ZH

Ladies and gentlemen, thank you for standing by, and welcome to the Zhihu Inc. Fourth Quarter and Full Year 2025 Financial Results Conference Call. [Operator Instructions] Today's conference is being recorded and webcasted. At this time, I would like to turn the conference over to you, Yolanda Liu, Head of IR and Capital Markets. Please go ahead, madam.

Thank you, Hadi. Hello, everyone. Welcome to Zhihu's 2025 Fourth Quarter and Full Year Financial Results Conference Call. Joining me today on the call from senior management team are Mr. Zhou Yuan, Founder, Chairman and Chief Executive Officer; and Mr. Wang Han, Chief Financial Officer. Before we begin, I'd like to remind you that today's discussion will include forward-looking statements made under the safe harbor provisions of the U.S. Private Securities Litigation Reform Act of 1995. These statements involve inherent risks and uncertainties. As such, actual results may be materially different from the views expressed today. Further information regarding these and other risks and uncertainties is included in our public filings with the U.S. Securities and Exchange Commission and the Hong Kong Stock Exchange. The company does not assume any obligation to update any forward-looking statements, except as required under applicable law. Additionally, the discussion today will include both GAAP and non-GAAP financial results for comparison purpose only. For a reconciliation of these non-GAAP measures to the most directly comparable GAAP measures, please refer to our earnings release issued earlier today. In addition, a webcast replay of this conference call will be available on our IR website at ir.zhihu.com. Today, Victor Zhou, an AI agent, representing Mr. Zhou Yuan, will deliver prepared remarks in English on his behalf. As Victor is still being refined, we appreciate your understanding. Victor, please go ahead.

