ChatGPT vs Google Statistics 2026: Search Volume and Market Share

For three years running, some corner of the internet has declared Google dead. ChatGPT was the killer in 2023, Perplexity in 2024, AI Mode's own cannibalisation in 2025. Yet in August 2026 Google is still fielding more searches than at any point in its history, and its share of classic search-engine referrals sits above 91%.

That does not mean nothing has changed. ChatGPT reached a billion monthly users in June 2026, Apple executives have testified under oath that Safari search volumes fell for the first time in 22 years, and Alphabet lost tens of billions in market value in a single afternoon on the back of one sentence in a courtroom.

This article puts the two platforms side by side using disclosed figures from Google and OpenAI, Statcounter and Similarweb measurement, panel data from Datos and SparkToro, and the antitrust-trial record. The gap is smaller than it was. It is also far larger than most commentary implies.

Key ChatGPT vs Google statistics for 2026

• Google processes more than 5 trillion searches a year, roughly 14 billion a day, per Google's own disclosure reported by Search Engine Land [1].
• ChatGPT users send about 2.5 billion prompts a day as of OpenAI's July 2025 disclosure, around 330 million of them from the US, per TechCrunch [2].
• Google received roughly 373 times as many searches as ChatGPT in 2024, and Google search volume grew 21.64% that year, per SparkToro and Datos panel research [3].
• Only around 30% of ChatGPT prompts are search-like in intent, based on Semrush clickstream analysis cited in the same SparkToro study [3].
• Google's worldwide search market share fell to 89.34% in October 2024, its first month below 90% since 2015, per Statcounter data reported by Search Engine Land [4].
• By July 2026 Google's Statcounter share had recovered to over 91%, with Bing at 4.47%, Yahoo at 1.24% and DuckDuckGo at 0.65% [5].
• Apple's Eddy Cue testified in May 2025 that Safari searches declined in April 2025 for the first time in 22 years; Alphabet stock dropped 7.51% that day, per MacRumors [6].
• google.com drew roughly 87.5 billion visits in May 2026 versus about 5.565 billion for chatgpt.com, a 15.7x gap, per Similarweb [7].
• ChatGPT hit 1 billion monthly active users in June 2026, whilst Sensor Tower put its share of the AI assistant market below 50% for the first time, per TechCrunch [8].
• Google's Gemini app passed 750 million monthly active users in February 2026, up from 650 million a quarter earlier [9].
• Unique visitors across generative AI platforms grew 57% year on year to 655 million in the year to May 2026, per Similarweb [10].
• OpenAI reported 50 million paying ChatGPT subscribers as of February 2026, per TechCrunch [11].

How many searches does Google handle compared with ChatGPT?

Google discloses more than 5 trillion searches a year, which works out to roughly 14 billion a day. OpenAI's most recent comparable disclosure is 2.5 billion daily prompts. On raw volume, Google still handles more than five times ChatGPT's entire prompt load.

The 5 trillion figure came from a Google blog post in early 2025, the company's first search-volume disclosure since it confirmed "more than 2 trillion" in 2016, as reported by Search Engine Land [1]. That is 158,548 searches a second.

OpenAI's number moved fast. Sam Altman said users were sending 1 billion messages a day in December 2024; by July 2025 TechCrunch reported the figure at 2.5 billion, a doubling in about eight months [2].

But prompts and searches are not the same unit. When SparkToro and Datos adjusted for the fact that only about 30% of ChatGPT prompts carry search-like intent, they estimated ChatGPT was fielding at most 37.5 million daily search-equivalent queries in 2024, against Google's 14 billion. That is the source of the striking 373x figure [3].

Even if you assume ChatGPT's search-like share has since doubled and apply it to the 2.5 billion daily prompts, you get perhaps 1.5 billion search-ish queries a day. Real, meaningful, and still an order of magnitude behind Google.

Has Google's search market share actually declined?

Barely, and then it recovered. Statcounter recorded Google below 90% worldwide for three months in late 2024, the first such dip since 2015. By July 2026 Google was back above 91%, which is higher than where it started the AI era.

The wobble was real but small. Statcounter figures reported by Search Engine Land show Google at 89.34% in October 2024, 89.99% in November and 89.73% in December, against Bing hovering at or just under 4% [4].

As of July 2026, Statcounter puts Google at 91.31% worldwide, Bing at 4.47%, Yahoo at 1.24%, Yandex at 1% and DuckDuckGo at 0.65% [5].

