News · 2026-09-30

Menlo estimates consumer AI reached $40 billion as existing users spend more

Menlo Ventures estimates that global consumer-AI spending reached $40 billion in 2026, up from its $12 billion estimate a year earlier, while estimated users rose from 1.8 billion to 2 billion. The report points to existing users converting to paid products and spending more, rather than a comparable surge in the user base. Its findings show substantial willingness to pay, with a pronounced dependence on heavy spenders.

Key facts

• Menlo’s global consumer-AI market estimate is $40 billion for 2026.
• Menlo and Morning Consult surveyed 5,067 US adults in July 2026.
• The 14% of payers spending at least $100 monthly account for 60% of spending.
• Primary source: Menlo Ventures’ 2026 consumer-AI report.

The important business change is not simply more people trying a chatbot. It is a larger payment relationship with people already using AI. A market can grow quickly when existing users subscribe, add a second tool, or move into a more expensive tier, even when adoption grows slowly. Menlo’s report, titled “The State of Consumer AI: 2026,” provides evidence for that pattern while exposing how concentrated the spending has become.

Think of a gym whose membership grows modestly while more members buy classes and personal training. Sales can rise much faster than the headcount. Menlo estimates global AI users grew about 11%, while its market-spending estimate more than tripled. The two figures have different construction methods, but the combined picture supports its interpretation that payment conversion and intensity, rather than sheer user growth, drove much of the estimated expansion.

The survey provides the most inspectable foundation. Morning Consult’s July sample was nationally representative, with demographic and regional weighting. Menlo says 55% of surveyed US AI users pay for at least one product. The report does not directly observe purchases by two billion people. It takes US survey evidence, triangulates it with third-party information, and adjusts for regional differences in adoption, internet access, age, and spending to build a global estimate.

The year-over-year headline therefore needs care. Menlo’s 2025 sample included 5,031 US adults, but spending questions were introduced only in 2026. There is no identical 2025 spending survey question to serve as a repeated baseline. The $12 billion-to-$40 billion comparison is between the firm’s market-size estimates. By contrast, its US adoption comparison used a consistent question and moved from 61% to 64% of adults. That is a different kind of evidence, even though both appear in the same report.

Concentration is the most useful anchor for product builders. Menlo says a relatively small group paying at least $100 a month accounts for most consumer spending. This suggests that high-intensity use cases can support premium products, but it also makes revenue sensitive to those buyers’ experience, budgets, and willingness to keep paying. A large nominal user base can coexist with a much smaller group providing the commercial engine.

Who pays is more varied than the word consumer might imply. Among payers, Menlo reports self-payment, support from friends or family, and employer or school payment. Respondents could select more than one source, so the shares are not mutually exclusive. This is evidence of overlapping funding arrangements, rather than a clean dollar breakdown of the global market. A personal assistant used at home can still be funded partly through work or education.

Today’s OpenAI Pro changes illustrate why that distinction matters. Higher tiers and changed allowances target precisely the users who consume substantial resources and may pay more for access. Yet willingness to pay does not establish provider profitability. A heavy user’s subscription can generate meaningful revenue while also creating meaningful computing cost. Inference economics determines how much of the payment remains after serving the workload.

The report is also narrower than the infrastructure financing debate. Bain’s 2031 scenario includes enterprise demand and future markets well beyond consumer subscriptions. Exponential View’s industry analysis uses another revenue boundary and separately models infrastructure depreciation. Menlo’s $40 billion cannot be substituted into either calculation without changing the question. It measures estimated consumer-market spending, not cash available to fund new data centers.

The strongest counterargument is methodological and economic: global extrapolation from a US survey is uncertain, a venture firm has incentives to highlight growing markets, and revenue says little about margins. Those cautions should accompany the result, not obscure the observed payment signal. Menlo’s report is valuable because it moves discussion from free-user counts toward paying behavior. The next test is whether the heavy-spending cohort remains satisfied enough to sustain repeat purchases as allowances, model choices, and competing products change.

Primary source, verified: read the paper →

Ground Truth. (2026, September 30). Menlo estimates consumer AI reached $40 billion as existing users spend more. Ground Truth. https://groundtruth.day/news/menlo-consumer-ai-forty-billion-heavy-payers.html

@misc{groundtruth:menlo-consumer-ai-forty-billion-heavy-payers,
  title  = {Menlo estimates consumer AI reached $40 billion as existing users spend more},
  author = {{Ground Truth}},
  year   = {2026},
  month  = {sep},
  url    = {https://groundtruth.day/news/menlo-consumer-ai-forty-billion-heavy-payers.html}
}

Topics: consumer-ai · industry · subscriptions · market-research