China's AI Search War Has Started — 800M Users, 38.7% Penetration, Six Contenders

The AI search war in China is no longer a question of whether. It is a question of how to win.

China Academy of Information and Communications Technology (CAICT) data shows that by Q2 2026, Chinese AI search users exceeded 800 million, with penetration reaching 38.7%. Globally, generative AI search hit 1.8 billion monthly active users in Q1 2026 — 32% of all internet users. Gartner predicts traditional search engine traffic will fall 25% in 2026.

Baidu, 360, Kimi, Doubao, Qwen, and Tencent Yuanbao — every major player has entered the arena. Each is redefining "search" in its own way. For overseas brands watching from outside, the confusion is real: six AI engines, six different citation behaviors, six different ecosystems. Where do you even start?

The answer is counterintuitive: the more chaotic the war becomes, the more value there is in the one entry point that still connects to everything.

📊 The Numbers Behind the War

The scale of this shift is measurable, and the numbers are not speculative:

• 800 million Chinese AI search users, 38.7% penetration (CAICT, Q2 2026) — up 17.4 percentage points from 21.3% a year earlier
• 1.8 billion global generative AI search MAU, 32% of internet users (Q1 2026)
• 60%+ of global search requests are now "zero-click" — users get the answer inside the AI response without visiting any website (SparkToro/Datos)
• 25% expected decline in traditional search engine traffic in 2026 (Gartner)
• RMB 32 billion China AI search marketing market in Q1 2026, up 150% YoY; traditional SEO revenue down 42% (iResearch)
• 17.7 billion AI Agent requests in Q2 2026, up 45% QoQ — agents are becoming a distribution channel of their own (DataDome)

These numbers describe one fact: the information entry point has moved. It is no longer "search then click." It is "ask then decide." And the brands that appear inside the answer are the ones that get chosen.

⚔️ Six Players, Six Strategies

China's AI search market is not a monolith. Six contenders with fundamentally different strategies now divide the traffic:

Baidu is the defender. Its search share fell from 87% in 2021 to 58% today — but its AI search share remains above 60%, and in April 2026 it relaunched its AI search engine around a Master Agent that plans, calls tools, and executes tasks. The user journey has shifted from "find and know" to "get and do."

Doubao is the traffic giant, powered by ByteDance's Douyin ecosystem. Qwen embeds into Alibaba's commerce. 360 positions itself as a super-agent. Kimi owns long documents. Tencent Yuanbao embeds into WeChat mini-programs and JD.

For overseas brands, the immediate implication is that "optimize for Baidu" is no longer the only question. But neither is "optimize for six engines." The real question is which entry point gives you both reach and a business model.

🪟 The Two-Window Moment

Here is the number most overseas brand teams miss: 38.7% penetration means 61.3% of Chinese users still get their information the traditional way.

The market has not flipped yet. It is in transition. That creates a rare structural window with two usable lanes:

Lane one — traditional search still works. Paid search on Baidu still reaches the majority of users. CPC economics, keyword targeting, and conversion tracking all still function. This is the lane where money buys visibility today.

Lane two — AI answers are being formed right now. Every brand's category positioning inside Wenxin Yiyan, Doubao, and the others is being written as you read this. AI answers have a lock-in effect: once a default answer solidifies and gets reinforced through cross-platform corroboration, displacing it becomes expensive.

The window closes when penetration crosses the majority threshold and traditional search volume collapses — which Gartner's 25% decline suggests is underway. Brands that build both lanes now capture the transition; brands that wait for clarity will find the AI lane already locked.

🎯 Why Baidu Remains the Only Full-Stack Entry

Among the six contenders, one structural fact stands out for overseas brands: Baidu is the only player that connects the traditional auction, the AI answer layer, and a complete commercial ecosystem in one place.

• Traditional paid search — still the highest-intent acquisition channel in China, with the majority of users still on it
• AI answer layer — Wenxin Yiyan citations, Baike, Baijiahao, news sources
• Commercial ecosystem — Baidu Maps, Baijiahao, knowledge graph, ICP-filed official presence

Doubao wins on traffic but has no ad auction you can buy into as an overseas brand. Qwen is commerce-native but requires Tmall/Taobao presence. Kimi is research-focused. 360 is a super-agent play. Only Baidu offers an overseas brand the ability to run paid search and build AI-citable assets under one roof.

This is not a statement that Baidu is the "best" engine. It is a statement about entry points: when you are an overseas brand with limited resources and no Chinese entity, you need one door that leads to both lanes. Baidu is that door.

🏢 The BPP Perspective

The practical sequence for an overseas brand entering China right now:

1. Start paid search immediately — the traditional lane still has the majority of users, and advertiser competition is thinner than it has been in years.
2. Build AI-citable assets in parallel — Baike entry, ICP-filed official site, Baijiahao activation, and Chinese news coverage are what Wenxin Yiyan cites and cross-verifies.
3. Let the two lanes compound — paid search captures intent today while the citation layer locks the AI default answer for tomorrow.

Steps one and two both require a Chinese business entity — an ICP-filed website, a verified account structure, compliance with Chinese advertising regulations. If you do not have that entity, or do not want to spend months setting it up, that is the gap BPP exists to bridge.

The AI search war is being fought with algorithms, traffic, and ecosystems. You do not need to win the war. You need to be present in the answers that get chosen — in both lanes, before the window closes.

The question is not which engine will win. It is whether you will be in the answer when Chinese consumers ask.