AI answers are concentrated: who captures the citations in late 2026

Short answer. AI answers are concentrated at the top and unstable below it. In a sample of 1,000 Google AI Overviews published on April 26, 2026, Wikipedia alone accounted for 24.3% of citations and Reddit for 21.6%. Semrush's 2026 AI Visibility Index found that only 36 brands held a top 100 position on all four platforms it tracks, in every month from January to April 2026. Beneath those leaders, platforms disagree, new names enter, and a growing share of citations comes from pages outside Google's top 10. For a challenger brand, the opening is real but narrow: specific prompts, specific engines, and the third-party sources those engines read.

• Sources are concentrated. Two domains supplied 45.9% of AI Overview citations in the April 2026 sample.
• Brand concentration depends on the category. The top three brands hold 82.9% of visibility in News and Media but 41.4% in Finance (Semrush, June 2026).
• Platforms disagree. Any two platforms shared 63 to 70 of their top 100 brands in the UK (Semrush, October 2026).
• Organic rank matters less. 37.9% of AI Overview citations came from top 10 results in March 2026, against about 76% in July 2025 (Ahrefs).

How concentrated are AI citations in late 2026?

The clearest single measurement comes from Digital Applied's study of 1,000 AI Overviews, published on April 26, 2026. The team collected US English desktop results between April 8 and 22, 2026, across ten intent types. Wikipedia took 24.3% of all citations and Reddit 21.6%. The average Overview cited 4.2 sources, and commercial queries only 3.1.

The authors also report that the top 1% of cited domains, about a dozen sites, captured 47% of citations. That headline does not reconcile cleanly with their own domain table, so we treat the exact split with caution. The direction is not in doubt.

Semrush's data points the same way across a much larger sample. In the edition released on October 13, 2026, covering January to July 2026, the ten most-cited sources in Google AI Mode accounted for 23.2% to 35.8% of citations among its 10,000 most-cited UK domains. ChatGPT was flatter, at 12.8% to 15.4%. The pattern is consistent with answer length: Semrush's June 26, 2026 release reported an average of 15 sources per ChatGPT response against 3 for Gemini. Fewer citation slots leave less room.

Who captures the citations?

Three groups take most of the space.

Reference and community platforms. Wikipedia and Reddit lead the AI Overview data above. Semrush's group of 36 brands visible on every platform, every month, includes YouTube, Google, Reddit and Amazon. A likely reason: they aggregate many independent voices on a topic, which gives an engine corroboration for a claim.

Category incumbents. In Semrush's June 2026 release, the top three brands held 82.9% of visibility in News and Media and 76.9% in Consumer Electronics. Concentration is not uniform: Finance sat at 41.4% and Industrial at 42.2%. Fragmented markets leave more room in the answer.

Pages that do not rank. This is the counter-trend. Ahrefs reported on March 2, 2026 that 37.9% of URLs cited in AI Overviews also appeared in the first 10 results, against about 76% in July 2025. Ahrefs changed its parsing method between the two studies and Google moved Overviews to Gemini 3, so the drop is directional. It still means that a page outside the top 10 now has a credible path into the answer.

Is concentration the same on every engine?

No, and that is the most useful fact for a challenger. Each engine has its own answer length and its own source mix.

BrightEdge, publishing on November 20, 2025, counted the average number of brands named per shopping query:

An AI Overview that names 1.4 brands is a winner-takes-most surface. An AI Mode answer that names 8.3 is a shortlist with room for a challenger.

Agreement between engines is partial. Semrush's October 28, 2025 trend update found 61 of the top 100 brands in both ChatGPT and Google AI Mode, while the two agreed on sources only 30% of the time. The October 2026 edition put brand overlap between any two platforms at 63 to 70 of the top 100 in the UK and 70 to 78 in the US. AI Overviews had the lowest brand overlap with the other platforms in every month. Semrush's own summary: there is "no single, global approach by which AI visibility is achieved."

What does the LLM Visibility Index show in B2B categories?

The index's June 2026 edition shows the same shape on buyer prompts. On the CRM and sales leaderboard, 3 of 20 brands are cited by all four engines, while 10 of 20 are cited by one engine or fewer. On the AI coding assistants leaderboard, a single brand is cited by all four engines, and 5 of 20 are cited by one engine or fewer.

Read together, the two leaderboards describe a thin consensus layer and a long tail that each engine fills differently. A brand absent from three engines is not absent from the market. It is absent from three answers, which the index records separately through its engine coverage dimension, described on the methodology page.

Does a concentrated answer leave room for challenger brands?

Yes, in three places.

Below the leaders, the list moves. Over three months in 2025, Semrush counted 25 new brands entering its top 100, of which only two reached the top 50. The top is sticky. The middle is not.

Being named and being cited are different. In the October 2026 edition, fewer than half of the top 100 mentioned brands also appeared among the top 100 cited domains, in every month and both markets. Engines name brands on the strength of what other sites say about them. Semrush's June 2026 release attributes one brand's stable score to consistent descriptions across independent review and community sites.

Rank in Google is no longer the entry ticket. With most AI Overview citations now coming from outside the top 10, the question is whether a page answers the sub-question the engine is resolving, not where it ranks for the head term. The mechanics are covered in how AI assistants choose which brands to cite.

What should a challenger brand do about it?

• Pick the prompts with room. Consideration prompts name more brands than informational ones on every surface BrightEdge measured. Start there.
• Pick the engines with room. Concentration differs by engine. A brand cited by one engine and not the others has a measurable, specific gap to close.
• Work on the sources engines read. Reference pages, community discussions, reviews and independent comparisons carry more weight than a brand's own site in the data above.
• Track mentions and citations separately. Fewer than half of the most-mentioned brands are among the most-cited domains. One number hides the other.

Methodology and limits

The studies above measure different things. Digital Applied and Ahrefs count cited URLs or domains in Google AI Overviews. Semrush measures brand mentions and cited domains across ChatGPT, Gemini, AI Mode and AI Overviews, mainly in the US and UK. BrightEdge counts brands named per shopping query. None of them covers Claude or Perplexity, which the LLM Visibility Index includes. Most are published by companies that sell search or AI visibility tools, and several rely on samples their authors chose. Figures should be compared within a study, not across studies. The index's own numbers quoted here are taken from its published June 2026 category pages and are not recalculated for this article.