On November 3, 2026, Adam Fry, OpenAI's consumer product lead for ChatGPT, told a San Francisco press briefing that search queries handled by ChatGPT now equal 20 percent of the volume running through Google. It is the first time OpenAI has benchmarked itself against the search incumbent with a quantified ratio, and the first time a four-year-old conversational product has reached statistically meaningful scale against a business Google has owned for over two decades. The significance lies less
Speaking at a Monday press event in San Francisco on November 3, 2026, Adam Fry, OpenAI's consumer product lead for ChatGPT, framed the comparison around user behaviour: in some circles, he said, "Ask Chat" has replaced "Google it" as the default verb. OpenAI did not disclose absolute query volume, the measurement window, or whether the figure includes API calls, repeated queries and automated retrievals fired by agents. The 20 percent is therefore a self-reported relative metric rather than audited third-party data. The disclosure lands roughly two years after ChatGPT's search feature went live in October 2024, a period in which OpenAI folded search into the main conversational flow rather than keeping it as a separate entry point, and pushed weekly active users to the 800 million range during 2025. The remark was read externally as a direct challenge to Google, whose parent Alphabet
Search has been a one-player business for more than two decades, with Google holding roughly 90 percent of global general search in third-party trackers and search ads funding most of Alphabet's cash flow. That changed in 2024: OpenAI prototyped SearchGPT, then folded search into ChatGPT in October, while Google answered with AI Overviews, stacking generated summaries above the ten blue links. Both companies did the same thing at the same time, swapping links for answers. The real catalyst came from distribution. In 2025 a US court ruled that Google could no longer lock in default search placement through exclusive agreements, giving ChatGPT, Gemini and Perplexity their first opening to
Liu Gong's take: Google will disclose a query share for its AI Mode within the next six months, most likely on an earnings call rather than at a developer keynote, because once advertisers accept OpenAI's 20 percent framing, the negotiating anchor moves from cost-per-click to impressions inside a conversation. This is not a share fight; it is a fight over the unit of account.
From a technical standpoint, 20 percent of queries is not 20 percent of information retrieval. Google's index spans hundreds of billions of documents with freshness and authority signals maintained continuously, while ChatGPT still runs a hybrid of licensed indexes and its own crawler, and the recall gap shows up on obscure entities and breaking-news queries. But conversational retrieval splits a single information need into a chain of follow-ups. A classic search session averages one or two queries; a chat session can fire five or more rewrites and clarifications. The 20 percent therefore likely maps to a materially smaller number of distinct needs. Liu Gong sees the same pattern in industrial vision inspection: a line that captures ten frames per part instead of one has ten times the captures, not ten times the defect-detection rate. Metrics that are not aligned on definition flatter whoever publishes them.
On the business side, OpenAI needs this number. Subscription growth flattens eventually, and search advertising is the largest pool of money that can be bolted onto a chat interface. Turning query volume into a citable public metric is, in effect, pre-selling ad inventory.
For the industry, the damage lands at the entry point. Once users default to asking rather than searching, the distribution value of browsers, default search deals and even app stores gets repriced. Liu Gong predicts that before mid-2027 at least one major Android OEM will drop a single search engine as the default knowledge source in its system assistant in favour of multi-model routing, and on that day 20 percent will look conservative.
- ChatGPT search volume equals 20 percent of Google's, a self-reported relative figure with no disclosed absolute volume or measurement window
- Google's global general search share has long sat near 90 percent in third-party trackers, against ChatGPT's current 20 percent of that query base
- ChatGPT reached roughly 800 million weekly active users in 2025; public estimates of Google's daily queries range from about 8.5 billion to 14 billion
- ChatGPT search launched in October 2024, roughly 25 months before this disclosure; Google Search has operated for nearly 30 years since 1997
- Inference costs fell by roughly an order of magnitude in two years, the precondition for running a generation
For the industry, the near-term reaction within three months will come from the advertising side: buyers will press Google to break out AI Mode and classic search metrics in earnings disclosures, or budget negotiations will lack a reference point. Perplexity, Microsoft Copilot and other second-tier players are likely to publish similar relative ratios, turning "benchmarking against Google" into standard rhetoric. Within a year, if the 20 percent figure is corroborated by third-party data, search ad pricing will likely split into two systems, link impressions and in-conversation impressions, and the value of Google's default-placement agreements will be repriced downward.
For ordinary users,
Questions worth exploring:
- The definitional question: whether 20 percent counts queries, sessions or distinct information needs determines how much the number is worth
- Monetisation: in-conversation ad pricing is still unformed, and when OpenAI rolls it out will reshape its revenue mix
- Entry-point migration: control over system assistant and browser defaults is becoming the key lever in the next distribution war
Further reading: 2026 Frontier LLM Comparison: A Practical Guide — a durable, evergreen reference. Bookmark it.
This item comes from The Verge AI (published 2026-11-03 18:30:42). Wenfeng Radar provides a Chinese/English summary and commentary on overseas AI developments; the original content is owned by its author.
Daily aggregation of overseas AI developments with in-depth analysis. Bookmark us to never miss an important signal; back to homepage.