The single most-quoted number for AI search reach is 2.5 billion, the count of people Google says see its AI Overviews every month. It is also the least meaningful. Nobody chose to become an “AI Overviews user”; the summary simply appeared above the ten blue links they were already scrolling. That gap between being counted and choosing to search is why every headline about AI search adoption in 2026 disagrees with every other one.
Ask five sources how many people use AI search and the honest range runs from 18% to 2.5 billion. Both numbers are correct. They measure different things. This report separates the four behaviors that get flattened into the phrase “AI search,” verifies each figure against its primary source, and shows why the numbers only look contradictory until the definitions are pinned down.
Key takeaways
- There is no single “AI search” number. Reported reach in 2026 spans passive exposure (2.5 billion monthly Google AI Overviews users), general-purpose chatbot scale (900 million weekly ChatGPT users), self-reported survey behavior (42–60% of U.S. adults), and query-level tracking (18% of Google searches produce an AI summary).
- About half of U.S. adults now use AI chatbots (49%), up from a third in 2024, and roughly one in four use them daily (Pew Research Center, Feb 2026).
- Searching for information is the single most common chatbot use (42% of U.S. adults), ahead of work tasks (38% of employed adults).
- 60% of U.S. adults say they read AI summaries at the top of search — but behavioral tracking found those summaries appeared on only 18% of actual searches.
- When an AI summary appears, clicks to websites roughly halve (8% vs. 15%), and links inside the summary are clicked just 1% of the time.
- “Market share” for the same product ranges from 45% to 77% depending on whether it is measured by app installs, web traffic, or chat sessions.
- The traffic that does arrive from AI is unusually valuable — Adobe found AI-referred retail visitors converted 42% better than non-AI traffic in March 2026, a reversal from 38% worse a year earlier.
- Gartner’s prediction of a 25% search-volume drop has not shown up in Google’s numbers — Google reported queries at an all-time high and search revenue up 19% in Q1 2
1. The headline numbers
Every major AI platform now publishes a reach figure, and the numbers are genuinely large. The problem is that they are not comparable to each other. A weekly-active-user count, a monthly-active-user count, and a “feature exposure” count answer three different questions, and stacking them into one ranking is the origin of most of the confusion in this category.
Here are the reported 2026 reach figures for the major AI search surfaces, with the metric type each one actually represents:
| Surface | Reported reach | What the metric counts | Source (date) |
|---|---|---|---|
| Google AI Overviews | 2.5 billion / mo | Passive exposure — summary appears above results | Google, I/O 2026 |
| Meta AI | 1 billion / mo | Cross-app monthly users in Meta’s apps | Meta (2025) |
| Google AI Mode | 1 billion / mo | Monthly users of the opt-in AI search tab | Google, I/O 2026 |
| ChatGPT (app) | ~1 billion / mo | App monthly active users | Sensor Tower / Reuters |
| ChatGPT | 900 million / wk | Weekly active users, general-purpose product | OpenAI (Feb 2026) |
| Google Gemini (app) | 900 million / mo | Standalone chatbot app monthly users | Google, I/O 2026 |
| Microsoft Copilot | ~420 million / mo | Product monthly users across surfaces | Microsoft Q3 FY26 |
| Perplexity | ~50 million / wk | Weekly search queries (not users) | Similarweb (2026) |
The top of that table is doing something subtle. Google’s own I/O 2026 disclosures put AI Mode — its dedicated, conversational AI search experience — at more than one billion monthly users a year after launch, with queries “more than doubling every quarter since launch.” That is a genuine adoption number: people actively chose that surface. The 2.5-billion AI Overviews figure is a different animal. It counts everyone who saw an AI-generated summary appear on a results page, whether or not they wanted it, read it, or even noticed it. Google reported AI Overviews reach climbing from 1.5 billion monthly users in May 2025 to 2 billion by the Q2 2025 earnings call to 2.5 billion by I/O 2026. Impressive as reach, but it is exposure, not usage.
