- An Epoch AI and Ipsos survey found that 80% of Claude’s weekly US users live in households earning more than $100,000 annually—the highest income concentration among major AI assistants tracked.
- By comparison, 64% of Microsoft Copilot’s weekly users and 56% each of ChatGPT‘s, Grok‘s, and Google Gemini’s weekly users fall in that income bracket; Meta AI stands at 37%.
- In absolute terms, Claude reaches only 6% of high-income US adults, trailing ChatGPT (37%), Gemini (24%), and Copilot (14%).
- The findings come from three survey waves conducted between March and April 2026; 44% of high-earning Americans report using no AI assistant at all.
What Happened
A survey conducted by Epoch AI and Ipsos across three waves between March and April 2026 found that 80 percent of US adults who use Anthropic’s Claude weekly live in households earning more than $100,000 per year. That figure is the highest income concentration recorded for any major AI assistant in the study, exceeding rivals by a margin of at least 16 percentage points.
Why It Matters
The income distribution findings arrive alongside a separate Anthropic study showing that users of more capable AI models tend to negotiate better prices in economic transactions, while users of weaker models were often unaware of the disadvantage. If that pattern holds, the concentration of more capable tools among higher-income users could reinforce existing economic asymmetries rather than flatten them.
Epoch AI explicitly cautioned against overstating the proportional figures, noting that Claude’s lead is “only relative—in absolute terms, its reach remains small.” That qualifier is important context when comparing across platforms.
Technical Details
Among Claude’s US weekly users, 80% live in households earning over $100,000 annually. Microsoft Copilot follows at 64%, while ChatGPT, Grok, and Google Gemini each register 56%. Meta AI records the lowest figure at 37% among the platforms surveyed.
When ranked by absolute reach among high-earning US adults—rather than proportional concentration—the picture reverses substantially. ChatGPT leads with 37% usage among that demographic, followed by Google Gemini at 24% and Microsoft Copilot at 14%. Claude accounts for just 6% of high-income users in absolute terms. The survey also found that 44% of high-income US adults report using no AI assistant at all, indicating the overall market remains far from saturated even in its highest-adopting income segment.
Who’s Affected
Anthropic faces a split reading of these numbers: Claude’s user base skews wealthy by proportion, but its total footprint among high-income Americans remains modest compared to ChatGPT and Gemini. Google and Microsoft each outpace Claude in absolute high-income reach despite lower concentration ratios, while Meta AI’s heavy representation among lower-income users reflects a distinct product-market trajectory tied to its free, broadly distributed deployment. Enterprise procurement teams and AI platform strategists evaluating adoption patterns will find the absolute-reach figures more operationally meaningful than the proportional concentration data.
What’s Next
Epoch AI has not announced additional survey waves beyond the three conducted between March and April 2026. Anthropic has not publicly addressed the income skew findings. The relationship between AI model capability, user income, and downstream economic outcomes—flagged by Anthropic’s own transaction-pricing research—has not been the subject of published longitudinal study, leaving the causal direction of these correlations unresolved.
