OpenAI says the volume of search-like queries running through ChatGPT has reached roughly one-fifth of Google’s query volume, a milestone its consumer product lead disclosed at an OpenAI-hosted dinner in New York on Monday night. Adam Fry, who leads ChatGPT’s consumer product, also said weekly active users climbed from 1 billion to 1.2 billion in about two months, while voice use reached 250 million people a week. The figures, shared by OpenAI itself, present ChatGPT as a search-scale product rather than only a chatbot, and they arrive as OpenAI prepares for a possible public listing.
The headline number
Fry told journalist Alex Kantrowitz that ChatGPT’s volume of search-like queries is now roughly one-fifth the volume of queries of Google. Kantrowitz published the conversation on Tuesday in his Big Technology newsletter, after the Monday-night dinner in New York hosted by OpenAI.
Fry stressed the speed of the change, noting that ChatGPT only gained up-to-date web information about a year and a half ago. Before that, the chatbot often answered that a question fell past its training date, a limitation users no longer meet in the same way.
The comparison is striking because Google still dominates traditional search, with more than 90 percent of the global search engine market on most measures. A fifth of its query volume would put ChatGPT ahead of Bing, whose search market share sits around 5 percent, according to the conversation itself.
What counts as a search
Fry’s definition is generous, and OpenAI has published neither the underlying data nor the counting method behind the claim. He described a search-like query as basically anything a user would traditionally have gone to a traditional search engine for, but now takes to ChatGPT instead.
Pressed on the detail, Fry said that every time a user asks a question and ChatGPT searches the web, that counts as a search. That rule matters because one conversation can generate several searches, since many prompts lead ChatGPT to check external sources for authority or fresh facts.
Fry did not put a specific number on how many prompts qualify, so no absolute daily total for ChatGPT can be derived from his words alone. Gizmodo, reporting the interview, noted that Google reportedly handles about 16 billion searches a day, which would imply roughly 3.2 billion comparable ChatGPT interactions.
That arithmetic is an estimate built on a reported Google figure, not an OpenAI disclosure, and it should be read that way. It shows the scale implied by the claim, without turning an outside estimate into an official total from OpenAI.
Growth beyond search
The other numbers Fry shared point to growth on several fronts at once, not only in search-style questions. Weekly active users rose by about 200 million in roughly 60 days, or close to 100 million new users every 30 days, after passing 1 billion in late August.
Users now create about 1.5 billion images with ChatGPT each week, and voice use, counting both dictation and spoken conversation, reaches about 250 million people weekly. Together, the figures describe a product growing as a daily habit, not only as an occasional question box.
One number Fry would not update was paying subscribers, which may be the most important gap in the story. He repeated the most recent public figure, 50 million paying subscribers, and said it has grown, without giving a new total.
That leaves the business question open: a huge free audience does not by itself show how many users convert to paid plans, or how fast that conversion is moving. Investors will watch that gap as closely as the user totals.
Chat and Work come together
Fry also revealed plans to merge ChatGPT’s Chat and Work modes, removing the toggle that currently sits at the top of the product. The move fits the search story, because Work mode is where longer, document-heavy tasks live, and folding it into the main chat would put more of a user’s working life inside one conversational surface.
The site has already covered the Chat and Work tabs merge, and Fry’s comment now confirms the direction of travel rather than announcing a date for the change. No timetable was given for when the toggle disappears from the product interface.
Why It Matters
If even the direction of Fry’s claim holds, the search market is no longer a contest between engines alone. It is now also a contest between an engine and a conversational assistant that answers first and cites sources second, a shift with real consequences for publishers.
Publishers and advertisers care about that distinction, because a chatbot answer can satisfy a question without sending a click to any website. That traffic question is already visible in monetization, where ChatGPT ads earn far less than Google search on a per-query basis.
Scale in queries does not yet mean scale in advertising revenue, and that gap frames the whole story. Google, for its part, is not standing still, and its AI Mode ads are now expanding to more countries as it tries to monetize AI answers on its own terms.
Readers can follow that response in the Google AI Mode ads expansion to 10 countries, which shows the incumbent defending the same ground from the other side. The contest is about who owns the first question, not only who owns the best index.
Our read
Treat the one-fifth figure as a claim about query behavior, not as market share, referral traffic, or proof that ChatGPT has replaced Google for any particular task. OpenAI’s counting rewards its own design, because ChatGPT now searches the web for many kinds of questions, including finance and health queries where freshness matters.
Ordinary conversation therefore inflates the total in a way a typed search box never would, and an independent measurement is still missing. A published method and a stated time period would be needed before the figure can be compared cleanly with Google’s own totals.
Even with that caveat, the underlying shift is hard to dismiss, because it matches how people describe their own habits. Users increasingly take person, place, product, and factual questions to ChatGPT first, and only fall back to a search engine when they want a list of links.
That behavior, if it holds, changes what search optimization means: ranking on a results page matters less when the answer itself is the destination. The next numbers to watch are paying subscribers and referral traffic, which will show whether attention is turning into money and visits.
Background
ChatGPT’s search story is short, which is part of why the claim landed so hard: web access arrived only about eighteen months ago, a standalone SearchGPT effort was folded back into ChatGPT, and its Atlas browser has drawn little attention since launch. OpenAI’s pitch now is that the chatbot, not a separate search product, is where the queries already live.
Source: Big Technology interview and Gizmodo report
0 Comments