OpenAI claims a fifth of Google’s search volume. Google is already rebuilding search

ChatGPT’s search volume is approaching 20% of Google’s, Yahoo Finance reports, as more people take their questions straight to AI. The figure comes from OpenAI itself: Adam Fry, ChatGPT’s consumer product lead, put the ratio at roughly one to five. For a product that was known mainly for writing essays and code not so long ago, that is a serious bid to become one of the internet’s main entry points.

In his conversation with Big Technology, Fry explained what OpenAI counts: each user question that sends ChatGPT to the web is one search. Several such questions in a conversation count as several searches. A definition or simple explanation answered without going online does not qualify. Fry did not specify where the Google query volume used for comparison came from. The headline 20% remains OpenAI’s estimate, and it does not represent ChatGPT’s share of the overall search market.

Its search volume, it says, is approximately 20% of Google’s.

Outside researchers also put ChatGPT’s search activity at a substantial scale, though they measure it differently. As Search Engine Journal notes, Ahrefs estimated in February that ChatGPT’s search-like queries amounted to roughly 12% of Google’s volume. Its analysts started with 2.5 billion daily prompts from July 2025 and classified about 65% as questions people might previously have taken to Google. That definition is broader: a prompt can serve the purpose of a search even when the model never accesses the web. Turning 12% and 20% into a tidy market-share growth chart would be a little too convenient.

A SparkToro and Datos study covering 41 platforms offers another view. Google accounted for 73.7% of the US desktop search activity measured across those sites in the fourth quarter of 2025. ChatGPT was measured in sessions, with a multi-turn conversation counted as a whole. OpenAI counts each question that triggers a web search separately. The independent research shows how much searching now happens beyond conventional search engines—and how much the result depends on where researchers draw the boundaries of a single search.

Google is hardly waiting for someone to make its search box obsolete. We recently examined its expensive shift to AI: in its third-quarter results, the company reported its fastest search query growth in five years. Search revenue growth was slowing, however, and Google’s chief business officer, Philipp Schindler, cited AI Mode’s lower monetization as one reason. Google is actively moving users into a conversational interface that currently earns it less advertising revenue.

Bernstein analyst Mark Shmulik highlights a particularly dangerous part of that transition: what happens to commercially valuable queries. An assistant can guide someone through a purchase in one conversation, absorbing several searches and opportunities to show ads along the way. To defend its audience against OpenAI, Google is rebuilding its own money-making machine while it is running. Alphabet’s wider spending plans give a sense of the commitment: capital expenditure is forecast at $195–205 billion for 2026, with a substantial increase promised for 2027.

The significance of OpenAI’s one-fifth claim extends well beyond another prediction of Google’s imminent demise. OpenAI is challenging the habit that built Google’s empire: have a question, open our service. Google is responding by putting AI into search at scale, accepting a hit to the old economics as the new interface leaves fewer opportunities to show ads before delivering an answer. Who wins? As with much of the AI race, there is still no reliable forecast.