Thank you, Yolanda. Hello, everyone, and thank you for joining Zhihu's fourth quarter and full year 2025 earnings call. I am Victor Zhou, and I'm pleased to deliver today's opening remarks on behalf of Mr. Zhou Yuan, Founder, Chairman and CEO. In 2025, we achieved our first ever full year non-GAAP profit. This historic milestone validates our strategic transformation and underscores the structural durability of our operational leverage. Full year 2025, adjusted net income reached RMB 37.9 million, on a substantial turnaround from the adjusted net loss of RMB 96.3 million in 2024. Our community engagement continues to thrive. In Q4, average daily time spent per user increased to over 41 minutes on the platform. Our ecosystem of trusted creators remains vibrant, consistently delivering authentic and high-quality content across diverse fields. At the same time, we accelerated AI integration within our community. The synergistic evolution of our high-quality content times the expert network times AI capabilities continuously strengthened Zhihu's competitive mode in the AI era. In 2025, we successfully optimized our business structure. With a healthier commercial ecosystem, total revenue trend improved meaningfully in the fourth quarter, driven by a double-digit sequential increase in marketing services. Entering 2026, amid the surging AI adoption, we are leveraging Zhihu's unique advantages to scale AI-driven commercialization, including rapidly building industry-leading export data solutions and deploying AI productivity tools to accelerate IP monetization of our Yan'an Stories franchise. These initiatives will unlock new commercial opportunities for Zhihu. These efforts are anchored by a robust self-sustaining ecosystem. The powerful synergies between high-quality content, our expert network and expanding AI capabilities have created a positive feedback loop, driving heightened community, activity and interaction. In the fourth quarter, our data engagement metrics strengthened significantly. Average daily time spent per user increased sharply both year-over-year and sequentially to over 41 minutes. Substantial year-over-year growth in positive user interactions also drove notable improvements in both short- and long-term new user intention. High-quality content on our platform continues to surge. In Q4, daily creation of high-quality content rose by over 20% year-over-year, contributing to over 31% growth for the full year. Notably, professional AI-related content increased by over 30% year-over-year. As the global AI landscape has shifted from capability races to architectural innovation and system integration, Zhihu remains a leading forum for prominent researchers and frontline engineers to share insights, unpack complex topics and debate key issues. At the vanguard of the AI revolution, our community hosted extensive high-level discussions on key topics such as DeepSeek's Engram architecture, Qwen's new RIF's winning mechanisms, and the continuous iterations of Kimi and Zhipu. The conversation has moved from stronger models to effective system deployment, emphasizing tiered agent architectures and workflow redesign in products like Open Cloud and Cloud Co-Work. The debut of Unitree Robots at the Spring Festival Gala, together with Tesla and the figures progress towards a mass-producing humanoid robots have filled a critical analysis of embodied AI road maps as founders and the employees from leading AI enterprises personally engaged on Zhihu to answer questions and address concerns. Our platform remains attractive space where AI innovations are first explained, validated and responsibly disseminated. We continue to leverage AI to upgrade our community governance and content mechanism. By replacing many operations with algorithm-driven automated workflows, we enhanced community governance ,efficiency and precision. We introduced new metrics for trustworthy contents recognition and promotion ,while integrating user feedback into our evaluation framework. These measures effectively reduce system noise, dynamically suppressing low-quality content and elevating the overall user experience. Professional creators remain the backbone of Zhihu's expert network. In the fourth quarter, daily active high-tier creators grew by double digit year-over-year. A number of verified honored creators rose by nearly 30% as we continue to strengthen incentives for top-tier creators while supporting their efforts to expand industry influence. Our Zhihu 2025 annual review highlighted exceptionally robust high-tier creator engagement. In AI and technology, leading AI companies, including DeepSeek, Moonshot, Tongyi Qianwen, ByteDance Seed, Zhipu and StepFun actively engaged on our platform through their official accounts. Creators with frontline industry and R&D backgrounds consistently shared cutting-edge insights on our platform, contributing to major industry discussions. For the full year, AI-focused creators grew by approximately 16%. In fundamental sciences such as astronomy and chemistry, high-profile creators actively joined our flagship online and offline science programs. Their authoritative content sparked a widespread discussion beyond our community, driving higher search interest for related topics. On the product side, Ideas remains the primary channel for high-frequency knowledge sharing by professional creators, while Circle facilitates engagement around common interest. For the full year, average daily content volume on Ideas grew 73.5%, and the average daily interactions doubled. This momentum persisted in the fourth quarter with double-digit sequential growth across both metrics. We also increased support for mid-tier creators during the quarter, fostering a dynamic growth-oriented ecosystem. Leveraging AI agents, we significantly improved our efficiency in identifying and nurturing talent. In Circles AI-powered proms and standardized tools lowered creation variants and enhanced content distribution. As a result, average daily content creation in Circles surged over 100% sequentially with daily views up 72%. Beyond the AI-driven efficiency gains in content operations, creator support and ecosystem management, Q4 also saw accelerated advances in our foundational AI capabilities enhancing experiences for both creators and the users. In search, creation and consumption, we continue to deepen the integration of AI into the Zhihu community experience. In search, we completed an AI upgrade to our integrated search in December, introducing cross topic content aggregation and hot trend summarization to create a new entry point for high-quality content discovery. We also tailored the answer formats to different query types, which drove a double-digit increase in click-through rates for our AI direct answer cards and meaningfully increased average AI search interactions per user through more multi-turn conversations. In creation, AI is increasingly becoming a practical tool for creators on Zhihu. Since the fourth quarter, we have rolled out features such as content publishing and one-click enhancement powered by intelligent editing, automated formatting and image pairing capabilities. These tools lower the barrier to creation, improve readability and distribution efficiency and help creators turn ideas into shareable content more efficiently. We are also introducing multimodal capabilities such as AI-generated illustrations and image summarization to make long-form content more visually engaging and improve user conversion in the feed. In consumption and circulation, AI is helping Zhihu content transcend traditional community boundaries through external ecosystem partnerships, we are extending our content capabilities into more intelligent assistant scenarios. Within the community users are beginning to use AI in common thrives for fact checking and professional explanation, which supports more authentic interaction and follow-up discussions. Meanwhile, our AI reading panel on PC has improved the efficiency of long-form reading through one-click summarization and terminology explanation and is beginning to generate more valuable interest signals for future recommendation and monetization. Now turning to commercialization. Our efforts to optimize our commercial structure have yielded notable results. With a healthier business ecosystem, total revenue has entered a recovery phase, reaching RMB 643.5 million in the fourth quarter as the pace of sequential decline continued to narrow. This shows a clear top line recovery trajectory. At the same time, we are exploring new scalable AI-powered monetization avenues with an unwavering focus on long-term value and operational excellence. Let's take a closer look at our performance by segment. In the fourth quarter, marketing services revenue reached RMB 234.8 million, up 24% sequentially as our adjustment cycle bottomed out. Disciplined execution in optimizing client mix and upgrading commercial products capitalizes momentum, strengthening our appeal to high-value clients. We elevated the overall client quality, deepened industry penetration and accelerated new customer acquisition. In the fourth quarter, ARPU rose significantly among clients in high-value verticals such as technology and e-commerce. We also reached the new segments in sectors such as automotive and health care. In December, we hosted the Electric Club New Knowledge Technology Conference, which brought together automotive engineers, autonomous driving specialists and leading tech experts from the Zhihu community to explore NEV safety and intelligent upgrades. The event drove 140% year-over-year increase in participating clients enabling industry leaders like BYD, Mitsubishi and Voyah to articulate their technological strength and the safety value through targeted engagement and build trusted content assets. On commercial product upgrades, we leveraged our trusted content and expert network to expand the community-driven monetization and amplify the commercial value of our key IPs such as Zhihu Science Season and Zhihu Reviewers Jewelry. Revenue from IP-related projects increased 21% year-over-year, supported by deeper brand collaborations across our IP portfolio. At the same time, our Idea Plus solution gained strong momentum during the quarter. By offering a lightweight precisely targeted format, Idea Plus extended our native advertising capabilities into short-form content, significantly shortening the path from discovery to purchase, capitalizing on 106% year-over-year increase in daily ideas, interactions. Idea Plus achieved a 62% sequential increase in client numbers and 200% sequential growth in average daily client spend. In 2026, supported by a healthier commercial ecosystem, we aim to drive continued recovery and sustainable long-term growth in marketing services. Next, turning to the business we currently report on the paid membership, which we increasingly see evolving into a broader content and IP operations business. Paid membership remains a revenue contributor of this segment. In the fourth quarter, average monthly paid members reached 12.2 million, generating RMB 333.5 million in revenue. Short-term membership fluctuations aligned with expectations as our structural adjustments prioritize fundamental improvements in service experience and profitability to support a smooth transition during this phase, we are exploring new growth drivers, initiatives to improve member retention and ARPU are yielding results. Q4 average ARPU increased by 1.4% sequentially and overall quarterly renewal rates improved by 2.7 percentage points. Beyond the paid memberships, we are maximizing content IP's value across media adaptations and licensing. IP monetization revenue, which is currently recognized in other revenues grew more than fivefold year-over-year in the fourth quarter and doubled for the full year, underscoring the significant growth potential of this business. The monetization potential of our Yan'an Stories IP continued to translate into tangible results. In December, 2 adapted short dramas Fang and Xia, and The Seventh Year Of Secret Love For My Childhood Friend premiered on Tencent Video, quickly ranking among the platform's top releases. Fang and Xia set all-time popularity record for vertical short dramas on the platform, while Seventh Year Of Secret Love For My Childhood Friend topped the charts and sparked widespread discussion across social media. These results demonstrate our IP's strong adaptation potential and mainstream appeal. During the quarter, we released our short story influence list for the third consecutive year recognizing 62 outstanding works and 20 authors. The selection includes both mature IP already adapted into film and television as well as a pipeline of high-quality titles with strong multi-format development potential. Together, these initiatives highlight our scalable pathway for long-term value creation, cultivating high-quality content, structuring an IP portfolio and extending it across multiple formats to unlock compounding growth. Looking ahead, rapid advances in multimodal AI and the rising industry productivity are expected to further expand monetization opportunities for Yan'an Stories IP creating new growth potential for our content and IP operations business. Building on this, we are exploring a new format for IP development, AI-powered comic dramas and emerging formats driven by demand for lightweight content and improved generative model efficiency. Positioned upstream, Zhihu leverages a dense network of high-quality creators and rich content assets giving us a natural advantage as a stable source of premium IP. Strategically, we will pursue a dual-track approach of IP licensing and in-house incubation. We will also collaborate with platforms and studios to unlock mature IP value, while building in-house AI production capabilities. Turning to other revenues. Beginning in the third quarter to improve profitability, we consolidated our vocational training and the new initiatives into other revenues, which totaled RMB 75.2 million in Q4. We believe 2026 will mark another leap in AI productivity complemented by rapid expansion of real-world applications. Leveraging Zhihu's unique strength we are accelerating exploration of AI-related monetization. We also see growing potential in export data solutions as competition among other ends, increasingly shifts from scale alone to alignment, quality and real-world generalization, high-value, traceable and structured data is now the core driver of model performance. With our long-standing expert network and authentic discussion scenarios, Zhihu is well positioned upstream in the supply of high-quality knowledge and insights and we believe we can be among the earliest platforms in China to systematically define and commercialize high-value data solutions. To support this opportunity, we are developing our export data solution capabilities. At the same time, we are also exploring how to engage experts more deeply in data construction and labeling processing that supports model training and alignment. In summary, achieving full year non-GAAP profitability in 2025 marks a pivotal milestone for Zhihu, validating the resilience of our strategy and the strength of our execution. In 2026, we remain committed to prioritizing disciplined operations, while accelerating AI integration across our community and commercial models. We are sharpening our strategic focus and optimizing resource allocation. In our established businesses, we will continue to prioritize ecosystem health and the user experience, leveraging AI to drive efficiency gains and elevate content quality. At the same time, we are doubling down on AI-driven monetization innovations to cultivate new scalable growth engines. We are confident that 2026 will usher in a new era for high-quality growth for Q4, defined by the further realization of our unique AI capabilities and monetization potential. With that, I will hand the call over to our CFO, Wang Han, whose remarks will be delivered through his AI voice agent. Han, please go ahead.

