rather than published by either company, and the US Department of Justice has since examined whether the structure avoided merger review.

How many users does Character.AI have?

The last figure the company put on the record was more than 20 million monthly users in October 2024, described as having doubled year over year. Higher and lower numbers circulate from third-party trackers. After removing open-ended chat for under-18s in November 2025, the company has not published an updated count.

Why do data-broker sites list different Character.AI revenue numbers?

Because they are modelling, not reporting. With no company disclosure, estimate sites work backwards from app-store downloads, subscription price and assumed conversion rates, then publish the output as fact. Different assumptions give different answers, which is why $30 million and $50 million both appear for the same year.

Can you build a business on Character.AI?

Not the way you can build on a platform with an API and a revenue share. Our index contains no businesses built on Character.AI. Builders in this category ship their own apps on general-purpose models instead, and the ones with real revenue charge for a narrow, specific job rather than open-ended roleplay.