The important caveat is what Statcounter measures. It tracks referrals to a network of websites, so it captures classic link-clicking search, not chatbot sessions that never send a click. ChatGPT does not appear in the ranking at all. Google's dominance of traditional search is intact; the open question is how much activity has moved to a category Statcounter cannot see.

Is Google search volume growing or shrinking?

Growing, on every measurement that exists. Google's own disclosures, its earnings-call commentary and independent panel data all point the same way. The "shrinking Google" narrative has no volume data behind it.

The SparkToro and Datos analysis of US desktop panel data found searches across Google grew 21.64% between 2023 and 2024, the same year ChatGPT adoption exploded [3]. Google's jump from a disclosed 2 trillion annual searches in 2016 to more than 5 trillion in 2025 tells the same long-run story [1].

On the July 2026 earnings call, Sundar Pichai told investors that AI features are "driving growth in queries", that Search usage "hit an all-time high during the World Cup this year", and that Google is "now sending billions of clicks to websites every week" through AI features, per Alphabet's published remarks [12]. Search and Other revenue grew 17% in the quarter [12].

Two things can be true at once, though. Total queries are rising whilst the value of each query to the open web falls. SparkToro's 2026 follow-up found fewer than a third of US Google searches now end in a click to the web, and that AI tools send out less than 1% of all traffic [13]. Growth in searches is not growth in traffic.

What did the antitrust trial actually reveal?

The single most cited piece of evidence for Google's decline came from Apple, not from Google. In May 2025, Apple services chief Eddy Cue testified that searches in Safari fell in April 2025 for the first time in 22 years, attributing the dip to users switching to AI tools.

Cue's courtroom remark, made on 7 May 2025 during the DOJ remedies proceedings, wiped 7.51% off Alphabet's share price in a day, per MacRumors [6].

Google issued a rare public rebuttal within 24 hours: "We continue to see overall query growth in Search. That includes an increase in total queries coming from Apple's devices and platforms" [6].

Both statements can be reconciled. Safari-specific searches can fall whilst total Apple-device queries rise, because Google reaches iPhones through its own app, Chrome, voice and Lens as well as Safari's search bar. Cue also had an interest in talking up AI alternatives, since Apple was defending the roughly 20 billion dollars a year Google pays it. The episode is the clearest documented crack in Google's volume story, and it is a crack in one browser's referral channel, not in Google overall.

How do Gemini and ChatGPT compare on users?

ChatGPT is still ahead, but the gap has closed dramatically. ChatGPT passed 1 billion monthly active users in June 2026; Google's Gemini app reached 950 million by the Q2 2026 earnings call, having tripled its daily actives in a year.

The trajectory matters more than the snapshot. TechCrunch reported Gemini at 650 million monthly actives in late 2025, 750 million by February 2026, and 950 million by July 2026 [9] [8]. ChatGPT went from 800 million weekly actives at DevDay in October 2025 to 900 million in February 2026 [11], and to 1 billion monthly actives by June 2026 [8].

Sensor Tower's H1 2026 State of AI report, cited by TechCrunch, put ChatGPT's share of the AI assistant market below 50% for the first time, with Gemini at 27.7% [8]. Similarweb's web-traffic lens agrees on direction: ChatGPT's share of generative AI web visits fell from roughly 76% in June 2025 to around 53% by May 2026 [10].

Google also has a distribution weapon OpenAI lacks. AI Mode, built into Search itself, passed 1 billion monthly users after going global in October 2025 [12]. Google does not need to win the chatbot download battle if it can convert its existing 14 billion daily searches into AI interactions in place.

One caution: the two companies report different units. OpenAI has favoured weekly actives, Google reports monthly actives for Gemini, and "users" of AI Mode are people exposed to a feature inside a product they already used. None of these figures are audited, and none are directly comparable.

How does referral traffic from each platform compare?

Google remains the web's traffic engine by an enormous margin, even in its zero-click era. AI assistants as a category still send well under 1% of website referrals; Google sends billions of clicks weekly by its own account.

The cleanest visit-level comparison comes from Similarweb: google.com took roughly 87.5 billion visits in May 2026 against chatgpt.com's roughly 5.565 billion, which itself grew under 1% month on month [7]. ChatGPT ranked fifth among world websites, behind Google, YouTube, Facebook and Instagram.

SparkToro's 2026 clickstream analysis found AI tools collectively "send less than 1% of all traffic out" to the web, whilst noting that fewer than a third of Google searches now produce a click either [13]. Ahrefs' server-side data across 107,300 sites tells a similar story, with AI assistants supplying 0.35% of tracked referrals against about 30% for search engines.