I will now go over our fourth quarter financials for a complete overview of our results, please refer to our press release issued earlier today. 2025 represents a structural upgrade in Zhihu's financial profile. As Victor noted, we achieved our first full year non-GAAP profitability milestone. Financially, this progress was driven by sustained cost discipline, improved operating leverage and tighter expense control, while maintaining healthy gross margins. For the full year, we recorded non-GAAP net income of RMB 37.9 million, and our non-GAAP operating loss narrowed by 33.6% year-over-year. These results reflect the cumulative impact of our multi-quarter structure optimization and provide a strong foundation to build on as we enter 2026. Now turning to the fourth quarter. Our total revenues for the quarter were RMB 643.5 million compared with RMB 859.2 million in the same period of 2024. The year-over-year decrease continue to reflect our ongoing efforts to optimize revenue mix and focus on sustainable, high-quality growth. Notably, the pace of sequential decline continued to narrow, reinforcing a clear top line recovery trajectory. Our marketing services revenue for the quarter was RMB 234.8 million compared with RMB 315.9 million in the same period of 2024, while the year-over-year decline reflects our proactive refinement of service offerings, the sequential trend was notably positive. Marketing services revenue grew 24% sequentially, marking a clear inflection point in our recovery. This momentum was driven by stronger client quality, deeper industry penetration and the successful ramp-up of new commercial products. Paid membership revenue was RMB 333.5 million compared with RMB 422 million in the same period of 2024. Average monthly subscribing members were 12.2 million. The year-over-year decline in membership was expected and reflects our deliberate prioritization of unit economics over scale. That said, we delivered sequential improvements in both ARPPU and renewal rates during the quarter, which we view as early validation that our retention initiatives are gaining traction. Other revenues were RMB 75.2 million compared with RMB 123.1 million in the same period of 2024. The decrease primarily reflected the strategic refinement of our vocational training business, partially offset by growth of revenues generated from our intellectual property derivatives business. Our gross profit for the quarter was RMB 344.8 million, compared with RMB 540.7 million in the same period of 2024. Gross margin was 53.6% compared with 62.9% in the same period of 2024. The decrease in gross margin was primarily due to our ongoing efforts to broaden and enhance content offerings for all users. Our total operating expenses for the quarter were RMB 608.7 million compared with RMB 528.8 million in the same period of 2024. The increase was primarily due to a onetime non-cash goodwill impairment charge of RMB 126.3 million, which was primarily associated with our prior acquisitions, mainly driven by lower valuations amid the current market conditions. Excluding this item, underlying operating expenses continued to decline year-over-year as we further streamline spending across key areas. Selling and marketing expenses decreased by 13% to RMB 275.2 million from RMB 316.2 million in the same period of 2024, driven by more disciplined marketing spend and lower personnel-related expenses. Research and development expenses decreased 16% to RMB 123.1 million from RMB 146.6 million in the same period of 2024. The decrease was primarily driven by ongoing improvements in our research and development efficiency. General and administrative expenses were RMB 84 million compared with RMB 66 million in the same period of 2024, primarily due to higher share-based compensation expenses. Our GAAP net loss for the quarter was RMB 210.8 million compared with RMB 86.4 million in the same period of 2024. On a non-GAAP basis, adjusted net loss was RMB 39.4 million compared with adjusted net income of RMB 97.1 million in the same period of 2024. As of the 31st of December 2025, we held RMB 4.5 billion in cash and cash equivalents, current and non-current term deposits, restricted cash and short-term investments compared with RMB 4.9 billion as of the 31st of December 2024. As of the 31st of December 2025, we repurchased 31.1 million Class A ordinary shares on the open market for an aggregate value of USD 66.5 million. In addition, throughout 2025, we repurchased a total of 16.6 million Class A ordinary shares through the company's trustee for an aggregate value of USD 23.4 million, representing 6.29% of the total issued ordinary shares. Looking ahead, we will further enhance earnings quality and scalability by prioritizing higher-margin, more capital-efficient revenue streams. We will continue to strengthen our monetization capabilities and explore new AI-powered revenue models, while leveraging Zhihu's core strength, high-quality content, a respected expert network and advanced AI capabilities, coupled with disciplined capital allocation, including share repurchases. These actions will reinforce our financial resilience and support sustainable long-term value creation.

[Operator Instructions] We will take our first question. Your first question comes from the line of Xueqing Zhang from CICC.

[Foreign Language] Thanks management for taking my question. And my question about your financial outlook. So firstly, what's the earnings outlook in 2026 and how to balance the investment with the cash flow and the profitability?

[Foreign Language]

[Interpreted] This is from Zhihu CFO, Wang Han. So first, 2025 demonstrated that Zhihu can achieve profitability. But more importantly, we believe, given our unique assets and positioning Zhihu's opportunity set in the AI era is meaningfully larger than what we current scale reflect. So we are not pursuing a single path of delivering profitability this year, more profitability next year. And then turning to dividends. At our current scale, that would not generate a particular meaningful level of returns for our shareholders. So what we want to do instead is stay focused on the opportunities created by AI and invest behind them. At the same time, this does not mean we will abandon the bottom line discipline that we worked hard to achieve or return to the old model of burning significant cash for growth. We will be disciplined in selecting new initiatives concentrating our investments on areas with visible ROI potential and a strong fit with Zhihu's core strength. In other words, we want to deliver growth in new AI-driven revenue stream. At the same time, to keep the overall bottom line on a healthy and responsive track. Thanks for the question.