ChatGPT’s numbers are the cleanest active-usage figures in the category. OpenAI reported 900 million weekly active users in February 2026, alongside 50 million paying subscribers — up from 300 million weekly users in December 2024 through 400 million, 700 million, and 800 million over roughly fourteen months.
| ChatGPT weekly active users | Reported |
|---|---|
| December 2024 | 300 million |
| February 2025 | 400 million |
| July 2025 | 700 million |
| October 2025 | 800 million |
| February 2026 | 900 million |
There is an important caveat buried in every row of both tables: these are self-disclosed platform metrics, released by the companies themselves, not independently audited. “Weekly active” and “monthly active” are defined internally and differently by each provider, and a “user” of an embedded feature is not the same as a “user” who opened an app on purpose. The numbers are real; their comparability is not.
Reported 2026 reach of major AI surfaces. The largest figure measures passive exposure, not deliberate search.
2. Fracture one: exposure is not intent
The single largest category error in AI-search reporting is treating Google AI Overviews exposure as equivalent to deliberate AI search. They belong on opposite ends of a spectrum.
Deliberate AI search is a person opening ChatGPT, Perplexity, Gemini, or Google’s AI Mode and typing a question because they want an AI-generated answer. This is an active choice, and the reach numbers are correspondingly hard-won — 900 million weekly for ChatGPT, roughly a billion monthly for both Google’s AI Mode and the standalone Gemini app.
Passive exposure is a person running an ordinary Google search when an AI Overview materializes above the results. They did not opt in. The 2.5-billion figure captures this population, and because AI Overviews now ship on a large share of Google’s roughly 5 billion daily searches, the reach naturally dwarfs any standalone product.
The behavioral evidence shows how passive that exposure often is. In Pew Research Center’s analysis of the browsing data of 900 U.S. adults, covering nearly 69,000 real Google searches in March 2025, users who encountered an AI summary ended their browsing session 26% of the time and clicked a link inside the summary just 1% of the time. Exposure was high; engagement with the AI answer as a destination was low.
The practical consequence for anyone building visibility strategy is that the surfaces demand different tactics — being cited inside a Google AI Overview (a re-ranked extract of existing search results) is a different problem than being recommended inside a ChatGPT answer (a synthesis drawn from training data and live retrieval). This is the core distinction behind Generative Engine Optimization: each engine selects, ranks, and cites sources by its own logic, and a single “AI search share” number hides all of it.
3. Fracture two: users, searches, and adults are three different denominators
Even setting exposure aside, the adoption numbers disagree because they use incompatible denominators. Some count platform users. Some count individual searches. Some count survey respondents. The same underlying reality produces wildly different percentages depending on which one is chosen.
The cleanest population-level data comes from Pew Research Center’s February 2026 survey of 5,119 U.S. adults. It found about half of U.S. adults now use AI chatbots at all, up sharply from a third two years earlier:
| U.S. adult AI behavior (Pew, Feb 2026) | Share |
|---|---|
| Ever use AI chatbots (ChatGPT, Gemini, Copilot, etc.) | 49% |
| Use AI chatbots daily | 24% |
| Use chatbots specifically to search for information | 42% |
| Ever read AI summaries at the top of search results | 60% |
| Ever use ChatGPT | 44% |
Now watch two of those rows collide. The survey says 60% of U.S. adults report reading AI summaries. But the behavioral study — which tracked actual searches rather than asking people to recall — found AI summaries appeared on 18% of Google searches. These are not contradictory; they answer different questions. “What share of adults have ever seen an AI summary?” is a lifetime-exposure question with a high answer. “What share of individual searches produce an AI summary?” is a per-query question with a much lower answer. Reporting that conflates them — claiming AI Overviews “appear on 60% of searches” — is quietly swapping the denominator.
Four survey figures share one denominator (U.S. adults); the fifth counts individual searches.
