Executive Summary / Key Takeaways

• The AI Paradox: Zhihu faces an existential threat from generative AI that can instantly answer questions, yet management positions this same technology as its greatest opportunity to amplify its core strengths—trustworthy content and expert networks—creating a binary outcome that will define the investment case.

• Profitability Without Growth: The company achieved a remarkable financial turnaround in 2024, delivering its first non-GAAP operating profit and net income in Q4, but this discipline has come at the cost of shrinking revenue (down 22% YoY in Q3 2025) and declining user metrics, raising questions about long-term sustainability.

• Valuation Disconnect: Trading at $3.44 per share with a market capitalization of $308.54 million against net cash of approximately $643.8 million, Zhihu's enterprise value is negative, signaling the market has priced in terminal decline—making any successful AI pivot potentially highly asymmetric.

• Segment Divergence: While marketing services appear to be bottoming with management forecasting Q4 2025 sequential recovery, paid membership remains in transition with uncertain revenue floors, and the vocational training business has been reclassified as management focuses on efficiency over growth.

• Critical Execution Risk: The success of Zhihu's "information + trust" strategy hinges on whether its AI integration (Zhihu Zhida) can create sufficient differentiation from general-purpose LLMs to retain users and attract advertisers, while competitors like Baidu (BIDU) and short-video platforms siphon away traffic.

Setting the Scene: The Professional Q&A Platform at an AI Inflection Point

Founded in 2010 and headquartered in Beijing, Zhihu Inc. established itself as China's premier online content community built on a simple but powerful premise: real experts sharing in-depth knowledge through validated discussions. This foundation created a unique asset in China's internet landscape—a professional, authentic content ecosystem where quality and trust served as the primary differentiators. For fourteen years, this model attracted urban, educated users seeking substantive answers to complex questions, positioning Zhihu as the go-to destination for professional knowledge sharing.

The business model rests on three pillars. Marketing services leverage Zhihu's brand influence and professional community discussions to connect high-value clients, particularly in technology verticals, with targeted advertising opportunities. Paid membership offers premium content through initiatives like the Yanyan Story long-form writing platform and voice live streaming. Vocational training, recently reclassified into "other revenues," provides creator-driven educational content. This structure historically generated stable revenue from a loyal user base willing to pay for quality.

However, the industry structure has fundamentally shifted. Generative AI has eliminated content creation barriers for short-video platforms like Douyin and made information retrieval faster for search engines, amplifying their user attractiveness. Simultaneously, large language models can now deliver high-quality, structured answers within seconds, making traditional Q&A platforms appear increasingly replaceable. Zhihu's average monthly paid members declined 13.3% year-over-year to 14.3 million by Q3 2025, while the company ceased reporting monthly active users altogether in 2025 after a 21.2% decline in 2024. This traffic erosion strikes at the heart of Zhihu's value proposition.

Against this backdrop, Zhihu's competitive positioning reveals both strengths and vulnerabilities. Bilibili (BILI)'s video-centric model captures younger demographics with visual content, while Weibo (WB)'s real-time microblogging dominates trending topics. Baidu's search-integrated Zhidao platform commands distribution through its search monopoly. Zhihu's text-based, expert-driven approach excels in depth and decision-making utility but lags in scale and speed. The company's 61.3% gross margin in Q3 2025 significantly exceeds Bilibili's 36.7% and approaches Weibo's 77.1%, demonstrating superior content cost efficiency. Yet Zhihu's Q3 revenue of RMB 658.9 million represents a 22% year-over-year decline, contrasting sharply with Bilibili's 5% growth and Baidu's AI segments expanding over 50%. This divergence highlights Zhihu's lag in both scale and technological adaptation.

Technology, Products, and Strategic Differentiation: The "Trust Moat" in the AI Era

Zhihu's response to AI disruption centers on a deliberate strategy to evolve into a dual-pronged medium for "information + trust," leveraging its community to become a "trusted information infrastructure" rather than being replaced by AI. This pivot represents management's core thesis: as knowledge becomes easier to access through AI, trust, expert networks, and genuine engagement will grow scarcer and more valuable.

The technological manifestation of this strategy is Zhihu Zhida, the company's AI-powered feature that has undergone continuous upgrades. In early March 2025, a major enhancement enabled users to trace AI-generated content back to actual community contributors and directly engage with specific experts. By Q3 2025, Zhihu Zhida evolved into an agentic mode designed to deliver more accurate search results while serving as a partner for deep thinking and creativity. The penetration rate exceeded 15% in Q3 2025, nearly four times higher than the same period last year. Critically, AI-generated responses now cite verified knowledge during the reasoning stage, significantly reducing hallucination and improving trust.