For a marketer the practical reading is uncomfortable in both directions. AI referrals are growing fast from a tiny base and tend to convert well, but they cannot replace Google-scale traffic for years on current trajectories. Meanwhile Google's own clicks are thinning as AI Overviews and AI Mode answer more queries on the results page.

Do ChatGPT users actually stop using Google?

No. The overwhelming pattern in panel data is that people add ChatGPT alongside Google rather than swapping one for the other. Overlap between the two user bases is near total.

Similarweb clickstream work has found that around 95% of ChatGPT's users also visited Google in the same measurement window, and SparkToro's Datos-based research found Google usage growing fastest in exactly the period ChatGPT went mainstream [3].

The audience for generative AI is also smaller than raw visit counts suggest. Similarweb measured 655 million unique visitors across generative AI platforms in the year to May 2026, up 57%, with visits growing faster than visitors at 70% [10]. Heavier use by existing adopters, not mass defection from search, is driving the category's growth.

This is the core error of the "Google killer" frame. Substitution is happening at the level of individual tasks, such as coding help, trip planning and product comparisons, not at the level of platforms. People are reallocating minutes and query types, and Google is simultaneously absorbing the chatbot behaviour into AI Mode.

What did the "Google killer" claims get wrong?

Almost everything except the direction of travel. Every predicted collapse metric, market share, query volume and search revenue, went up instead. What actually changed was the click economy underneath Google, which is a different and quieter crisis.

The scoreboard since ChatGPT launched: Google's Statcounter share is higher in July 2026 than in late 2024 [5], disclosed search volume rose from over 2 trillion to over 5 trillion a year [1], and Search revenue grew 17% year on year in Q2 2026 [12]. Bing, whose AI-powered relaunch prompted the first wave of obituaries in 2023, holds 4.47% [5].

The claims failed for structural reasons. Chatbots and search overlap on perhaps a third of use cases [3], Google ships defaults on billions of devices, and habit is sticky. Perplexity, once the favourite "Google killer" candidate, does not register a visible share on Statcounter's search ranking at all [5].

What the sceptics got right is subtler. Safari search volume did fall [6]. Zero-click Google searches passed two-thirds [13]. ChatGPT built a billion-user product in under four years [8]. The threat to Google was never extinction; it is the erosion of the click-through model that funds the open web, and on that front the pessimists are winning the argument.

How to read these numbers

Treat every figure here as an estimate with a lens attached. Statcounter counts page-view referrals across a tracking network, so it under-represents apps, voice and anything that never sends a click. Similarweb models traffic from a browsing panel plus partnerships, which is why its numbers rarely match server logs exactly.

Company disclosures are accurate but curated. Google chose the moment to reveal "5 trillion", OpenAI chooses when to update prompt counts, and each firm picks the flattering unit: weekly actives for OpenAI, monthly actives for Google, "users" of a bundled feature for AI Mode. None are independently audited.

Panel studies such as Datos and SparkToro's cover US desktop-and-mobile panellists, not the world, and prompt-intent classification (the 30% search-like figure) involves judgement calls. The 373x ratio is a defensible 2024 estimate, not a law of nature, and the true 2026 ratio is certainly lower.

Finally, disagreement between credible sources is normal. Cue said Safari searches fell; Google said Apple-device queries rose. Both were probably telling the truth about different slices. When two honest measurements conflict, the explanation is usually the denominator.

What this means going into 2027

Plan for a bigger Google that sends fewer clicks. Query volume and search revenue are rising whilst click-throughs shrink. Budget for visibility inside AI Overviews and AI Mode, not just blue links.

Treat ChatGPT as a second, smaller search channel, not a replacement. Even generous estimates leave it an order of magnitude behind Google on search-intent volume. It deserves optimisation effort proportionate to its high-converting but tiny referral stream.

Watch Gemini harder than ChatGPT. Gemini closed a 10x user gap to near parity in about 18 months by riding Google's distribution [9] [8]. If one assistant ends up bundled with the dominant search engine, the "vs" in this article's title dissolves.

Expect the next market-share story to come from courtrooms, not Statcounter. The Safari disclosure moved Alphabet's valuation more than two years of Statcounter drift. Remedies, default-payment rulings and Apple's browser choices remain the genuine swing factors.

Distrust any 2027 prediction stated without a measurement caveat. The last three years punished everyone who extrapolated a single panel or a single quarter. The honest position is that both platforms are growing, and the open web sits between them, paying for it.

If you want help working out how much of your visibility budget belongs in Google, in AI Overviews and in ChatGPT, you can book a call with me.