The top 1% of AI spenders pay about $903 a month, mostly for automation and building tools like n8n
Most people who pay for AI spend $25 a month. The people spending $900 are building automations.
What happened
Andreessen Horowitz published the seventh edition of its Top 100 Gen AI Consumer Apps, and for the first time added observed US consumer card spending from YipitData's panel.
As of August, 4.5% of US consumers had an active paid personal subscription to ChatGPT, Gemini or Claude, up from 2.1% a year earlier. Nearly half report using AI, but only 25% use it daily.
Spending is concentrated. The top 1% of payers account for 19.5% of observed consumer AI spend, more than the bottom 50% combined. That top 1% averages $903 a month, up 80% in 18 months. The median payer spends about $25 a month and has barely moved.
a16z says the top 1% disproportionately buy tools for automation and product building, naming n8n, fal, Manus and Nous Research's Hermes Agent, plus creative tools like Figma and HeyGen. The Decoder notes the figures come from panels and do not represent total revenue.
My take
The gap between $25 and $903 is not about owning a smarter model. Both groups can reach the same models. The difference is what the AI is wired into.
A chat subscription answers questions. An automation reads the new lead, updates the CRM, drafts the follow up and books the call without anyone opening a tab. That is why the heavy spenders buy n8n and agent tools: they turn the model into work that runs on its own.
For a small business owner, that is the practical read. If your team pays for a chat plan and still copies data between the inbox, the CRM and a spreadsheet by hand, you have the expensive part and are missing the useful part. The next step is not a pricier plan. It is one workflow, connected to the systems you already use, with a clear before and after.
Pick the task your team repeats every day and hates the most. Send me a message describing it, and I will tell you honestly whether it is worth automating. Start at romielwillautomate.dev.
More posts
- METR says Anthropic is improving its models about twice as fast as before, mostly thanks to AI. Plan your automations for faster model turnoverNov 9, 2026
- A free open-weight model nearly matched OpenAI's restricted cyber model at breaking into simulated company networks, for under $3 a runNov 9, 2026
- Karpathy built a video site from 43 million Hacker News comments in a weekend, and Simon Willison benchmarked a SQLite parser with ClaudeNov 9, 2026