This addresses the fundamental weakness of general-purpose LLMs: their inability to attribute information to credible sources. While Baidu's Ernie bot or other AI platforms can generate answers quickly, they lack Zhihu's foundation of expert-validated content. The contributor attribution feature launched in Q1 2025 directly counters this, creating a feedback loop where AI enhances rather than replaces human expertise. As CEO Zhou Yuan stated, "By strengthening the attribution of content to trusted creators across the knowledge base and search, AI-generated responses now cite verified knowledge during the reasoning stage, significantly reducing hallucination and improving trust."

The product ecosystem reinforces this moat. The "Ideas" product saw average daily content volume and interactions increase by 21.7% and 33.1% quarter-over-quarter respectively in Q3 2025. The "Circles" product's average daily views more than tripled sequentially. The Yanyan Story long-form writing marathon generated tens of thousands of submissions, while IP licensing revenue maintained triple-digit year-over-year growth and high double-digit quarter-over-quarter growth. These initiatives create a content flywheel where AI tools lower creation barriers for mid-tier creators while premium content attracts high-value subscribers.

The strategic implications are twofold. First, Zhihu is positioning itself upstream of Chinese LLMs as a trusted data source, with management noting the company is "gaining prominence" as a content provider for AI applications. This could create new revenue streams through data licensing or API access. Second, the AI integration aims to make Zhihu Zhida a primary entry point rather than a secondary feature, with pilot features like cross-topic content aggregation and community trend summaries planned for launch. If successful, this transforms Zhihu from a destination site into an embedded knowledge layer across the AI ecosystem.

However, the differentiation remains fragile. External analyses note that Zhihu Zhida offers limited differentiation from rapidly evolving general AI models, potentially failing to provide a compelling reason for users to choose Zhihu over more convenient alternatives. The initiatives have not yielded significant or sustained growth in traffic or paid users since their launches, suggesting the market remains skeptical of the trust moat's durability.

Financial Performance & Segment Dynamics: Cost Discipline Masking Revenue Decline

Zhihu's financial results in 2024 marked a historic turnaround, with Q4 delivering the company's first non-GAAP operating profit and net income, achieving an 11% net profit margin. This momentum continued into Q1 and Q2 2025, with the company posting its first Q1 non-GAAP profit since IPO and maintaining profitability for three consecutive quarters. Q2 2025 adjusted net income reached RMB 91.3 million, a dramatic swing from a RMB 44.6 million loss in the prior year, while gross margin expanded to 62.5% from 59.6%.

This performance reflects ruthless cost discipline rather than revenue growth. Total operating expenses in Q3 2025 decreased 19.4% year-over-year to RMB 503.5 million, driven by a 36.2% reduction in R&D expenses and 14.9% cut in selling and marketing expenses. The 88% year-over-year reduction in user acquisition costs, initiated in April 2024, exemplifies this strategy of forgoing peripheral users to focus on core engagement. The result is a leaner, more profitable operation—but one that is shrinking.

Segment performance reveals divergent trajectories. Marketing services revenue fell to RMB 189.4 million in Q3 2025 from RMB 256.6 million year-over-year, but management expects sequential recovery starting Q4 2025, targeting each 2026 quarter to exceed the Q3 2025 baseline. The year-over-year decrease has narrowed, indicating the bottoming of its adjustment cycle. Marketing services still represents the largest revenue contributor, and its stabilization is critical for overall financial health.

Paid membership revenue declined to RMB 385.6 million in Q3 2025 from RMB 459.4 million year-over-year, but average monthly paid members increased 8.1% sequentially to 14.3 million. This divergence suggests ARPU pressure as the company retains users but at lower price points or with higher promotional activity. IP licensing revenue's triple-digit growth provides a bright spot, nearly doubling year-to-date compared to the prior year, indicating premium content can command value. However, management acknowledges the segment remains in transition and cannot confirm if revenue has bottomed.

Other revenues, including reclassified vocational training, fell to RMB 83.9 million from RMB 129 million year-over-year. The strategic shift here is profound: vocational training is no longer positioned as a growth driver but as a creator ecosystem enhancer. The "column" product focuses on super creators rather than commercial expansion, with AI tools driving sequential growth in leading creators and user engagement. Monetization models are diversifying, with overall GMV more than doubling quarter-over-quarter, but the segment's contribution to overall revenue remains modest.

