Zhihu Profits in 2025, Pivots to AI
💡Zhihu's profitable AI pivot: strategy for knowledge platforms vs. novel giants
⚡ 30-Second TL;DR
What Changed
First full-year profitability (Non-GAAP) of 37.87M CNY after Q4 2024 turnaround.
Why It Matters
Zhihu's AI embrace counters revenue woes and competition from free novel apps, potentially boosting content supply via AI writing. Success hinges on balancing cost control with AI investments amid traffic uncertainty.
What To Do Next
Test DeepSeek-R1 integration in your search bar for Q&A platforms using custom corpora.
Key Points
- •First full-year profitability (Non-GAAP) of 37.87M CNY after Q4 2024 turnaround.
- •Revenue fell 23.6% to 2.75B CNY; ad income down 32.3%, members down 10%.
- •Integrated DeepSeek-R1 in Feb 2025 for search/creation using 50M docs + 870M Q&A corpus.
- •Plans 2026 AI focus: short/manhua dramas, data services without heavy spending.
- •User base: 73% under 30, 60% female; competing with Tomato Novel, Xiaohongshu.
🧠 Deep Insight
AI-generated analysis for this event — not the original article.
🔑 Enhanced Key Takeaways
- •Zhihu's 2025 revenue decline was primarily attributed to the strategic abandonment of low-margin, high-traffic advertising segments and the deliberate reduction of marketing expenditures to prioritize bottom-line stability.
- •The integration of DeepSeek-R1 has enabled Zhihu to launch an 'AI-Agent' ecosystem, allowing power users to create personalized knowledge bots trained on their specific historical Q&A contributions.
- •Zhihu's pivot into short dramas utilizes its proprietary 'Salt' literature IP library, creating a closed-loop monetization model that bypasses traditional third-party distribution fees.
📊 Competitor Analysis▸ Show
| Feature | Zhihu | Xiaohongshu | Tomato Novel |
|---|---|---|---|
| Core Content | Professional Q&A/Knowledge | Lifestyle/UGC | Web Novels/Short Dramas |
| AI Integration | DeepSeek-R1 (Knowledge) | Proprietary (Visual/Search) | Generative Plotting |
| Monetization | Data Services/Members | E-commerce/Ads | Ads/Subscription |
🛠️ Technical Deep Dive
- •Implementation of DeepSeek-R1 utilizes a RAG (Retrieval-Augmented Generation) architecture specifically optimized for Zhihu's long-form, high-context Q&A data.
- •The system employs a multi-stage filtering pipeline to sanitize the 870M Q&A corpus, removing low-quality or 'troll' content before vectorization for the LLM's context window.
- •Zhihu's AI-driven search utilizes a hybrid retrieval approach, combining traditional keyword-based indexing with semantic vector search to improve the accuracy of complex, multi-faceted queries.
🔮 Future ImplicationsAI analysis grounded in cited sources
⏳ Timeline
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 虎嗅 ↗
This is a summary, not the original. Read the source, or get the weekly briefing.
Weekly AI briefing
One email a week. Unsubscribe anytime.