The ChatGPT adoption curve shows the same sensitivity. Pew found 44% of U.S. adults have used ChatGPT, more than double the 2023 figure:
| Year | U.S. adults who have ever used ChatGPT |
|---|---|
| 2023 | 18% |
| 2024 | 23% |
| 2025 | 34% |
| 2026 | 44% |
That 44% is a share of the U.S. adult population. OpenAI’s 900 million is a global weekly-active count. Both describe ChatGPT. Neither can be substituted for the other, and averaging them would be meaningless. Adoption also splits hard by age: Pew found adults under 50 were roughly twice as likely as those 50 and older to use ChatGPT (57% versus 28%).
4. Fracture three: market share depends entirely on the meter
If reach numbers are slippery, market-share numbers are worse, because share requires both a numerator and a denominator to be defined — and the proprietary panels that publish these figures each define them differently. The same product can be reported at 45% or 77% “share” in the same quarter, and both providers are being internally consistent.
The table below shows ChatGPT’s reported “share” across four measurement methods in early-to-mid 2026. These are directional panel estimates, not audited figures, included precisely to show how far they diverge:
| How “share” is measured | ChatGPT share | Source (date) |
|---|---|---|
| AI-chatbot usage share | ~77% | StatCounter (Apr 2026) |
| Global web-traffic share among AI tools | ~53% | Similarweb (May 2026) |
| U.S. mobile-app market | ~45% | Sensor Tower / Fortune (Jan 2026) |
| “AI search” market share | ~60% | First Page Sage (early 2026) |
A 32-percentage-point spread on a single product is not a data error. It is four honest answers to four different questions: what share of chatbot sessions run through ChatGPT, what share of web visits to AI tools land on it, what share of app installs it holds, and what share of a loosely defined “AI search” category it occupies. None is wrong. None is comparable to the others.
The direction of travel is consistent across panels even where the levels disagree: ChatGPT’s share has been drifting down through 2026 — Similarweb’s web-traffic read moved from roughly 76% a year earlier to about 53% by May 2026 — as Gemini, Copilot, Claude, and Perplexity absorbed the redistributed share. Share fell while absolute usage rose, because the category grew faster than any one player.
What people actually use AI search for
“AI search” also blurs a distinction between searching and everything else a chatbot does. When Pew asked U.S. adults what they use chatbots for, information search led — but only just, and it sat alongside a long list of non-search uses:
| What U.S. adults use AI chatbots for (Pew, Feb 2026) | Share |
|---|---|
| To search for information | 42% |
| For tasks at work (among employed adults) | 38% |
| For fun or entertainment | 25% |
| To create or edit images or videos | 24% |
| To get medical advice | 20% |
| To get diet and fitness information | 20% |
| To get news | 13% |
| For emotional support or advice | 10% |
| For companionship | 4% |
This matters for interpreting the platform reach numbers. ChatGPT’s 900 million weekly users are not 900 million searchers — a large share are writing, coding, brainstorming, or generating images. Treating a general-purpose assistant’s entire user base as “AI search” inflates the search-specific population.
There is a behavioral signature that separates AI search from classic search, and Google has quantified it: the average AI Mode query is roughly three times the length of a traditional Google search. The Pew behavioral study found the same pattern from the search side — longer, question-shaped queries were far more likely to trigger an AI summary:
| Query type | Share that produced an AI summary (Pew, Mar 2025) |
|---|---|
| One- or two-word searches | 8% |
| Ten-plus-word searches | 53% |
| Question-form searches (who / what / why) | 60% |
| Full-sentence searches (noun + verb) | 36% |
| All searches | 18% |
The takeaway for content strategy is that AI answers concentrate on the informational, question-shaped queries that historically drove the most organic traffic to publishers — which is exactly why the traffic impact has landed where it has.
The traffic impact: zero clicks, higher value
The most consequential AI-search statistic for any business is not how many people use it — it is what happens to the click. Here the primary evidence is unusually clear and points in two directions at once: AI answers send fewer clicks, but the clicks they do send are worth more.
