On Monday night, at an OpenAI-hosted dinner in New York, ChatGPT’s consumer product lead Adam Fry told Big Technology’s Alex Kantrowitz a striking figure about ChatGPT search volume: “ChatGPT’s volume of search-like queries is now roughly one-fifth the volume of queries of Google.”
One-fifth. Twenty percent. For scale, consider what Microsoft’s chief executive said at Google’s 2023 antitrust trial: that Bing had about 3 percent of search engine market share after more than a decade of trying. That was a share of the whole search market, not a ratio to Google’s query volume, so the two figures are not directly comparable. Now OpenAI says its chatbot handles a fifth of Google’s query volume.
It is a spectacular claim. It is also, for now, OpenAI’s claim alone — and that is the story underneath the story.
The ChatGPT search volume number comes with an asterisk
Start with what Fry actually disclosed, because the gaps are as informative as the figure. He did not specify the source of the Google search volume number behind the ratio. OpenAI has not published the data, the time period, or the method behind the comparison. What readers can assess from the announcement is the claim itself, not the comparison behind it.
Then there is the definition. On what counts as a search, Fry said: “Every time you ask a question and go search the web, that’s a search.” Under OpenAI’s definition, a search is any user input that results in a web search. Prompts that ask for a definition or a simple how-to explanation do not count. That is a defensible definition — arguably a conservative one, since it excludes definition and how-to prompts. But it is OpenAI’s definition, applied to OpenAI’s counted searches, while the source and method for the Google denominator remain undisclosed. Whether the two sides of the ratio were measured on the same basis is, at this point, an open question.
None of this makes the figure false. Missing public sourcing limits independent verification; it does not establish that no measurement underlies the ratio. But it does mean the ratio cannot be independently checked from the announcement alone, and any article that repeats “20 percent” without that qualifier is doing OpenAI’s marketing for it.
The independent estimates are lower
There is a reason to keep the qualifier attached. Independent estimates of ChatGPT’s search share run well below OpenAI’s figure. Search-data company seoClarity estimated in October 2026 that ChatGPT’s daily search activity was about 7.5 percent of Google’s. Ahrefs, an SEO tool company, estimated about 12 percent in February 2026.
These numbers come from different times, and they may not share a methodology — which is precisely the point. “Search” may not be a single, settled unit: a prompt count, a web-search-triggering query, a daily active searcher, and a share of referral traffic could be different measurements wearing the same name. Until OpenAI publishes the methodology behind its comparison, the honest statement is separately attributed estimates with a footnote, not a headline number.
For scale, Gizmodo cites a reported figure of about 16 billion Google searches per day, which would put ChatGPT’s claimed volume at roughly 3.2 billion search-like interactions daily. That is an enormous number by any standard — and still, as The Verge notes, a long way from overtaking Google.
The challenger may be riding the old challenger’s engine
There is a second irony tucked inside the announcement. OpenAI likely uses Microsoft Bing for a significant part of its searches, along with its in-house search index, and there is speculation that some queries may route through Google itself. The extent of that reliance is unclear. But if the plumbing matters, then the 20 percent figure describes searches made through ChatGPT, not necessarily searches served entirely by OpenAI’s own systems.
Read that way, the story gets stranger and more interesting. The company claiming a fifth of Google’s volume may be standing, in part, on the infrastructure of the company whose chief executive reported about 3 percent market share in 2023. Bing never beat Google. But Bing’s index may now be powering the product that claims to be doing what Bing could not. The exact split has not been established, and the speculation about Google routing remains speculation. Still, the question of whose engine answers a ChatGPT query is now a real question about the search market, not a technical footnote.
Google warned about this itself
Perhaps the most striking context for the announcement is Google’s own record. During its 2023 antitrust trial, Google described AI as a potential disruptor and argued that ChatGPT was poised to seriously compete with it. Set against that record, OpenAI’s number reads as Google’s own warning quantified — by the company it warned about.
Google’s behavior since then reads like a company that believed its own argument. It continues to reduce conventional referrals to third-party sites while investing heavily in AI tools: not just AI Overview, but additional generative capabilities, including personalized search results with visualizations and charts. OpenAI, meanwhile, calls its equivalent “Intelligent UI” — a visual and interactive interface inside chatbot replies. The two sides are converging on the same product: not a list of links, but an answer, dressed up and delivered in place.
That convergence is why the definitional fight matters more than the ratio. If search becomes answers rather than links, then it is an open question worth testing whether whoever defines what counts as a search also shapes who is seen as winning it — distinct from the established point that the undisclosed denominator prevents independent checking of this ratio. Fry also used the dinner to report that ChatGPT has grown from 1 billion to 1.2 billion weekly active users in about 60 days. User growth on that scale is a fact about distribution; the 20 percent is a claim about a contest. Conflating the two is how a press line becomes conventional wisdom before anyone has checked it.
What to watch
Three questions will settle whether this week marked a turning point or a talking point. First, was the Google figure itself disclosed anywhere, and what source and methodology sit behind it? Second, does the ratio hold if “search” is defined the way Google or an independent measurer would define it? Third, why do independent estimates sit well below OpenAI’s figure — is the gap definitional, methodological, or both?
Until those questions have answers, the fair version of the headline is not “ChatGPT catches Google.” It is narrower, and more interesting: OpenAI says its chatbot now handles a fifth of Google’s search volume, on its own definition, with the comparison’s sourcing unspecified. The race is real. The scoreboard is not yet.