Thank you. We will take our next question. Your next question comes from the line of Daisy Chen from Haitong International.

[Foreign Language] I'll translate it myself. As of current stage, what is your strategy in terms of our commercialization? And what are the company's core priorities for 2026?

[Foreign Language]

[Interpreted] So thanks for your question. I will get started with my answers. This is from Zhihu CFO, Wang Han. So in terms of the priority and the strategy in 2026, these are mainly centered on 2 tracks. First of all, in our core community business, we want to continue using AI to improve efficiency and deliver a better product experience for our users and the content creators. At the same time, to maintain stable revenue and a healthier level of operating profitability. In other words, we want to -- our core business to maintain steady, while becoming increasingly AI-enhanced and financially stronger over time.

[Foreign Language]

[Interpreted] At the same time, we want to fully leverage Zhihu's unique assets to develop new AI businesses. As I mentioned earlier, the new initiatives we choose will not be built around aggressive cash burn. We will focus on areas where we can see a path to a healthy cash flow. Right now, we are mainly focused on 2 areas.

[Foreign Language]

[Interpreted] The first is AI-enabled short-form drama and comic adaption. As text to video and image to video models continue to evolve. The production chain is becoming increasingly streamlined. In that process, the scarce asset is high-quality upstream IP, and that is not something that can be acquired overnight, simply by spending heavily. Zhihu's advantage is not only that we have accumulated a large library of high-quality copyrighted content, but also that we have a highly active creator ecosystem that continue to generate new ideas and new IPs. More importantly, AI-generated short drama and the comic style content have already shown that users are willing to pay for this type of AI content. So we believe this is one of the most promising areas where focused investment could generate meaningful and scalable AI revenue for us.

[Foreign Language]

[Interpreted] The second area is data -- AI data services. At a time when many AI applications are still operating with heavy cash burn, there are only a few categories in the ecosystem that can capture structurally attractive economies. One, of course, is represented by companies like NVIDIA. Another on a relatively smaller but still very attractive scale is high-quality data area. In U.S. companies such as Scale AI, Surge AI and McClure has grown rapidly within just a few years by providing high-quality data services to leading LLM developers, while also demonstrating a healthy cash flow characteristics. So with our strong export network and depending on understanding of high-quality model data, we believe Zhihu is well positioned to provide differentiated data solutions for all of these AI developers. At the same time, our community can continuously service new areas of expertise, emerging knowledge, and involving capabilities that LLMs have not yet fully covered. This gives Zhihu a very differentiated advantage in this field. And we believe this is also a business with a clear opportunity to generate positive cash flow.

[Foreign Language]

[Interpreted] So in a word, what we want to deliver is a stable core business that continue to upgrade through AI with improving product capability and a healthy financial profile. Alongside new AI revenue streams that can grow in a disciplined way. The goal is not to pursue growth through excessive spending, but also -- but to build a new AI business with visible monetization potential and a path to positive cash flow. Thank you for the question.

[Operator Instructions] We will take our next question. The question comes from the line of Vicky Wei from Citi.

[Foreign Language] So could management share some data that will help us better understand the impact of AI on the Zhihu community? And additionally, with regard to product upgrades and user experience enhancement in the coming year, what new initiatives does Zhihu have in place?

[Foreign Language]

[Interpreted] Thanks for your question. I will take this question. This is from Zhihu CEO, Zhou Yuan. So first of all, the impact of AI on our community has not been passive. Over the past few quarters, we have been actively driving this accelerating and deeper integration between AI and the Zhihu community with a clear focus on improving such as content consumption, creator experience and so on. Broadly speaking, the positive changes from AI adoption can be seen across 2 groups: our core retained users and our new users.

[Foreign Language]

[Interpreted] Starting with our core retained user, AI is helping users and creators better understand and connect with each other, which further strength the social nature of a real human interaction on our platform. In 4Q, both the coverage and frequency of the positive user interactions on the platform increased year-over-year. We are also seeing users actively call on AI capabilities, aka Zhida. In the comments section for things like fact checking, explaining professional topics and the following training discussions. Importantly, this is happening without disrupting the community atmosphere. Instead, it is helping drive -- is helping drive more interaction and the follow-on discussion among real users. More recently, we launched AI reading panel on PC, with features such as one-click summaries and explanations of professional terms. It has meaningfully improved the reading efficiency of long-form content and significantly enhance the deep reading experience for our core users.

[Foreign Language]

As we mentioned earlier, daily newly added high-quality content in the community grew by over 20% year-over-year in 4Q. But beyond content volume, while we are more -- what we care more about is a positive shift in user and the creators' behaviors. So through like AI capabilities, such as intelligence editing and multi-model associated creation, we're continuing to lower the barrier. So we can see like in the per user's interaction improved significantly in this quarter.

[Foreign Language]

[Interpreter] For new users entering the community, AI is also lowering the barrier to content discovery, joining discussions and participating in interactions. In 4Q '25, the direct MAUs of Zhihu Zhida continued to grow by more than 260% year-over-year, while next month's retention improved by about 83% year-over-year. In February '26, average daily search queries per DAU increased by more than 16% compared with November '25.

[Foreign Language]

[Interpreter] As we shared previously, we mentioned that we completed another upgrade of AI capability within Zhihu's main search to further integrate Zhida with our broader search experience and making it a new entry point for high-quality content for our users. So we see this happened in December. And after this upgrade, search can present more suitable answer formats based on different types of queries. Since launch, user coverage of a AI Zhida cards have increased meaningfully. CTR, click-through rate saw double-digit improved and average AI searches per user also increased noticeably.

[Foreign Language]

Continue to standby, the conference will resume shortly. [Technical Difficulty]

[Interpreter] Okay. I will continue to deliver answers from our CEO, Zhou Yuan. So for the looking forward perspective, our plans are focused on 2 areas. First, we will continue investing in the experience gains we are already seeing from AI, both in terms of enabling more social interaction and efficiency for core return users and new users. So this direction here is already quite clear, and we have been building toward it step-by-step. On top of that, we are preparing to upgrade the Zhida's core capability from AI search towards an agent-based experience. We believe this could bring broader product experience upgrades to users across community. Although there is still an innovation and execution process ahead of us, and we will continue to work through that rollout. So this is from Zhihu CEO, Zhou Yuan. Thanks for your question again.

