a16z's new AI spending ranking finds spending concentrated among power users

Why it matters

The new spending view separates audience size from monetization: consumer AI's most valuable early customers are concentrated users buying tools for work and creation, leaving startups to solve how to serve everyone else.

Olivia Moore, a former early-stage investor at CRV who now focuses on AI at Andreessen Horowitz, has spent three years tracking which consumer AI products draw users. In the seventh edition of a16z's ranking, published October 5th, Moore adds a new measure alongside traffic: observed spending on U.S. consumer cards. The result is a sharper view of a market where attention is broad, but paid use is concentrated.

The report ranks 50 web products by monthly visits and 50 mobile apps by monthly active users, using Similarweb and Sensor Tower data. Its new spending view, supplied by YipitData, captures a different slice of the market, including desktop products and agents used inside messaging apps. The measures are not interchangeable: card spending in a panel is not a company's total revenue, and traffic rankings do not show how much a user pays.

Twenty-nine of the 50 highest-spending vendors do not appear on either the web or mobile lists, according to a16z. Only seven companies appear across all three rankings: ChatGPT, Claude, Suno, Perplexity, Photoroom, Canva and Notion. A traffic leaderboard alone misses products that make money from a smaller group of intensive users.

The market has users; the paid base is thinner

The report says nearly half of U.S. adults now use AI, while about a quarter use it daily. Pew Research Center's June survey similarly found that 49% of U.S. adults had used AI chatbots, with roughly one-in-four using them daily. The paying base remains narrower: in YipitData's U.S. e-receipt panel, 4.5% of eligible consumers had an active personal subscription to ChatGPT, Gemini or Claude in August, up from 2.1% a year earlier.

Among AI payers, spending is unusually concentrated. The top 10% of spenders account for roughly half of the observed spend. The top 1% alone account for 19.5%, spending an average of $903 a month on AI products through consumer cards in August. The median payer spent $25. Those top buyers lean toward coding, automation, productivity and creative tools, including n8n, fal, Manus, Higgsfield, Figma and HeyGen, a16z says.

The figures describe a panel, not every consumer or every dollar earned by the named companies. Raw popularity alone can miss how much users rely on specific products. ChatGPT remains the leader by web traffic and mobile activity in a16z's rankings, but products built for specific jobs can draw substantially more spending from users who rely on them heavily.

Moore's own path gives the report a useful lens on that divide. Before joining a16z, she invested in consumer and small-business technology at CRV, after working on alternative investments at Goldman Sachs. She studied economics at Stanford, earned an MBA at Stanford Graduate School of Business and co-founded Cardinal Ventures, an on-campus startup incubator. At a16z, her investments have included consumer and AI companies such as Krea and HappyRobot. The report approaches consumer AI as both a usage market and a business-building problem: who returns, who pays, and what the product helps them accomplish.

A crowded leaderboard, a quiet business-model problem

ChatGPT still leads Gemini by roughly two to one in August web visits and six to one over Claude, according to a16z. Its mobile lead is wider: 2.5 times Gemini's monthly active users and 14 times Claude's. The report's panel also puts ChatGPT's U.S. consumer paid subscribers at about three times either rival's. Claude, meanwhile, has moved from outside the first web ranking in 2023 to passing DeepSeek and Perplexity in traffic, and a16z says it passed Gemini in U.S. consumer subscribers earlier this year.

Beyond the three model companies, a16z says 84% of the 44 AI-native products in its web ranking offer subscriptions and 64% charge by usage or sell extra credits; only 14% offer advertising and 2% collect transaction or platform fees. Products can use more than one model. That mix asks consumers to pay directly for software at a time when many people use AI only intermittently.

For startups, the opportunity in these numbers is specific. Specialized creative models, multi-model interfaces and software designed around a particular professional's workflow can earn meaningful spend without becoming general-purpose chatbots. But a business built around subscriptions may struggle to serve casual users who will not pay monthly. a16z sees personal agents and transaction-based fees as possible alternatives, though the report describes that path as an emerging bet rather than an established model.

The findings echo a16z's broader recent market argument. On October 1st, RuntimeWire reported on the firm's view that AI investment is shifting toward chips, power and robotics as public software faces a tougher growth test. Moore's consumer data puts the same question at the product level: AI usage is real, but turning occasional use into durable revenue remains unfinished work.

The seventh edition also has the fewest first-time entries of any of a16z's seven lists: 11. That suggests the visible set of consumer AI products is settling, even as the spending rankings reveal businesses that traffic alone obscures. The report excludes products primarily designed for NSFW use cases from future editions, treating them as a separate category; a16z says the change should not be read as evidence that demand has fallen.