360doc to shut down on May 1 — a signal that knowledge services must move from “store” to “understand”

Long-lived Chinese archive bows out as AI reshapes value

360doc Personal Library (360doc个人图书馆), a 20‑year‑old web archive long tied to many Chinese netizens’ reading habits, will cease service on May 1, it has been reported. The site and its owner — Beijing Liuzhi Information Technology Co., Ltd. (北京六智信息技术股份有限公司, “六智信息”) — have told users to back up content and apply for refunds; users are being warned that platform access will end. It has been reported that the company’s registered office appeared closed during recent on‑site reporting and that a January plan to transfer platform assets without charge to a willing partner failed to produce a buyer.

Why did a near‑century‑million‑page archive come to this? Founder Cai Zhi told reporters her small team “cannot provide the resources” to carry 360doc into the AI era. The platform’s traffic and ad model — once heavily dependent on search engines like Baidu and advertising income — has collapsed as large language models and conversational agents (ChatGPT, Kimi, Deepseek, OpenClaw and others) let users skip searching and go straight to answers. Financial statements show revenue sliding from roughly RMB 10.0m in 2023 to RMB 8.1m in 2024, and mounting losses; H1 2025 revenue fell further to about RMB 2.5m with widening net losses, underscoring the commercial squeeze.

A case study in what “knowledge” must become

Experts see 360doc’s fate as emblematic, not accidental. Liu Xingliang of DCCI calls it a classic transition from an internet‑era “container” for information to an era that prizes understanding and interaction; Zhang Yi of iiMedia likewise points to structural issues — copyright, monetization and low interactivity — that AI simply accelerated. Cai argues the real asset is the unpolluted user‑content relationships accumulated before large‑scale model training: human‑origin data that, she says, users must control if a next‑generation personal knowledge product is to be viable.

The competitive landscape is global and fierce. Google’s NotebookLM and Gemini, Microsoft’s Copilot Notebooks, and domestic moves by Tencent, ByteDance and Alibaba are all racing to make personal knowledge bases interactive, searchable and provenance‑aware. How will trust, privacy and commercial models be balanced? Who will hold the user’s data in an era marked by data‑localization policies and geopolitical scrutiny over AI training datasets? Those questions will determine which approaches survive.

Where personal knowledge goes next

360doc’s closure is more than nostalgia; it’s a prompt. Will personal knowledge live in vendor‑locked silos, in decentralized, privacy‑preserving vaults, or inside assistants owned by global cloud giants? The answer will shape who owns not just content, but the next layer of personal cognition. For now, the lesson is clear: storage alone no longer suffices — understanding, provenance and sustainable business models do.