Thank you. That concludes today's Q&A session. I will now turn the call back to Yolanda for additional or closing remarks.

Thank you once again for joining us today. If you have any further questions, please contact our IR team directly or Christensen Advisory. Thank you. Thank you all.

This concludes today's conference call. Thank you for participating. You may now disconnect. [Statements in English on this transcript were spoken by an interpreter present on the live call.]

Ladies and gentlemen, thank you for standing by, and welcome to the Zhihu Inc. Third Quarter 2025 Financial Results Conference Call. [Operator Instructions] Today's conference is being recorded and webcasted. At this time, I would like to turn the conference over to Yolanda Liu, Director of Investor Relations. Please go ahead, ma'am.

Thank you, operator. Hello, everyone. Welcome to Zhihu's Third Quarter 2025 Financial Results Conference Call. Joining me today on the call from the senior management team are Mr. Zhou Yuan, Founder, Chairman and Chief Executive Officer; and Mr. Wang Han, Chief Financial Officer. Before we begin, I'd like to remind you that today's discussion will include forward-looking statements made under the safe harbor provisions of U.S. Private Securities Litigation Reform Act of 1995. These statements involve inherent risks and uncertainties. As such, actual results may be materially different from the views expressed today. Further information regarding these and other risks and uncertainties is included in our public filings with the U.S. Securities and Exchange Commission and the Hong Kong Stock Exchange. The company does not assume any obligation to update any forward-looking statements, except as required under the applicable law. Additionally, the discussion today will include both GAAP and non-GAAP financial measures for comparison purpose only. For a reconciliation of these non-GAAP measures to the most directly comparable GAAP measures, please refer to our earnings release issued earlier today. In addition, a webcast replay of this conference call will be available on our IR website at ir.zhihu.com. This quarter, Victor Zhou, Zhou Yuan's AI agent will once again deliver the prepared remarks in English on his behalf. Victor is still in training, so we appreciate your patience as he continues to improve. Victor, please go ahead.