On the “fewer clicks” side, the Pew behavioral data is the strongest single dataset because it tracked real behavior rather than self-reports:
| User action | With AI summary | Without AI summary |
|---|---|---|
| Clicked a traditional result link | 8% | 15% |
| Clicked a link inside the AI summary | 1% | — |
| Ended the browsing session entirely | 26% | 16% |
Clicks to websites roughly halve when an AI summary is present, and the links inside the summary are almost never clicked. This is the mechanism behind the “zero-click” concern that dominates 2026 SEO discussion, and the reason enterprise SEO strategy has shifted toward being the cited source inside the answer rather than the tenth blue link beneath it.
Pew Research Center behavioral study, 68,879 Google searches, March 2025.
On the “worth more” side, the picture is genuinely surprising. Adobe Analytics — drawing on more than a trillion visits to U.S. retail sites — reported that the small stream of traffic arriving from AI assistants is dramatically higher-intent than average. During the 2025 holiday season, AI-referred shoppers converted 31% better than non-AI traffic, and by early 2026 the gap had widened into a reversal:
| AI-referred retail traffic vs. non-AI (Adobe) | Figure |
|---|---|
| Conversion rate, March 2026 | +42% (vs. −38% in March 2025) |
| Revenue per visit, March 2026 | +37% |
| Time on site, March 2026 | +48% |
| Bounce rate, 2025 holiday season | 33% less likely to leave immediately |
The conversion gap flipping from −38% to +42% in twelve months is roughly an 80-percentage-point swing. The mechanism is structural: a shopper who arrives from an AI assistant has already done comparison research inside the chat, so they land pre-qualified. The volume is still modest next to paid search or email — AI referral traffic to retail grew 693% year over year in the 2025 holiday season and 393% in Q1 2026, but off a small base. The signal is quality, not quantity.
Adobe Analytics; AI-referred vs. non-AI retail conversion. Vendor-stated, not independently audited.
Two caveats keep this honest. First, the Adobe figures are vendor-stated and not independently audited, though Shopify has reported a directionally similar pattern. Second, retail-shopping behavior does not necessarily generalize to B2B, informational, or research queries, where the value of a click is measured differently.
Is search dying? The prediction versus the data
The most-cited forecast in this space is Gartner’s February 2024 prediction that traditional search-engine volume would drop 25% by 2026, with search marketing losing share to AI chatbots and virtual agents. It has been quoted in nearly every AI-search article since. The 2026 data does not support it — at least not in the form it was made.
Google’s own reporting runs the opposite direction. At I/O 2026 the company said search queries had reached an all-time high, and Alphabet’s Q1 2026 results showed Google Search revenue up 19% year over year to $60.4 billion — an acceleration, not a decline. Google’s framing is that AI features are adding searches: once people use AI Overviews and AI Mode, they search more, not less, for the kinds of queries those features serve.
Both things can be true. Traditional blue-link search behavior is being displaced — the click data above makes that undeniable — while total search volume rises because the same searches now happen inside AI surfaces that Google also owns. The 25% “decline” did not materialize as a headline drop in queries; it showed up as a redistribution of behavior within search (toward AI Overviews and AI Mode) and a smaller migration out of search (toward standalone chatbots). A prediction about volume was really a prediction about interface.
This is the most important reframing in the whole category. AI search in 2026 is not primarily subtracting from search — it is changing where answers get delivered and who gets the click. The strategic question is no longer “will people still search?” (they will, more than ever) but “when an AI answer intercepts the query, is your brand the source it cites?”
Frequently asked questions
How many people use AI search engines in 2026?
There is no single correct figure because “AI search” spans four distinct behaviors. Deliberate AI search products report roughly a billion monthly users each (ChatGPT, Google AI Mode, the Gemini app). Google’s AI Overviews reach 2.5 billion monthly users, but that counts passive exposure rather than active search. In the U.S., 42% of adults say they use chatbots to search and 60% report reading AI summaries. The defensible summary: AI search has crossed into mainstream use, with hundreds of millions of deliberate users globally and billions more exposed to AI answers inside ordinary search.
