AI news ·
Half of US uses AI tools but only 4.5% pay for them
Nearly half of US consumers use AI tools, but only 4.5% pay for a subscription. The top 1% of spenders accounts for almost a fifth of all AI spending, averaging $900 a month.
Nearly half of US consumers now use AI tools, but only 4.5 percent pay for a subscription to services like ChatGPT, Gemini, or Claude, according to Andreessen Horowitz's seventh Top 100 AI ranking. The data, pulled from YipitData card-level transactions, reveals a market where broad adoption masks an extreme concentration of revenue among a small group of professional users.
The ranking, released October 10, marks the first time a16z has tracked actual consumer spending rather than just traffic and app downloads. The list itself is stabilizing: only eleven products appeared for the first time this round, the fewest new entries since the ranking began three years ago. ChatGPT maintains a wide lead over Gemini and Claude.
Where the money actually goes
The top one percent of spenders accounts for nearly a fifth of all observed AI spending - more than the entire bottom half combined. That top tier averages about $900 per month and continues to climb. The typical paying user spends roughly $25 per month, a figure that has held steady for some time.
This split means the subscription numbers tell two different stories. For most users, AI is a low-cost utility. For a narrow slice of professionals, it functions as work infrastructure with spending patterns closer to SaaS platforms than consumer apps.
A power-user business in consumer clothing
The 4.5 percent conversion rate from nearly half of US consumers signals an enormous gap between casual use and willingness to pay. High model costs are pushing the industry toward subscription models over advertising, making this monetization gap a structural problem rather than a temporary one.
For founders building on these platforms, the concentration of spending rewrites assumptions about who the customer actually is. A small group of professionals spending $900 a month props up averages that would otherwise look thin with typical payers at $25. That is not a mass-market subscription business yet. It is a power-user business wearing a consumer mask.
Why this matters for executives and strategy
The spending data forces a hard question about pricing and product design. Building for the median user - the casual experimenter paying nothing or $25 - means competing for a pool of revenue that barely moves. The users driving real revenue treat AI as core workflow infrastructure, not novelty. For professionals in AI for Executives Courses, the takeaway is direct: the market is telling you to design tools and team workflows for the power-user tier, not the mass adopter.
OpenAI's lead in this panel data also matters for competitive positioning. Gemini and Claude are fighting for a distant second while the economics of running these models make the conversion problem more urgent. The question for 2027 is not whether usage grows - it will. It is whether the paying slice ever widens beyond the professionals already treating AI as a line-item expense.