Thank you, Yolanda. Hello, everyone, and thank you for joining Zhihu Third Quarter 2025 Earnings Call. I am Victor Zhou, and I am pleased to deliver today's opening remarks on behalf of Mr. Zhou Yuan, Founder, Chairman and CEO of Zhihu. The third quarter marked another meaningful step toward our goal of achieving non-GAAP breakeven on a full year basis. As our structural optimization initiatives continue to take effect, we further refined our service offerings and balanced commercialization with community health. We also maintained disciplined cost control and improved operating efficiency. As a result, our non-GAAP operating loss narrowed by 16.3% year-over-year in the third quarter. At the same time, our community ecosystem continues to strengthen user mix and engagement improved, while MAUs increased modestly from the second quarter. Daily time spent continued to trend higher year-over-year and quarter-over-quarter. Our users and creators remain highly active, supporting improved core user retention and a steady stream of reliable high-quality content on the platform. With our high-quality content, expert network and AI capabilities working greater synergy. We are accelerating our agentic AI upgrades to deliver trusted and differentiated experiences to users, both within and beyond the community. As the AI industry enters a new phase of real-world integration and accelerated deployment, Zhihu as a trusted source of high-quality content and data upstream of Chinese LLMs and AI applications is gaining prominence, creating expanding opportunities for collaboration. With rising high-quality content, a highly active base of professional creators and accelerating AI integration, our community ecosystem radiates vitality. Our competitive moat of trusted content continues to strengthen. In the third quarter, daily creation of high-quality content increased by over 25% year-over-year, with professional AI-focused content up by more than 30% compared to the same period last year. As AI technologies and applications rapidly advance in China, Zhihu remains a go-to platform for frontline engineers and researchers for sharing and lively discussions. AI-focused content covers a range of subjects, including deep technical analysis, innovative product applications, emerging industry trends, personal growth, career development and a growing array of emerging topics driven by rapid AI adoption from the technical debate between MiniMax and Moonshot AI over efficient attention, which sparks heated discussions on Zhihu and highlighted China's diverse approaches to LLM innovation to the in-depth engineering analysis of new models shared by leading companies. Zhihu has become a trusted source for authentic first-hand exchanges. These discussions have made our platform a place where AI innovations are first interpreted, validated and shared. Meanwhile, we continue to strengthen our trustworthy content ecosystem through ongoing improvements to content governance mechanisms and recommendation algorithms. Professional creators are a vital force in our community. In the third quarter, daily active high-tier creators increased significantly on both year-over-year and a sequential basis. The number of verified honored creators also grew by 29% year-over-year. Engagement among AI-focused creators also continues to strengthen. Zhihu now brings together more than 60 million continuous learners and 3.56 million proficient creators in science and AI and 150,000 ecosystem builders. These contributors not only add consistent high-quality input to our AI content ecosystem, but also show significant potential as future service providers for enterprises. Beyond the science and AI, creator activity in humanities and social sciences also remains strong across the platform. In September, we launched the co-benefit co-creation initiative [Foreign Language] in collaboration with leading institutions such as Alibaba Foundation, Tencent Charity Foundation, One Foundation, and Greenpeace alongside the psychologists, medical experts and the writers. This initiative generated a wide range of high-quality content across disability rights, mental health, environmental protection and more joining over 80 million views. We also hosted the 2025 Zhihu Humanities Season, Zhihu Renwenji event, which brought creators together through a blend of online and offline engagement. The campaign attracted nearly 30 influential creators, driving a 7.5% quarter-over-quarter increase in creator activity in the humanities category and generating 5.82 million topic views, reinforcing Zhihu's professional influence and cultural relevance. To better support professional creators, we continue to enhance the content creation and distribution experience. Our ideas product supports knowledge-based expression from high-tier creators and enables more diverse short-form content creation among mid-tier creators. As a result, average daily content volume and interactions increased by 21.7% and 33.1% quarter-over-quarter, respectively. Our Circles product also continues to serve as a focused space for users with shared interests to gather and interact with average daily views more than tripling sequentially during the quarter. We also continue to advance our agentic AI upgrades across the community. From a product perspective, Zhihu Zhida evolved into the agentic mode at the end of September, delivering more accurate and smarter search results. Most notably, Zhihu Zhida now serves as a helpful partner for deep thinking and creativity, capable of understanding user intent, performing multistep reasoning and synthesizing information across research, learning and content creation. Our advancements in agentic AI are also amplifying the value of our creators. By strengthening the attribution of content to trusted creators across the knowledge base and search, AI-generated responses now sites to verify the knowledge during the reasoning stage, significantly reducing hallucination and improving trust. This strengthens creator influence within the generative AI landscape and gives Zhihu a distinct advantage as a trusted content provider in the emerging AI ecosystem. Now moving on to commercialization. In the third quarter, our commercialization continued to recover on a healthier base with total revenues reaching RMB 658.9 million in the third quarter. We also made notable progress in exploring new monetization avenues by leveraging our core strengths. Let's take a closer look at our performance by business unit. In the third quarter, marketing services revenue was RMB 189.4 million. Notably, the year-over-year decrease narrowed, indicating the bottoming out of our adjustment cycle. We expect marketing services revenue to begin growing on a sequential basis in the fourth quarter. During the quarter, we made a solid progress in both optimizing our client mix and upgrading our advertising products. We continue to optimize client mix by deepening our focus on high-value accounts with our brand power and expanding commercial IP, driving strong uptake from enterprise clients, particularly in technology and other high-value verticals. In late September, we hosted the TechClub Conference, bringing together AI experts and some of the most influential tech creators from the Zhihu community to explore the latest developments and future applications of AI. The event showcased the technology's transforming role in everyday life and our unique ability to connect professional content with meaningful brand engagement, further expanding our high-value client base. Through the Zhihu platform, leading companies such as Gree, China Mobile, Huawei and FY Tech further strengthened their brand positioning in technological innovation and product excellence. Backed by the credibility of our brand and strong commercial efficiency created by professional discussions across our community, we made a solid progress in acquiring new clients across diverse sectors such as automotive, consumer and health care. This quarter, we also further upgraded a wide range of our commercial products by integrating AI more deeply across our portfolio. Our dual ecosystem optimization and product efficiency engines drove a significant increase in positive feedback from clients. For example, we launched the upgraded CCS for idea scenarios and introduced the product to more clients. By offering this short content plus precise scenarios format, it bridges authentic experiences and purchase decisions for brands and merchants. At the same time, it makes content consumption and the decision-making for users substantially more efficient. We are also seeing rising demand from clients to improve brand and product presentation in AI-generated answers. Leveraging our trusted content and high citation rate across the Internet, we launched our new GEM marketing solution in early November. This new solution provides core insights such as visibility across AI platforms and citation analytics. Leading technology clients we have worked with include Lenovo, FlightTech, Vivo and Proa. We have received a positive endorsement as we help enhance both their SEO and GEO performance for brands and new products. Looking ahead, with a healthier ecosystem, stronger client base and more robust service offerings, we will continue to leverage AI to drive a steady recovery and long-term growth in our marketing services business. And now for our paid membership business. In the third quarter, average monthly paid members increased by 8.1% sequentially to 14.3 million, with revenue reaching RMB 386 million. Our efforts to boost member retention and ARPU through diversified initiatives continue to generate positive feedback from both creators and users. The Yanyan Story long-form writing marathon came to a successful close in late October after 6 months campaign, generating tens of thousands of submissions in the third quarter alone. This initiative opened up new development pathways for aspiring creators and provided a steady pipeline of content for our library and the future IP development. At the same time, voice live streaming saw a further improvement in paid conversion rates. We also unlocked further commercial potential for our IP adaptations in China and overseas. During the quarter, revenue from IP licensing maintained its triple-digit growth rate year-over-year and generated high double-digit growth quarter-over-quarter. Year-to-date, revenue has nearly doubled compared with the same period last year. In mid-October, Yanyan Story debuted at the Frankfurt Book Fair, showcasing Chinese digital literature on a global stage for the first time. It also draw coverage from the U.K. magazine, the bookseller, which noted the new growth path for Chinese short-form digital literature in the international markets. By the end of October, Yanyan Story licensed more than 100 titles for publication across major Asian markets, including Japan, South Korea, Thailand and Vietnam. A number of works have also been adapted into short dramas for overseas markets and performed well, reflecting the growing popularity of Chinese short-form content abroad. Meanwhile, Yanyan Story has established partnerships with international platforms such as Mobile Reader and GoodNovel to translate works into English, Spanish, Japanese, Korean, Portuguese, Thai, Indonesian and other languages, further expanding its international reach. Going forward, we will pursue a diversified set of initiatives to improve member retention and ARPU. By enhancing content supply, membership benefits and personalized experiences, we aim to strengthen long-term member value. As AI enables more efficient content creation, the potential for IP development and commercialization will expand, unlocking new growth opportunities for our membership business. Starting this quarter, we are simplifying our revenue breakdown and will begin reclassifying vocational training revenue into other revenues to align with our overall strategy. Other revenues were RMB 83.9 million, of which we will continue to adjust our vocational training business with a focus on improving operational efficiency and prioritization. Although our vocational training business has been reclassified, we continue to build on its creator-driven foundation with the development of our column product. Designed primarily to serve super creators, column is intended to enhance the creator ecosystem rather than act as a new commercial growth driver. During the quarter, we enhanced the product by rolling out a PC version and AI tools that help creators generate column descriptions and cover designs. This enhancement drove sequential growth in both the number of leading column creators and creator user engagement. Monetization models for column creators is also becoming more diversified with overall GMV more than doubling compared with last quarter. Going forward, we will continue to operate with discipline, maintaining stability while investing prudently for sustainable growth. With the ongoing enhancements in efficiency and steady cost optimization, we are confident in achieving our full year profitability target. Building on this foundation, we will continue to invest with a long-term view to strengthen our AI capabilities and improve the efficiency of our core operations. Deeper AI integrations will drive greater synergies across content creation, distribution and monetized on Zhihu. Meanwhile, we will further refine our product and marketing strategies to capitalize on new growth opportunities from high-quality users and enterprise clients. With a healthier operating structure and ongoing innovation, we are well positioned to thrive in this next stage of high-quality growth. With that, I will hand the call over to our CFO, Wang Han. Han, please go ahead.