What percentage of U.S. adults use AI chatbots?
About 49% — roughly half — per Pew Research Center’s February 2026 survey of 5,119 U.S. adults, up from 33% in 2024. Around 24% use chatbots daily.
How many people use ChatGPT?
OpenAI reported 900 million weekly active users in February 2026, and Sensor Tower data reported by Reuters put ChatGPT’s app at roughly 1 billion monthly active users by June 2026. Separately, 44% of U.S. adults told Pew they have used ChatGPT. The weekly figure is a self-disclosed global platform metric; the 44% is a share of the U.S. adult population — different measures that should not be combined.
What is the difference between AI Overviews, AI Mode, and a chatbot like ChatGPT?
AI Overviews are the AI-generated summaries at the top of an ordinary Google results page — users see them without opting in. AI Mode is Google’s separate, conversational AI search tab that users choose to open. ChatGPT (and Perplexity, the Gemini app, etc.) are dedicated AI products a person opens deliberately. Their reach numbers are not comparable: AI Overviews measures exposure; AI Mode and the chatbots measure deliberate use.
Do AI Overviews actually reduce clicks to websites?
Yes. Pew’s behavioral study of nearly 69,000 real Google searches found users clicked a traditional result 8% of the time when an AI summary was present, versus 15% when it was not — roughly half. Links inside the summary itself were clicked only 1% of the time.
Is AI search replacing Google?
Not in the way the phrase implies. Google reported search queries at an all-time high and search revenue up 19% year over year in Q1 2026. What is changing is how answers are delivered — increasingly through AI surfaces, many of them Google’s own — and who gets the click. Behavior is shifting within and around search rather than abandoning it.
What do people use AI chatbots for most?
Searching for information is the top use (42% of U.S. adults), narrowly ahead of work tasks (38% of employed adults). But a large share of chatbot use is non-search — writing, image creation, entertainment, and advice — which is why platform user counts overstate the “AI search” population.
Which AI chatbot has the most users?
ChatGPT, by a wide margin, on nearly every meter. In Pew’s U.S. survey, 44% of adults used ChatGPT, versus 24% for Gemini, 17% for Copilot, 14% for Meta AI, 8% for Grok, and 6% for Claude. Globally, ChatGPT’s weekly active users lead all standalone assistants.
Did search volume really drop 25% in 2026, as Gartner predicted?
Gartner’s February 2024 forecast of a 25% decline in traditional search volume by 2026 has not appeared in Google’s reported numbers, which show queries at record highs. The prediction is better read as a forecast about interface change than a literal drop in search activity, and it was contested by SEO analysts when published.
Is traffic from AI search worth anything if people don’t click?
The clicks that do occur appear unusually valuable. Adobe reported AI-referred retail visitors converting 42% better than non-AI traffic in March 2026 — a reversal from 38% worse a year earlier — with higher revenue per visit and lower bounce rates, because those shoppers arrive pre-researched. The volume is small relative to paid search or email, and the data is vendor-stated, so it is best treated as a strong quality signal rather than a settled fact.
What share of searches trigger an AI Overview?
Pew’s behavioral study found 18% of all Google searches produced an AI summary in March 2025. Trigger rates rise steeply with query length and phrasing: 8% for one-to-two-word searches, 53% for ten-plus-word searches, and 60% for question-form queries. Later 2026 SEO-panel estimates report higher prevalence, but they measure different keyword sets and are not directly comparable to Pew’s behavioral sample.
Why do AI search statistics vary so much between sources?
Because the sources measure different things and rarely say so. Reach figures mix weekly-active, monthly-active, and passive-exposure counts. Adoption figures mix platform users, individual searches, and survey respondents as denominators. Market-share figures come from proprietary panels that each define sessions, visits, or installs differently — producing a 30-plus-point spread on the same product. A figure without its definition and date is close to meaningless in this category.
