How much do people spend on AI every month

From everyday questions to professional tasks

How much users spend on AI can depend on what they use it for. Some need answers to everyday questions, others use several tools to write code, produce videos, generate voiceovers and automate workflows.

The second group stands out in the seventh edition of venture capital firm Andreessen Horowitz’s (a16z) consumer AI report. For the first time, its authors supplemented popularity rankings with YipitData figures on U.S. consumer spending.

In August 2026, the top 1% of spenders paid an average of $903 a month for AI. Median spending among paying users was $25.

This top 1% accounted for almost a fifth of all observed consumer spending on AI, outspending the bottom half of paying users combined.

Why one chatbot is not enough

General-purpose assistants can handle an increasing range of tasks, but specialized music, video and voice tools have retained their audiences.

The report’s authors suggest that a particular sound, visual style or training dataset can matter more for a creative task than access to the strongest general-purpose model.

Products that offer access to models from different developers are also finding an audience, allowing users to choose a model to suit each task.

However, paying for multiple AI services remains uncommon. According to the report, only 13% of users who pay for an AI product also pay for at least one other.

More visitors do not necessarily mean more money

AI tools are often judged by how many people use them. Spending rankings, however, tell a different story.

Music generation service Suno, for example, ranks 19th in web traffic but 7th in observed consumer spending. Voice AI platform ElevenLabs ranks 25th and 10th, respectively.

Of the 50 vendors attracting the most spending from panel participants, 29 appeared in neither the web traffic rankings nor the mobile app audience rankings.

The spending data also captures products that traffic rankings can miss, including desktop applications and AI assistants operating within messaging platforms.

Beyond subscriptions: advertising and commissions

Paid subscriptions to the largest chatbots remain relatively uncommon. In August 2026, 4.5% of eligible consumers in YipitData’s U.S. e-receipt panel had an active paid personal subscription to ChatGPT, Gemini or Claude, up from 2.1% a year earlier. These figures do not cover subscriptions to other AI products.

Among the 44 AI-native products in the web rankings, 84% offer subscriptions, while 64% charge for usage or additional credits. These approaches can overlap. Only 14% use advertising, and 2% earn revenue from transaction or platform fees.

The authors see potential in assistants through which users buy goods and book trips. Such services could earn commissions on purchases. Instinct and Muse have described plans to introduce this model, but neither was collecting these fees when the report was prepared.

The cost of running AI models remains a barrier to expanding free access, the authors note. Advertising and transaction fees could provide alternative revenue streams, although these business models often require a large user base.

The spending figures are drawn from U.S. consumer panels and should not be interpreted as total company revenue or global market spending.

Earlier, Qazinform News Agency reported that OpenAI would begin introducing invisible watermarks in ChatGPT and Codex text outputs in the European Union.