Now I will review the details of our third quarter financials. For a complete overview of our third quarter 2025 results, please refer to our earnings release issued earlier today. In the third quarter, we maintained disciplined cost management and drove further improvements in operational efficiency. As a result, our non-GAAP operating loss narrowed by 16.3% year-over-year. We continue to invest in areas that reinforce our long-term growth potential, striking a healthy balance between efficiency and investment. Our total revenues for the quarter were RMB 658.9 million compared with RMB 845 million in the same period of 2024. The decrease was mainly the result of our continued efforts to optimize revenue mix and focus on sustainable, high-quality growth. Notably, the year-over-year decrease narrowed for the third consecutive quarter, in line with our expectations. Our marketing services revenue for the quarter was RMB 189.4 million compared with RMB 256.6 million in the same period of 2024. This decrease was mainly driven by our proactive refining of service offerings and optimization of client mix. Encouragingly, the year-over-year decrease narrowed meaningfully, indicating that our adjustment cycle has bottomed out. Paid membership revenue was RMB 385.6 million compared with RMB 459.4 million in the same period of 2024. While the number of average monthly subscribing members fell year-over-year, they rebounded and grew 8.1% sequentially to 14.3 million. We also continued to enhance retention and ARPU through diversified content and membership initiatives. Other revenues were RMB 83.9 million compared with RMB 129 million in the same period of 2024. The decrease was primarily due to the strategic refinement of our vocational training business. Our gross profit for the quarter was RMB 403.6 million compared with RMB 540.1 million in the same period of 2024. Gross margin was 61.3% compared with 63.9% in the same period of 2024. Our total operating expenses for the quarter decreased by 19.4% year-over-year to RMB 503.5 million. The decrease was primarily due to a more efficient cost structure and disciplined resource allocation across key operating areas. Selling and marketing expenses decreased by 14.9% to RMB 330.1 million from RMB 388 million in the same period of 2024. The decrease was mainly due to tighter control over promotional spending and optimized personnel-related expenses. Research and development expenses decreased by 36.2% to RMB 114.4 million from RMB 179.3 million in the same period of 2024. The decrease was primarily driven by continued improvement in research and development productivity and efficiency. General and administrative expenses were RMB 59 million compared with RMB 57.2 million in the same period of 2024. Our GAAP net loss for this quarter was RMB 46.7 million compared with RMB 9 million in the same period of 2024. On a non-GAAP basis, our adjusted net loss was RMB 21 million compared with RMB 13.1 million in the same period of 2024. As of the 30th of September 2025, we had cash and cash equivalents, term deposits, restricted cash and short-term investments of RMB 4.6 billion compared with RMB 4.9 billion as of the 31st of December 2024. As of the 30th of September 2025, we repurchased 31.1 million Class A ordinary shares for an aggregate value of USD 66.5 million on the open market. Additionally, we repurchased a total of 22.5 million Class A ordinary shares for an aggregate value of USD 34.5 million through the trustee of the company as of the end of the third quarter. Looking ahead, we are on track to achieve full year breakeven on a non-GAAP basis. We will continue to further strengthen our monetization capabilities and pursue new revenue opportunities that leverage Zhihu's strength in high-quality content creator expertise and AI-driven innovation. Together, these efforts will reinforce our business resilience and support sustainable long-term growth. This concludes my prepared remarks on our financial performance for this quarter. Let's turn the call over to the operator for the Q&A session.

[Operator Instructions] We will now begin with our first question, and this is from Vicky Wei from Citi.

Will management share some color about the AI progress of Zhihu? For example, the penetration rate of Zhihu Zhida and the progress of the AI integration with the Zhihu Community.

[Interpreted] Thank you for question. This is from Zhou Yuan, Zhihu CEO. First of all, I would like to say sorry about my weak voice because I didn't recover yet from my cold. Anyway, I would just start with your first question. So as you can see, Zhida remains one of our key products. Its overall usage and penetration rate continued to increase in the third quarter with the penetration rate existing 15%, nearly 4x higher than the same period last year. This not only reflects the ongoing evolution of our foundational AI capability across the community, but also demonstrates strong user endorsement of a very strategic depending of AI plus community. This also gives us a very strong confidence to continue upgrading this AI plus community experience updates across more touch points. Now let me just share some recent progress and upcoming plans. First of all, in the search scenario, by late November, Zhida will fully augment our general AI search capability to include Zhida-generated content for all users. Additionally, we will soon launch pilot features such as cross-topic content aggregation and the community trend summaries. This will formally navigate Zhida from secondary entry point to a primary one, further boosting AI adoption across the entire community. And second of all, on the content creation side, we are empowering professionals with strong AI copilot. In this quarter, we launched a suite of AI assistant writing tools for our creation assistant, which includes smart headlines, grammar and fact checking and lead paragraph generation. This will help creators optimize long-form structures and expert-level content. So by the end of 3Q, adoption of these new AI features has already surpassed 20%. Looking ahead, we plan to introduce additional capabilities such as AI-powered multi-model content conversion, intelligent formatting and short-form content generation and et cetera. These tools will significantly lower the barrier to entry for mid-tier creators, enabling more users to express themselves effortlessly to increase posting frequency, creation frequency and engage more actively. In addition, on the content consumption and distribution side, we are also expanding Zhida into high-frequency consumption scenarios. For example, AI-powered daily briefing on Zhihu's training topics and other vertical-specific topics as well as the ability to mention Zhida in threats or to also summarize discussions and surface key insights will help users quickly grasp complex conversations and participate more meaningfully. We believe this will further strengthen user engagements and community stickiness. Thank you.

We'll now take the next question. This is from Luqing Zhou from Goldman Sachs.

So my question is on how do you see the current status of Zhihu's user ecosystem? And based on that, could management share more color on the directions for improving Zhihu's future product design and how is the progress so far?