Sources and methodology
This report distinguishes between what was directly verified at a primary source and what is directional or secondary. Every figure is tagged with one of the flags below.
Pew Research Center — behavioral study. “Google users are less likely to click on links when an AI summary appears in the results” (July 2025), analyzing the browsing data of 900 U.S. adults across 68,879 Google searches in March 2025. Source of the 18% trigger rate, the 8%-vs-15% click figures, the 1% in-summary click rate, the 26%-vs-16% session-end figures, and the query-length breakdown. Report.
Pew Research Center — Americans and AI 2026. Survey of 5,119 U.S. adults, February 17–23, 2026. Source of the 49% chatbot-use figure, 24% daily use, the use-case breakdown, the 60% AI-summary readership figure, the 44% ChatGPT figure and its 2023–2026 trend, the per-chatbot brand shares, and the age split. Report.
OpenAI — ChatGPT weekly active users. 900 million WAU and 50 million paying subscribers, announced February 2026; earlier milestones from OpenAI’s own prior disclosures. Reported via TechCrunch, quoting OpenAI. Self-disclosed platform metric.
Google — AI Overviews, AI Mode, and Gemini reach. AI Mode surpassing 1 billion monthly users, from Google’s I/O 2026 Search blog post; AI Overviews at 1.5 billion rising to 2 billion, per Google’s Q2 2025 earnings call reported by TechCrunch; the 2.5 billion AI Overviews and 900 million Gemini-app figures and the all-time-high-queries statement from Google’s I/O 2026 disclosures. Self-disclosed platform and earnings metrics.
Adobe Analytics (Adobe Digital Insights) — AI referral traffic and conversion. Retail AI-referral growth and the conversion reversal (−38% in March 2025 to +42% in March 2026), plus revenue-per-visit, time-on-site, and bounce figures. From Adobe’s own blog: 2025 holiday recap and Q1 2026 update. Drawn from 1 trillion-plus U.S. retail-site visits; vendor-stated, not independently audited.
Forecasts (🟠)
Gartner — search-volume prediction. “By 2026, traditional search engine volume will drop 25%, with search marketing losing market share to AI chatbots and other virtual agents,” Gartner press release, February 19, 2024. A prediction, presented here against subsequent measured data.
Secondary / directional (🟡)
- Market-share-by-meter figures (StatCounter ~77% usage share, Similarweb ~53% web-traffic share, Sensor Tower/Fortune ~45% U.S. app share, First Page Sage ~60% “AI search” share) are proprietary panel estimates that each define numerator and denominator differently; included to illustrate divergence, not as precise measures.
- Meta AI (1 billion monthly) and Microsoft Copilot (~420 million monthly) reach figures, reported via company statements and third-party trackers; included for competitive context.
- Sensor Tower / Reuters ChatGPT ~1 billion monthly app users (June 2026), a third-party estimate of a different (monthly, app-based) metric than OpenAI’s weekly figure.
Method notes
Figures were prioritized in this order: primary research organizations (Pew), then companies’ own disclosures and earnings materials (OpenAI, Google, Adobe, Microsoft), then proprietary panels reported through third parties. Where a widely quoted statistic could not be traced to a primary source, it was excluded or flagged 🟡 and attributed to its named panel. Reach, adoption, and market-share numbers are kept in separate tables because they are not interchangeable. All figures carry a date because, in a category moving this fast, an undated statistic is not usable.
Get cited, not buried
The pattern across every dataset here is the same: AI answers increasingly intercept the query before the click. Winning in 2026 means being the source AI engines cite. SERPsculpt builds GEO and AEO strategy for B2B SaaS — structuring content to appear in ChatGPT, Google AI Overviews, and Perplexity answers, not just the blue links. For related data, see the B2B customer retention statistics report, or explore the GEO and AEO approach for SaaS.