The balance sheet provides strategic flexibility. Cash and equivalents totaled RMB 4.6 billion (US$643.8 million) as of September 30, 2025, down from RMB 4.9 billion at year-end 2024 but still substantial. With minimal debt (debt-to-equity ratio of 0.04) and a current ratio of 3.69, Zhihu has the resources to invest in AI integration without immediate liquidity concerns. CFO Wang Han emphasized this strength, stating, "We have a solid—very solid cash position, and we are not reverting to the old model of spending aggressively just forth go."

The implication is clear: Zhihu has achieved financial stability through disciplined cost management, but this stability is predicated on eventual revenue recovery. The company can afford to invest in AI initiatives and weather short-term losses, but cannot sustain profitability indefinitely while revenue continues to contract. The non-GAAP operating loss narrowing 16.3% year-over-year in Q3 2025, despite revenue decline, demonstrates operational leverage, but this leverage only creates value if revenue stabilizes and grows.

Outlook, Management Guidance, and Execution Risk

Management's guidance for 2025 reflects cautious optimism rooted in the belief that AI represents a historic opportunity. CFO Wang Han stated there is a "very high likelihood of achieving our first full-year non-GAAP profitability in 2025," supported by strong cost control and the marketing services segment turnaround. This confidence stems from the company's performance in the first half of 2025, where it achieved profitability on both GAAP and non-GAAP bases in each quarter.

The key near-term catalyst is marketing services revenue beginning sequential growth in Q4 2025. Management expects the segment to recover from its Q3 bottom, with each 2026 quarter staying above the Q3 2025 baseline of RMB 189.4 million. This recovery is predicated on optimizing the client mix toward high-value accounts and enterprise clients, plus upgrading advertising products through deeper AI integration. The new GEM marketing solution, launched in early November 2025, provides core insights such as visibility across AI platforms and citation analytics, directly addressing the shift toward generative engine optimization (GEO) as advertisers seek presence in AI-generated answers.

For paid membership, management is actively experimenting with initiatives to improve retention and ARPU through enhanced content supply, membership benefits, and personalized experiences. AI is expected to enable more efficient content creation and IP development. However, the admission that revenue may not have bottomed introduces execution risk—any decline would come from products or cohorts with lower ROI and weaker profitability, suggesting a strategic pruning that could further reduce near-term revenue.

The vocational training business is no longer considered a drag on the bottom line, having achieved a 90% year-over-year operating profit increase in Q2 2025. The strategic shift toward a more social, knowledge-sharing oriented model leveraging community strengths indicates management's focus on efficiency over scale. This aligns with the broader theme of disciplined resource allocation.

The AI integration roadmap is ambitious. By late November 2025, Zhida will fully augment general AI search capability to include Zhida-generated content for all users. Pilot features such as cross-topic content aggregation and community trend summaries aim to elevate Zhida from secondary to primary entry point. AI assistant writing tools have achieved over 20% adoption, with plans to introduce multi-model content conversion and short-form generation to lower creation barriers.

The critical question is whether these initiatives can outpace the competitive threat. Management's view, articulated by CEO Zhou Yuan, is that "as knowledge becomes easier to access, 'trust, expert networks and genuine engagement' will grow scarcer in the AI era." This conviction underpins the strategy, but external analyses suggest Zhihu's features offer limited differentiation from rapidly evolving general AI models. The initiatives have not yielded significant or sustained growth in traffic or paid users since their launches, creating a timing risk: can Zhihu achieve product-market fit before its user base erodes further?

Risks and Asymmetries: When Cost Discipline Meets Existential Threat

The primary risk is straightforward: generative AI could render Zhihu's core Q&A model obsolete. Large language models deliver high-quality, structured answers within seconds, making traditional knowledge-sharing platforms look increasingly replaceable. As users grow accustomed to querying AI directly, Zhihu's crucial entry-point traffic is being siphoned off. This isn't theoretical—paid members declined 13.3% year-over-year, and the company stopped reporting MAUs after a 21.2% drop in 2024. If this trend accelerates, no amount of cost discipline can preserve profitability.

Competitive dynamics compound this risk. Baidu's AI-powered Zhidao benefits from search distribution and over 50% growth in AI segments. Short-video platforms like Douyin eliminate content creation barriers, capturing user attention that might otherwise seek knowledge. Bilibili's 5% revenue growth and Weibo's massive scale create network effects that Zhihu's niche cannot match. Zhihu's 61.3% gross margin advantage is meaningless if revenue continues declining while competitors grow.

The AI integration strategy itself carries execution risk. While management touts contributor attribution and hallucination reduction, external analyses note that Zhihu Zhida offers limited differentiation from general AI models. The features have not demonstrably reversed user decline. If AI tools simply accelerate content creation but fail to create a defensible moat, Zhihu becomes a commoditized data provider in a race to the bottom.

