Menlo Ventures dropped its annual State of Consumer AI report in mid-September, and the headline number is hard to ignore: consumers worldwide are now spending an estimated $40 billion a year on AI, up from $12 billion in 2025. That's a tripling in twelve months, spread across roughly two billion users paying about $20 each on average. But if you run an adult AI chat directory like we do, the number that actually matters is buried further down. The top 14% of payers account for 60% of all consumer AI revenue. The casual crowd is window-shopping. A small, committed minority is footing the bill, and if you're reading this site, you're probably in that minority. So it's worth asking what that money buys, and what the fine print costs.
Key points
- Global consumer AI spending hit an estimated $40 billion in 2026, up from $12 billion the year before, per Menlo Ventures' survey of 5,067 U.S. adults.
- The top 14% of payers generate 60% of consumer AI revenue, meaning committed users, the core of the adult AI chat market, carry the industry.
- ChatGPT's ad-supported tier reportedly passed $1 billion in run-rate revenue roughly 200 days after ads launched, and ad economics create structural pressure to collect behavioral data.
- The report found users resist AI on finances, private records, and personal relationships, a trust boundary that sits at the center of the unfiltered chat debate.
- Menlo's authors hold positions in consumer AI companies and the global figures are the firm's own estimates, so treat the headline numbers as a directional signal, not gospel.
The report, authored by Shawn Carolan, Amy Wu Martin, and Sam Borja, surveyed 5,067 U.S. adults and benchmarked the results against last year's edition. Global AI users grew about 11%, from 1.8 billion to 2 billion, or roughly a quarter of the world's population. Average spend per user climbed from under $7 to around $20. One more milestone: ChatGPT introduced ads in February 2026, and per the report it crossed $1 billion in run-rate revenue roughly 200 days later, around late August or early September. Before you build a thesis on any of this, two flags. The survey covers U.S. adults, but the spending and user figures are global estimates from Menlo itself, not measured totals. And Menlo's authors invest in consumer AI companies, a conflict the firm discloses. We're a directory that verifies claims for a living, so we'll say it plainly: these are directional numbers from an interested party. They're probably roughly right. They're not audited.
The most useful finding in the report is that trying AI and relying on AI have split into two different behaviors. The pool of experimenters has stayed roughly flat. What's growing is how much the committed users give: more time, more tasks, more money, year over year. Growth is coming from existing users deepening usage, not from new signups. Now map that onto adult AI chat. A market where 14% of payers drive 60% of revenue is a market built on high-intent users, and that's precisely what the unfiltered companionship segment is. People don't stumble into an unrestricted AI companion the way they stumble into a free general-purpose chatbot. They seek it out, they stay, and they pay. Casual users churn. Committed users subscribe, and often subscribe across more than one platform. The catch, and there's always a catch: when a platform's economics depend on a small whale cohort, the product gets tuned for retention, not transparency. Conversation logging, moderation backends, retention hooks dressed up as features. None of that appears in the marketing copy. It appears in the privacy policy, if you read it, and in the matrix we publish after we test what a platform actually does versus what it claims.
ChatGPT passing $1 billion in run-rate ad revenue in about 200 days is the single most consequential data point in this report for our beat. It proves the ad-supported model works in consumer AI at scale. It also guarantees imitation. Expect more 'free' AI chat platforms, including ones in the adult and companionship space, to bolt on ad tiers over the next 18 months. Here's why that matters to you specifically. Ad economics run on behavioral profiling. An ad-supported general assistant needs to know what you shop for. An ad-supported intimate chatbot needs to know what you talk about when you think nobody's listening. The incentive structure is the exact opposite of the privacy posture an adult user should demand. Every platform that adds an ad tier becomes a new row in our filter-transparency matrix, and the question we ask is always the same: what does 'free' cost you in data? Sometimes the honest answer is fine. Often it isn't, and the platform's own policy language gives it away if you know where to look.
The report measures four dimensions of usage (time, money, tasks, autonomy) plus user confidence, and the trust finding is the interesting one. People accept AI drafts for low-stakes work but resist it on finances, private records, and personal relationships. Accuracy and privacy doubts decide which uses spread and which stall. On its face, that's awkward for our industry. The report says people don't trust AI with personal relationships, yet unfiltered companionship platforms keep growing. We think the reconciliation is straightforward: the trust line is about the vendor, not the task. Users don't distrust AI conversation as a category. They distrust specific companies with specific track records, and they distrust platforms whose marketing promises ('no filters, no limits, total privacy') don't match observed behavior. A user who believes a platform actually doesn't log conversations will go deep. A user who suspects otherwise stays shallow, churns, or never pays at all. That's the whole reason this site exists. Claims of 'unfiltered' are cheap. Verified behavior is expensive to fake, which is why our matrix cells only publish after we've tested the claim, and why we name the catch in plain terms every single time.
The global average of $20 per user is a smoothing that hides the shape of the market. If 86% of payers contribute 40% of revenue, the committed cohort is spending multiples of that average, and adult-oriented platforms almost certainly sit in that heavy-spend tier. Subscription stacking across multiple 'unfiltered' platforms is common in this space, and the per-dollar question is rarely asked: what does a $20 or $30 monthly subscription actually buy in filter behavior? Sometimes unrestricted output as advertised. Sometimes a moderated backend with the same content policies as mainstream platforms, plus a premium price tag for the branding. If you're spending in that committed cohort, the value test isn't whether a platform says it's uncensored. It's whether the output, the logging policy, and the moderation behavior hold up under inspection. That test is what we do, and the industry's concentration economics mean the platforms that fail it are failing their most valuable users first.
Menlo's report describes an AI market that's stopped being a novelty and become a line item, with a small committed minority carrying the economics and a trust boundary that decides where usage deepens. For the unfiltered chat segment, that's both a validation and a warning. The committed users are the market. The platforms that win them will be the ones whose privacy and moderation behavior survives verification, and the ones that don't will keep spending on marketing while quietly monetizing the gap between what they claim and what they do. We'll keep testing. As always, our matrix updates only after verification, and FlirtAlly, our own 18+ product, is in beta with the waitlist open at flirtally.com, disclosed in our footer and about page because disclosure is the minimum bar we hold everyone else to.