[Interpreted] Thank you for your question. This is from Zhou Yuan, Zhihu CEO. We believe, overall, the community ecosystem is very healthy. We do not rely on any single metric to assess its health. Instead, we focus on content quality, user structure and user quality and whether our content creator incentives are forming a virtuous cycle. We have also deployed AI as a core product driver at a strategic level. Over the past few quarters, we have made the synergistic development of high-quality content, multiply expert network, multiply AI capabilities as a core path for driving our ecosystem in a positive direction. From this perspective, our ecosystem is stable and continuously improving. This is fully in line with our expectations as well. First of all, the trustworthiness and professionalism of our content are very crucial. They are crucial indicators of the ecosystem health. Over the past few quarters, we have continued to strengthen our trustworthy content ecosystem and our expert network while also cracking down on low-quality content and traffic to keep the ecosystem healthy at its core and reinforce the virtuous cycle. As a result, we have delivered several consecutive quarters of double-digit growth in daily high-quality content creation. The AI category is the most reflective of this progress with the professional AI-related content regarding double-digit growth for 4 consecutive quarters. On this basis, users' trust in our content has also continued to increase steadily. And secondly, our user structure and user quality have improved and users' need across different scenarios has been addressed. As we can see from last Q4, our MAU has remained stable on a sequential basis for 4 consecutive quarters. And building on that, average daily user time spent, which we believe as a proxy for engagement and retention has delivered double-digit year-over-year growth for 6 consecutive quarters. Our users remain mainly young and focused on learning and growth with user age between 18 to 30, accounting for more than 65% of our total user base. Among them frontline professionals in technology and AI have become one of the most representative groups. They have long-term professional learning, frontier exploration and interest development needs and contribute more content and provide a stronger positive feedback to the ecosystem. And last but not the least, the content -- the creator ecosystem continues to grow and expand. Output from top-tier professional creators have remained stable over 7 consecutive quarters. At the same time, by using AI tools, we are continuously lowering the creation threshold for mid-tier creators and increasing the creation frequency of the entire creator group. This makes the supply side of the community more diverse and keep social interaction within the community growing. So in summary, ecosystem health is foundational to Zhihu. Going forward, we'll continue to invest in trust content and expert network so that as a community scales, it can maintain its professionalism, vibrancy and trustworthiness. Let me just turn to the second question you mentioned. It's about our core product going forward plan. Here, we hold a few key beliefs. First of all, over the next 3 years, people will consume more AIGC content. At the same time, human-to-human interaction will become more valuable. So we believe both trends will coexist. And the second belief we hold here is that stronger AI becomes, the more people will experience a sense of diminished presence, which means the participation, social capital and relationships enabled by community will become increasingly scarce and increasingly demanded. And the third belief here is that high-quality human-generated content and data will become extremely scarce as well as valuable on the supply end. This supply matters on both ends. It's crucial for the advancement of AI as well as for human development. So going forward, Zhida will definitely integrate with our users' functional social needs. For example, when a user wants to ask a question, search or look for resources, AI will dramatically raise efficiency. And Zhida will push the community further towards utility, enabling even the first day users to get a meaningful experience immediately. At the same time, we will double down on the social needs that come from real human connection things like building recognized, growing together and finding people who share your identity. We want to build these things with user feel like a sense of belonging in an environment grounded in real people, real culture and trusted interactions. So our future product direction is built around 2 pillars: utility and identity. My hope is for Zhihu to become the connection layer for humans in AI era as a place where people can use AI tools to understand the world as well as a community where they can find renaissance and understanding from one another. At the same time, we plan to build a trusted content and expert network as 2 foundational layers of infrastructure. Thank you. Thank you again for your question.

We will take next question. This is from Daisy Chen from Haitong International.

Could management update the progress of the adjustments in each business line? Did you see any signs that the revenue has bottomed out or started to rebound? In particular, how do you expect the future of the advertising business? And also, could you share your outlook on the company's profitability?

[Interpreted] Thank you for your question, Daisy. This is from Wang Han, Zhihu CFO. So I will just pick up your second question. Here's a quick take on our profitability outlook. After delivering solid profits in the first 2 quarters, we now see a very high likelihood of achieving our first full year non-GAAP profitability in 2025. So with that buffer in place, we are using Q3 and Q4 as a window to keep fine-tuning and investing where needed. That's why you will see -- you can see a small loss in Q3, which is well within what we can comfortably take. Let me just walk through the adjustments across our major revenue lines. First, about the marketing services. As we mentioned last quarter, this Q3 is -- it will become the bottom. And we expect a sequential recovery starting in Q4. What we see now give us confidence to maintain that guidance. Looking ahead to next year, our goal is for each quarter to stay above the baseline set by Q3 this year. And second, about the pay membership. This segment is still in a transition period. As we said before, even the best libraries and bookstores separate fiction from nonfiction, the real challenge here is how to differentiate and integrate them in a way that feels natural to users. We will continue experimenting here. So we cannot say membership -- pay membership revenue has hit its bottom yet. But even if there is some decline, it will be about products or cohorts with lower ROI and weaker profitability or less than ideal retention. Search is about vocational training. This business is no longer a drag on our overall bottom line. Given this relatively low base or small scale, we have now reclassified it into others. So overall, you've seen us deliver several consecutive quarters of profitability followed by the small loss in Q3. Even though we remain confident in achieving full year profitability. With that foundation, we are taking this period to make necessary adjustments and targeted investments. As we approach our first full year of profitability, we also want to use this moment to shed some legacy inefficiencies and to start fresh. We have no intention of staying where we are and simply just squeezing out profits. We are now operating from a healthier foundation and getting back onto a trajectory that aligns with Zhihu's long-term development. Also, we have a solid -- very solid cash position, and we are not reverting to the old model of spending aggressively just forth go. And this new AI cycle or in this AI era, our focus is on strengthening Zhihu's position in real people interactions, expert network and trusted content areas. And these capabilities are becoming increasingly important and carry real social value. Thank you for your questions.

We will now take the next question. And this is from [ Jing Yi Wang from Guangfa ].

Could management share some more color about the shareholder return pay in progress.

[Interpreted] Thank you for your question. This is from Wang Han, Zhihu CFO. We can see over the past 2 years, we've been one of the most active buyback companies among U.S.-listed Chinese names. That conviction came from our confidence in reaching profitability. And this year, we expect to demonstrate that our outlook and the targets set 2 years ago are being delivered. Even so Zhihu's current market cap remains significantly below the cash on our balance sheet. So we believe we are super undervalued. Therefore, we intend to maintain our buyback program and expect to remain one of the most active repurchase in this sector. Thank you again for your question.

That concludes today's Q&A session. At this time, I will turn the conference back to Yolanda for any additional or closing remarks.

Thank you once again for joining us today. If you have any further questions, please contact our IR team directly or Christensen Advisory. Thank you so much.

The conference is now concluded. Thank you for attending today's presentation. You may now disconnect. [Portions of this transcript that are marked [Interpreted] were spoken by an interpreter present on the live call.]