Andreessen Horowitz's latest Top 100 AI list includes, for the first time, an analysis of actual US consumer spending on AI. It found that while nearly half of US consumers use AI, only 4.5% pay for a subscription, with the top 1% of payers spending around $900 per month, mostly on professional development and automation tools.
SynthePulse Insight · AI deep reading
Version 1 · 1 source
The seventh edition of a16z's consumer AI ranking introduces credit card spending data for the first time, showing that nearly half of Americans use AI, but only 4.5% pay for ChatGPT, Gemini or Claude; among payers, the top 1% spend about $900 a month on average and contribute nearly one-fifth of total spending. The industry relies almost entirely on subscriptions, while ad monetization remains negligible.
The venture capital firm Andreessen Horowitz released the seventh edition of its ranking of the most-used consumer AI products. Unlike previous editions, which relied only on web traffic and mobile user numbers, a16z used YipitData for the first time in its latest ranking to track actual AI spending on US consumers' credit cards. This change lets outsiders observe the real monetization of AI products from the payment side for the first time.
But the limits must be made clear: this spending data comes from a sample panel and does not represent companies' total revenue. In addition, a16z's consumer spending figures cover only the US market; Gemini does not appear on the new spending list because its revenue cannot be separated from Google's other subscription businesses. The ranking is based on three types of data—web traffic, mobile users and consumer spending—which use different definitions and cannot be compared directly across categories.
According to a16z, nearly half of US consumers use AI, but only about a quarter use it daily. More critical is paid conversion: in August 2026, only 4.5% of US consumers had a paid individual subscription to ChatGPT, Gemini or Claude, roughly double the figure a year earlier. The paid ratio counts only individual subscriptions to these three products and excludes other AI products.
This set of figures comes from a16z's own tally and lacks independent third-party verification. But even by this measure, there is still a huge gap between AI adoption and payment: nearly half use AI, yet fewer than one in twenty are willing to pay for the leading chat assistants.
Within the narrow paying group, spending is extremely unevenly distributed. According to a16z, the top 1% of payers contribute nearly one-fifth of observed AI spending, more than the bottom 50% of payers combined. This 1% spends about $900 a month on average and that figure is still rising, while the typical paying user spends about $25 a month and has stayed flat for a long time.
Growth also comes almost entirely from this 1%: their spending rose 79% from early 2025. a16z believes consumer AI has already found its first genuine paying market among this group of professional users. Note that the $900 figure is a panel observation and may be affected by sample selection bias.
The product choices of high-spending users reveal who they are. According to a16z, they favor building and automation tools such as n8n, Manus and fal, as well as creative tools such as Higgsfield, Figma and HeyGen. In other words, they are buying software for building, designing and working, and professional use clearly drives the spending.
This product preference data comes from a16z's ranking analysis, which did not disclose the sample size, so it can only be treated as directional reference. But it is consistent with the concentration of spending: those who actually pay are not ordinary consumers but professional users who treat AI as a production tool.
At the top, the landscape looks stable. According to a16z, ChatGPT's web traffic is about twice Gemini's and six times Claude's, and the gap is even larger on mobile. But competition on the subscription side is fiercer: Claude surpassed Gemini in US paid subscribers in March 2026, and after Google adjusted its plans in June the two are neck and neck.
Monetization efficiency also differs. According to a16z, Anthropic's monetization efficiency is better than its rivals', and the share of Claude users choosing the top-tier plan starting at $100 a month is far higher than for comparable plans from OpenAI and Google. However, the lead between Claude and Gemini changed with Google's June plan adjustment, and the subsequent trend is unclear.
According to a16z, the vast majority of leading AI-native products rely on subscriptions or usage-based charges, with only a very few using advertising or transaction commissions. This stands in sharp contrast to the pre-AI-era internet: back then users themselves were the product, and advertising drove most of Meta's and Alphabet's revenue.
a16z believes high model costs prevent AI companies from doing what pre-AI-era internet companies did—giving up early revenue first to scale up users—so they have turned almost entirely to subscriptions. Alternative paths are beginning to emerge: OpenAI reported that as of August 2026 the annualized revenue run rate from ChatGPT ads reached $1 billion, but this figure is self-reported by the company and unaudited. The path of monetizing personal agents through affiliate commissions or transaction commissions is still early, and whether it can work is unknown.
Despite a steady stream of new players, the top of the market is becoming locked in. According to a16z, only seven companies made the top 50 on all three lists, including ChatGPT, Claude, Perplexity and Canva, and ChatGPT ranked first on all three.
a16z believes AI may worsen social skill and income inequality, but also acknowledges that AI could become easier to use and cheaper, making the problem disappear. This judgment itself carries uncertainty. Equally uncertain: the 4.5% paid subscription ratio and the nearly half usage rate both come from a16z's own tally and lack independent third-party verification; the top 1% of payers' average of $900 a month is a panel observation and may be affected by sample selection bias.
All data and judgments in this article come from the seventh edition of a16z's consumer AI ranking and THE DECODER's report. The spending data comes from the YipitData sample panel and does not represent companies' total revenue; the paid subscription ratio counts only individual subscriptions to the three products ChatGPT, Gemini and Claude; a16z's consumer spending figures cover only the US market. Key numbers such as the 4.5% paid subscription ratio, the nearly half usage rate and the top 1% of payers' average of $900 a month all come from a16z's own tally and lack independent third-party verification, and should be treated as source_claim rather than confirmed. OpenAI's ad revenue run rate is self-reported by the company and unaudited.
a16z tracked actual US consumer AI spending for the first time, revealing a split market: extremely broad use, an extremely narrow paying base, and revenue highly concentrated among a small number of professional users. The industry relies almost entirely on subscriptions rather than advertising, with high model costs as the core constraint. Whether this model can last depends on whether that 1% of high-spending users keeps growing, and whether alternative paths such as advertising and agent commissions can mature.
Primary report
Primary source