Fewer than 5% of Americans pay for AI products—top spenders hit $903 monthly
Only 4.5% of Americans pay for AI products, but the top 1% spend an average of $903 per month—the real consumer AI business is propped up by a tiny fraction of power users.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
Payment data reveals moneymakers traffic metrics completely miss
This season's traffic leaderboard added only 11 new products—the fewest in seven seasons—suggesting the consumer AI market is consolidating. But when the report introduced payment data for the first time, 29 of the top 50 revenue-ranked products never appeared on any traffic leaderboard. The gap between traffic metrics and actual spending has become vast: many products have modest traffic but loyal power users willing to spend serious money. Traditional measures like website visits no longer capture real business scale.
— Olivia MooreInstinct captures early monetization faster than Muse despite less hype
Instinct hit 100K users and grew at 10% daily, with 40% binding credit cards within three weeks and average first-month spending exceeding $1000. Muse attracted bigger headlines: 500K downloads and 250K active users by day 12, but measured over 22 days across US and Canada, Muse had roughly 5M downloads while Threads hit 16M. The comparison reveals that buzz in tech circles doesn't always translate to real growth or monetization speed.
— Olivia MooreThe top 1% of paying users fund the entire category
About half of Americans report using AI, but only roughly 4.5% actually pay subscriptions for AI products—up from half that share a year ago. Within this paying cohort, the distribution is brutally steep: the top 10% generate over half of all revenue, the top 1% account for 20%, and the bottom half contribute just 16%. Median monthly spending among paying users sits at $25. But the top 1% average $903/month, predominantly on productivity and creative tools like N8N, Granola, and Manus.
— Olivia MooreSubscriptions are a placeholder until inference costs drop
Among consumer AI products on the leaderboard, 85% charge via subscription, 62% use pay-per-token or credit models, and only 13% rely on ads or ‘you are the product’-style monetization. This inverts consumer internet history: the largest consumer tech companies built revenue primarily on ads or transaction fees, not subscriptions. Subscriptions dominate now because inference still costs too much. Once unit economics improve further, the model will likely flip again—ads and transaction fees should drive the largest revenues, as they have in nearly every major consumer tech company.
— Olivia MooreOpenAI reached $1B in annual ad revenue faster than any platform
As of August, OpenAI's advertising business was running at a $1B annualized rate (likely higher now) while still being deliberate about ad expansion and working with only a handful of partners. No platform has reached this scale in so little time. This is powered by real user density: ChatGPT reports 1.2 billion weekly active users. More importantly, ChatGPT understands its users far better than any historical ad network—with login-based identity and a wallet product coming. This could let it achieve pricing power and conversion rates that beat what Meta or Google ever achieved.
Claude's paid subscriptions now exceed Gemini's despite distribution gap
ChatGPT still leads decisively in web traffic (6 times Claude, twice Gemini) and paid subscriptions (roughly 3x the others). But when looking at US payment data for the first time, Claude's paid subscriber base surprisingly exceeds Gemini's, even though Gemini has more installs and Google's native distribution advantage. Anthropic pursued a different strategy: no advertising, aggressive subscription focus, and a striking 7.5% of its subscribers chose the $100+ Max tier—compared to roughly 1% for both ChatGPT and Gemini. This suggests Anthropic's audience skews toward power users willing to pay premium prices.
— Olivia MooreStartups win where incumbents won't cannibalize their own interfaces
Products like Granola, Whisperflow, and Superhuman scaled from consumer tools to enterprise business through product-led growth—they became so good at helping individuals that companies felt compelled to buy enterprise versions, then pushed upstream to add privacy, compliance, and team features. The pattern repeats: established companies rarely reinvent their flagship interfaces. Google didn't rebuild Docs or Gmail for the AI era; even OpenAI and Anthropic have rarely spun out entirely new products beyond their core offerings. The organizational friction of large companies—governance, competing stakeholders, legacy revenue protection—makes it hard to cannibalize existing products, even when incumbents see the threat.
— Josh ElmanAI hasn't touched the products people spend the most time on
Nearly every breakout consumer AI product saves time: email composition, search, task management. But the largest time sinks in consumer internet—Netflix, TikTok, and their peers—remain largely AI-free. Two-sided markets like dating and hiring need genuine matching; no standout AI-native product exists in either yet. Social media currently uses AI only to generate content that gets posted to existing platforms, not to reimagine the experience itself. Shopping and e-commerce remain unsettled: will they become standalone AI products or be absorbed by personal assistants like Instinct and Muse? Entertainment and game content generation still lacks compelling output quality. These untouched categories represent the frontier for consumer AI's next wave.
In their own words · checked verbatim
The top 1% user is spending $903 per month personally on their personal credit cards on AI. And the median is spending $25.
Olivia Moore15:35
most people aren't looking to save time. They're looking for ways to spend their time.
Olivia Moore18:41
substantially all of the consumer AI revenue thus far has come from direct subscriptions.
Olivia Moore20:41
They're at a billion dollars now in annual run rate on advertising.
Claude has actually passed Gemini in terms of number of paid subscribers, which is kind of crazy given Gemini has a much bigger install base and user base overall.
Olivia Moore30:49
incumbents being unwilling or unable to cannibalize their existing interfaces.
Josh Elman39:06
the difference between an email that is 99.9% sounds like me and even an email that's 85% sounds like me is the difference between spending like 10 seconds to fix it and like 10 plus minutes.
Josh Elman44:07
Figures
| Share of Americans using AI | ~50% | 14:32 |
| Americans paying for AI subscriptions | ~4.5% | 14:32 |
| Top 1% of paying users' average monthly spend | $903 | 15:35 |
| Median monthly spend among paying users | $25 | 15:35 |
| Top 10% of paying users' share of revenue | >50% | 15:35 |
| ChatGPT weekly active users | 1.2 billion | 24:43 |
| Instinct users binding credit cards in first 3 weeks | 40% | 6:07 |
| Anthropic subscribers on Max tier ($100+) | 7.5% | 31:54 |
Glossary
- Harness
- A dismissive term for a product that merely wraps a language model in a user interface without adding independent value.
- PLG
- Product-led growth: acquiring users through a product's inherent quality rather than through sales efforts.
- Run rate
- Projecting current revenue over a full year.
- Power law
- A distribution where a tiny fraction of users or products generate the majority of revenue or traffic.
How to listen
Founders and investors trying to figure out which business model—subscriptions, ads, or transaction fees—works for consumer AI; and anyone tracking how ChatGPT, Claude, and Gemini compete.
The middle section where Josh describes buying keychains with ChatGPT is personal anecdote and skippable.