ChatGPT Is Not the Whole AI Market (Q2 2026 Share Data)

ChatGPT held a user share in the mid-30% range among US desktop AI users through Q2 2026. In Europe it fell from 44.82% in March to 41.91% in June. Both of those numbers are large, and neither of them is a majority.

I mention it because almost every AEO check I see, including the ones I was running, opens ChatGPT, asks a question, and treats the answer as the state of AI visibility.

Key takeaways

• ChatGPT remained the leading AI platform in Q2 2026 but held only a mid-30% user share in the US, and its European share declined across the quarter.
• Claude posted the strongest growth in both regions, rising from 8.54% to 11.51% in the US and from 9.61% to 12.60% in Europe.
• Gemini reached a European high of 22.88% and peaked above 20% in the US, making it the clear second platform in both markets.
• Users increasingly move between assistants rather than staying in one, so visibility in a single tool is a partial picture of AI visibility.

Where the US market actually sits

ChatGPT stayed the market leader through the quarter with a broadly stable user base in the mid-30% range. Stability is the interesting part. Its share did not fall while competitors grew, which suggests the overall market expanded rather than users defecting.

Gemini reached a peak of 20.34% in May before easing slightly to 19.53% in June. Claude recorded the strongest growth of any platform, climbing from 8.54% in March to 11.51% by the end of the quarter.

Perplexity, Copilot, and DeepSeek stayed in low single digits with no meaningful movement. That matters for anyone who has been told to optimize for Perplexity specifically. It is a real audience and a small one.

Europe is running a different race

The European picture diverges sharply, which matters if you sell there. ChatGPT’s user share declined from 44.82% in March to 41.91% in June, the only region where its position weakened.

Gemini finished the quarter at a new high of 22.88%. Claude grew from 9.61% to 12.60%. And DeepSeek overtook Perplexity for the first time, ending the quarter ahead of both Perplexity and Copilot.

Europe also adopts AI tools slightly faster than the US overall, at 1.91% of desktop activity against 1.83%, and it runs further ahead on Google AI Mode. If your market is European, the assumption that ChatGPT is the default is already shakier there than here.

AI tools became destinations for each other

The finding I keep coming back to is not about share at all. Among the domains people visit from AI tools, ChatGPT climbed from #10 to #4 in the US, while Gemini and Claude both entered the top ten for the first time.

Three of the top ten destinations from AI tools are now other AI tools. A year earlier it was one. Europe shows the same shape, with ChatGPT moving from #13 to #5 and Gemini and Claude joining the top ten.

People are cross-checking. They ask one assistant, then open another and ask again, the way you would get a second quote on a roof. That behavior makes single-platform visibility a genuinely incomplete measurement rather than just an imprecise one.

What this changes about checking your visibility

If you check one assistant, you are sampling roughly a third of US AI users and about four in ten European ones, and you are blind to the cross-checking behavior entirely. A business that gets recommended by ChatGPT and ignored by Gemini looks fine in a one-tool audit and loses half its opportunities.

My own habit here was lazy, and I fixed it after reading this report. Running the same five questions through ChatGPT, Gemini, and Claude takes about ten minutes and the answers diverge more than I expected, particularly on local and service queries where each tool leans on different sources.

The good news is that the underlying work does not fragment. Nobody needs three content strategies. Clear structure, a direct answer near the top, and content a machine can extract without guessing are what earn citations across all of them, which is the argument in AEO being mostly SEO with new acronyms and the practical version in getting cited by AI as a service business.

Keeping the scale honest

All of this sits inside a category that accounts for 1.83% of US desktop activity and 1.91% of European. Traditional search still commands 10.52% and 11.63% respectively, several times larger, and it rose slightly this quarter rather than falling.

US AI tool usage also paused in June after growing almost uninterrupted for over a year. One month is not a trend and it may well be a blip, though it is the largest single decline in the series so far.

So the reasonable position is that AI assistants are a fast-growing minority channel where competition is intensifying, and that treating any one of them as the whole channel is the mistake. That is a narrower claim than most AI visibility pitches make, and it is the one the data supports. I have made the case before that the data disagrees with the Google is dying narrative, and this quarter does nothing to change that.

FAQ

The AI market stopped being one product with a long tail somewhere in the last year. Checking one tool and calling it AI visibility made sense when a single platform held most of the users. It does not describe this quarter.

Sources

1. Datos (a Semrush company), “State of Search Q2 2026: Behaviors, Trends, and Clicks Across the US & Europe.” Desktop clickstream panel, April 2025 to June 2026. datos.live
2. Same report, state of AI search section, on user share for ChatGPT, Gemini, Claude, Perplexity, Copilot, and DeepSeek across both regions.
3. Same report, top AI destination domains, on AI assistants entering the top ten destinations visited from other AI tools.