Financial asymmetry works both ways. The RMB 4.6 billion cash position provides runway for experimentation, but also represents a value trap if the core business continues shrinking. The negative enterprise value suggests the market expects permanent decline. Any sign of revenue stabilization could trigger re-rating, but continued contraction could lead to cash burn despite current profitability.

A key mitigating factor is management's recognition of these risks. The strategic pivot from growth to profitability, the AI integration focus, and the client mix optimization all reflect clear-eyed assessment. CFO Wang Han's comment that "what were one thing as the challenges for Zhihu in traditional Internet era has now become our biggest competitive advantage, and are in the most AI native way" captures this optimism. However, history shows that recognizing disruption doesn't guarantee survival.

Valuation Context: Pricing in Terminal Decline

Trading at $3.44 per share, Zhihu's market capitalization of $308.54 million stands at a stark discount to its balance sheet. With cash and equivalents of approximately $643.8 million as of September 30, 2025, the company trades at 0.48 times book value and 0.74 times sales. The enterprise value of -$289.17 million indicates the market assigns negative value to the operating business, effectively pricing in liquidation or terminal decline.

This valuation compares extremely unfavorably to peers. Bilibili trades at 2.45 times sales, Weibo at 1.41 times, and Baidu at 2.34 times. Even after accounting for Zhihu's smaller scale and declining revenue, the 0.74 P/S multiple suggests deep skepticism about the company's future. The price-to-book ratio of 0.48, against a book value of $7.19 per share, implies the market doubts the carrying value of assets, likely reflecting concerns about intangible value erosion as AI disrupts the core business.

For an unprofitable company with inconsistent earnings, traditional P/E metrics are misleading. The trailing P/E of 22.93 reflects a brief period of profitability rather than sustainable earnings power. More relevant metrics include:

• Cash runway: With minimal cash burn in profitable quarters and RMB 4.6 billion on hand, Zhihu has over a decade of runway at current spending levels, providing strategic flexibility.
• Revenue multiple: At 0.74x sales, any stabilization or recovery would likely drive multiple expansion toward peer averages of 1.5-2.0x, implying 100-170% upside before considering growth.
• Asset value: Net cash exceeds market cap by over 100%, creating a floor that limits downside unless the business begins burning cash.

The valuation context reveals a market that has given up on Zhihu's growth prospects. This creates a highly asymmetric risk/reward profile: if the AI strategy fails, the company can liquidate with limited downside; if it succeeds, the re-rating could be substantial. Management's confidence is evident in continued share repurchases, with the CFO stating, "At this clear undervalued stage, we will continue to conduct open market buybacks at appropriate times."

Conclusion: The Trust Moat's Last Stand

Zhihu stands at a critical inflection point where disciplined financial management meets existential technological disruption. The company has achieved what many loss-making platforms never do—consistent non-GAAP profitability and industry-leading gross margins—yet this accomplishment rings hollow against a backdrop of shrinking revenue and eroding user metrics. The market's verdict is clear: trading below net cash with a negative enterprise value, investors have priced Zhihu as a melting ice cube.

The central thesis hinges on whether Zhihu's "trust moat"—its fourteen-year accumulation of expert-validated content and professional community norms—can survive the AI onslaught. Management's strategy to position Zhihu as "trusted information infrastructure" is logically sound: in an era of AI hallucination and synthetic content, verified human expertise should become more valuable. The Zhihu Zhida integration, contributor attribution, and agentic search capabilities are tangible steps toward realizing this vision.

However, execution risk is paramount. While the AI tools enhance content quality and reduce hallucination, they have not yet demonstrably reversed user decline or attracted new paid members at scale. Competitors with superior distribution—Baidu's search integration, Bilibili's video engagement, Weibo's real-time virality—continue to capture user attention. The window for Zhihu to prove its AI differentiation is narrowing.

The valuation creates a compelling asymmetry. With net cash exceeding market cap by over two-fold, downside is limited unless the business begins burning cash. Yet upside requires more than cost discipline—it demands revenue stabilization and eventual growth. Management's guidance for Q4 marketing services recovery and full-year 2025 profitability provides near-term catalysts, but the longer-term question remains: can Zhihu's trust moat prove defensible, or will it become a footnote in the AI era?

For investors, the decision reduces to two variables: the pace of AI-driven traffic erosion, and Zhihu's ability to monetize its expert network in ways that general-purpose AI cannot replicate. If management executes, the current valuation could prove a generational entry point. If not, Zhihu risks becoming a case study in how even the most trusted communities can be disrupted by algorithms. The next two quarters will likely determine which path the market believes.