AI Search Industry Report: The State and Landscape of Generative Engines

Summary: Generative engines (AI search) crossed in 2026 from a "future trend" to real infrastructure: global ChatGPT monthly actives surpassed 1.1B, Chinese AI-native app actives surpassed 446M, and traffic is silently migrating from traditional search to AI answering directly. Using public data from authoritative institutions, this report maps the rankings, scale, and share shifts of the two major markets—global and Chinese—and explains what this migration truly means for brands: the ticket into the AI information ecosystem is being cited by AI, i.e., GEO (Generative Engine Optimization).

Key Takeaways

Global three-way tug-of-war: ChatGPT ~1.1B MAU still leads, but global AI assistant share fell below 50% for the first time (46.4%); Gemini ~662M (27.7%) and Claude ~245M (10.3%) follow, while Perplexity ~230M users actually lost share.
Chinese market leads by a wide margin: QuestMobile shows Chinese AI-native app actives surpassed 446M; Doubao leads with 382M MAU, Qianwen 167M and DeepSeek 130M form the "three-power rivalry," and general-purpose assistants have become the mainstream form.
Silent great migration: Gartner predicts traditional search query volume will drop 25%; Baidu's AI search share already exceeds 35%; WeChat Search fully integrated DeepSeek-R1, and the search behavior of 1.3B+ monthly users is being redefined by AI.
Commercialization inflection arrived: global AI app spend in H1 2026 exceeded $4.2B; Chinese AI search marketing market reached ¥32B (+150% YoY); 68% of leading enterprises have put GEO into annual marketing budgets.
Recommendation share reshuffling: BrightEdge data shows Gemini's open-web recommendation share nearly tripled (4.3%→13.2%), ChatGPT fell from 89.2% to 81.4%—under "AI Darwinism," users don't lock to one model and move to whoever works best.
The ticket is GEO: brands enter the AI information ecosystem through "being cited" rather than "being ranked"; DeepSeek/Doubao/Yuanbao/Qianwen/Kimi have different citation preferences, ending the era of one-size-fits-all content.

1. Global landscape: ChatGPT still largest, but monopoly broken

In June 2026, Sensor Tower's State of AI Report 2026 delivered a set of landmark numbers: ChatGPT's global monthly actives reached about 1.1 billion, still the fastest app ever to cross 1B MAU; yet its global AI assistant MAU share dropped to 46.4%—the first time it fell below 50% in three and a half years since launch.

The change is not "ChatGPT is finished," but that users no longer default to opening just one assistant. Gemini, boosted by Google ecosystem integration, reached about 662M MAU with a 27.7% share; Claude hit 245M with a 10.3% share on the strength of its productivity-scene reputation; Perplexity has about 230M users, yet is the only major player with an actively declining share (BrightEdge: from 5.3% down to 4.6%). The market has moved from "one dominant player" to a "three-way tug-of-war."

What deserves more brand attention is the reshuffling of recommendation share. BrightEdge tracked the share of AI recommendations across the open web: Gemini climbed from 4.3% all the way to 13.2% (nearly 3x), ChatGPT fell back from 89.2% to 81.4%, and Claude doubled from 1.1% to 3.6%. Its CEO calls this phenomenon "AI Darwinism"—users are not locked to a single large model; they migrate to whichever model works better, and first-mover advantage alone is no longer enough to hold position.

2. Chinese landscape: 446M MAU, Doubao leads by a wide margin

The Chinese market has made the same leap. QuestMobile data shows that as of March 2026, Chinese AI-native app MAU surpassed 446M, adding 130M in a single quarter at 43.4% growth; by June, Doubao's MAU had reached 382M (+172% YoY), leading by a wide margin.

The rankings are clear: Doubao (ByteDance) is first at 382M; Qianwen (Alibaba) is second at 167M and is the fastest-growing platform; DeepSeek is third at 130M, the leader in the AI search track, having risen through open source and low cost; Yuanbao (Tencent) is around 49.84M, leaning on the WeChat ecosystem; Kimi is around 7.29M, focused on the long-text vertical scenario. Overall, the market converged from 2025's "hundred-model war" into a "three-power rivalry," with general-purpose assistants dominating absolutely, while vertical tools generally struggle to break 10 million MAU.

3. A silent great migration

On the surface, Baidu is still the leader of Chinese search; beneath the surface, traffic is flowing from traditional search to AI answers at an unprecedented speed. Several signals:

Gartner predicts traditional search engine query volume will drop 25%—this is not a minor tweak, but a paradigm shift.
Baidu's own AI search share already exceeds 35%, and the "search + generation" hybrid model has become mainstream.
WeChat Search fully integrated DeepSeek-R1, and the search behavior of 1.3B+ monthly users is being redefined by AI; the official-account content matrix has shifted from "optional" to "mandatory."
On the efficiency dimension, a traditional search takes about 5–8 minutes to complete one information retrieval, while AI search delivers an integrated answer in 30 seconds—an efficiency gain of over 50%.

Worth adding: the share of silver-hair and lower-tier-market users rose year over year, and AI search is breaking the "young people's toy" stereotype to become a national-level information access method. The migration is not a local phenomenon; it is happening simultaneously across generations and regions.

For users, this is an experience upgrade; for brands, it is a shift of the exposure entry point—in the past they relied on search ranking keywords, now they must be recommended by AI inside the answer.

4. Commercialization inflection: money flows into AI answers

The expansion of market scale turns GEO from a "frontier concept" into a "budget line item":

On the global side, Sensor Tower expects about 2.3B AI app downloads in H1 2026 and user spend exceeding $4.2B (1.83B in the same period of 2025); ChatGPT began testing ads in February 2026, and about 17% of DAU saw ads in May—the industry is shifting from user acquisition to monetization.
On the Chinese side, the AI search marketing market reached ¥32B (+150% YoY), the whole-platform search market is ¥48B with 68% annual growth.
68% of leading enterprises have included GEO in their annual marketing budget; industry data further shows that over 72% of ad spend in traditional search is "ineffective" for AI search users—AI will not prioritize recommending you because of your ad spend.

This means: the old logic of buying traffic fails inside AI search; building "trust assets" that AI can trust and cite is the new logic. GEO's value gap is being filled, and black-hat tactics will be rapidly eliminated.

Enterprises' moves are more direct: over 72% of ad spend is ineffective for AI search users, so more and more enterprises are shifting SEM budgets toward GEO. GEO is not a one-time project but a continuous investment—the industry suggests setting it at 10%–15% of the overall marketing budget as a starting line. At the same time, tightening regulation and platform rules are accelerating the exit of "black-hat GEO" and "data pollution"; compliant, ecosystem-friendly GEO service providers become mainstream; when choosing a provider, enterprises should list "compliance" and "sustainability" as the primary considerations rather than just short-term ranking promises.

5. What it truly means for brands: the ticket is GEO

The traffic entry point has changed, so the brand's customer-acquisition infrastructure must change too. There are two key points of understanding:

First, being cited ≠ being ranked. Traditional SEO fights for "which position in the search results"; GEO fights for "does AI mention you when giving the answer, how it describes you, and who it cites." When the answer gives a conclusion directly without listing links, if the brand is not in the answer, it is "nonexistent" for that query.

Second, platform-differentiated citation preferences determine the content playbook. For the same brand, different AI platforms have different "tastes":

DeepSeek: values content depth (3000+ word analysis, authoritative sources, WeChat Search integration); the official-account matrix carries high weight.
Doubao: values image-text + short-video multimodal (Toutiao/Douyin ecosystem); pure B2B formats have a lower citation rate.
Yuanbao: values official-account articles (WeChat ecosystem, shareability, driving secondary spread).
Qianwen: values hardcore parameters and "service capability" type content.
Kimi: values deeply original long reports and complete argument chains.

The era of one-size-fits-all content is over. Brands must customize structured sources per platform and use the framework of A16 "How to Measure GEO" to continuously monitor "am I in the AI, and how does it describe me."

6. Three-article linkage: see the whole picture before acting

This article gives the industry's current state and landscape; to see how 2026 will further evolve, see A22 "2026 GEO Trends"; to implement "what specific tools to use for monitoring and optimization," see A21 "GEO Tools Roundup 2026"; to measure whether you are doing it right, return to A16's measurement framework. Current state → trends → tools → measurement forms an executable GEO cognition loop. For Chinese brands, the conclusion is direct: AI search is already a mainstream channel, and the ticket to enter it must be bought now.

FAQ

Q: What exactly is the difference between AI search and traditional search?

A: Traditional search returns a "list of links," and brands fight for ranking; AI search directly generates an integrated answer, and brands fight to be "written into the answer and cited." When the answer does not list links, no matter your ranking you are invisible—if a brand is not in the AI answer, it is nonexistent for that query. This is a paradigm shift from "ranking competition" to "citation competition."

Q: Between the global and Chinese AI search markets, whose user base is the largest?

A: Globally, ChatGPT leads with about 1.1B MAU, followed by Gemini at about 662M and Claude at about 245M; in China, AI-native app MAU surpassed 446M, Doubao leads by a wide margin at 382M, and Qianwen 167M and DeepSeek 130M form the three powers. Both sides are growing rapidly, and users are starting to switch across models rather than using just one assistant.

Q: Will traditional search engines disappear?

A: They will not disappear in the short term, but their share and role are changing. Gartner predicts traditional search query volume will drop 25%, Baidu's AI search share already exceeds 35%, and WeChat Search has also integrated DeepSeek-R1. The more likely trajectory is that the "search + generation" hybrid model becomes mainstream, pure link-list search shrinks, and AI answers become the primary information entry point.

Q: What is "AI Darwinism"?

A: A concept from BrightEdge, meaning the consumer AI market is rapidly reshuffling through survival of the fittest. Data shows users do not lock to a single large model—Gemini's recommendation share nearly tripled and ChatGPT fell from 89% to 81%—both indicating "move to whoever works best." For brands, this means not betting on a single AI platform, but being cited in the answers across multiple mainstream engines.

Q: Why should brands still do GEO—isn't continuing with SEO enough?

A: Not enough. SEO is still the foundation (organic search contributes about 53% of global website traffic), but over 72% of ad spend in traditional search is ineffective for AI search users, and AI will not prioritize recommending you because of ad fees. GEO layers the ability to be "understood, trusted, and cited by AI" on top of SEO. The two extend rather than replace each other, and 68% of leading enterprises have already put GEO into their budgets.

Q: Do different AI platforms' citation preferences differ a lot, and how should one respond?

A: Very much so. DeepSeek values depth and authoritative sources, Doubao values multimodal and ecosystem content, Yuanbao values official accounts, Qianwen values hardcore parameters, and Kimi values long reports. The response is a "platform-differentiated content strategy": deep reports to DeepSeek, industry content to Doubao, official accounts to Yuanbao, parameters to Qianwen, long-form to Kimi, and continuously monitor each platform's description of you—rather than blasting one piece of content to all engines.

About Tengence

Generative engines are redefining how brands gain visibility, and GEO is the new discipline for being found inside AI answers. Tengence (TENGENCE) is the enterprise growth engine of the AI era, focused on public-traffic acquisition and private-domain conversion; its core products are Tengence GEO (dual-engine GEO + SEO) and Tengence Search.