Good day and thank you for standing by. Welcome to Agora, Inc. Fourth Quarter and Fiscal Year 2025 Financial Results Conference Call. Please be advised that today's conference is being recorded. The company's earnings results press release, earnings presentation, SEC filings and a replay of today's call can be found on its IR website at investor.agora.io. Joining me today are Tony Zhao, Founder, Chairman and CEO; Jingbo Wang, the company's CFO. During this call, the company will make forward-looking statements about its future financial performance and other future events and trends. These statements are only predictions that are based on what the company believes today and actual results may differ materially. These forward-looking statements are subject to risks, uncertainties, assumptions and other factors that could affect the company's financial results and the performance of its business and of which company discussed in detail in its filings with the SEC, including today's earnings press release and the risk factors and other information contained in the final prospectus relating to its initial public offering. Agora, Inc. remains no obligation to update any forward-looking statements the company may take on today's call. With that, let me turn the call over to Tony. Please go ahead.

Thanks, operator, and welcome, everyone, to our earnings call. I'll begin by reviewing our operational performance for the past quarter. We are pleased to report our fifth consecutive quarter of GAAP profitability in Q4, marking our first full year of GAAP profitability since 2018, driven by sustained double-digit revenue growth, improved operating leverage and disciplined cost management. Total revenue for the fourth quarter was $38.2 million, representing 10.7% year-over-year growth. Our GAAP net profit for the quarter was $4.9 million with a GAAP net margin of 12.9%. Next, I would like to share with you our recent business update, which highlights both the strength of our core real-time engagement business and the accelerating momentum of our conversational AI initiatives. Our platform's scalability and reliability were recently validated during a high-profile live streaming event over the Super Bowl weekend. MrBeast, the world's most followed content creator, hosted a broadcast session on Whatnot, the leading video-based shopping platform and a long-standing customer of Agora. We delivered high-quality full HD video to nearly 600,000 peak concurrent viewers worldwide while enabling their interactions at sub-second latency. To quote our customers' own words from their technical blog, 'On event day, Agora's real-time media pipeline performed reliably at peak. Time to first frame stayed under 1 second, latency remained consistently low, and video quality held stable throughout the stream, even as we pushed systems to their limits at extreme load.' We believe this is the largest live video shopping event in U.S. history. Events of this magnitude are the ultimate stress test for real-time infrastructure. Our ability to deliver stable, high-quality video with ultra-low latency at a global scale demonstrates our leadership in network resilience, distributed architecture and real-time routing. This event was powered exclusively by our platform, as no competitor can match our performance and scale. This is why industry leaders in e-commerce, social, entertainment and education continue to trust our infrastructure for their most critical moments. At the same time, we are witnessing rapid adoption of our conversational AI engine product. Since its launch in March 2025, usage has more than doubled each quarter. We are also encouraged to see early experimentation among our customers quickly evolve into real-world deployment across multiple verticals, including customer services, smart devices, education and AI-powered consumer applications. Companionship toys powered by our solution, such as Fuzozo, are driving accelerated shipments with high user stickiness. Validating this momentum, a leading consumer hardware giant recently launched a companionship toy built on our technology. Furthermore, our conversational AI kit, integrating a voice module and an emotion-display screen, has set an industry trend and is now widely adopted by manufacturers. We started the year with a strong reception of our conversational AI solutions for Physical AI at CES 2026 in January. At the event, we introduced the latest upgrade of our conversational AI device kit, featuring enhanced multimodal capabilities, including vision understanding and motion control. These new capabilities enable the development of embodied AI hardware and robotics across multiple use cases. For example, our customer Luwu Dynamics is developing a desktop embodied AI robot powered by this solution. Many of our customers also showcased products at CES that leveraged our solutions, ranging from AI companion devices and robotics to next-generation Physical AI products. The strong market interest and media coverage coming out of CES further validates the growing demand for real-time, human-like interaction embedded directly into smart devices. Beyond one-on-one interaction between humans and AI agents, we are also expanding into multi-agent collaboration scenarios. During the quarter, we supported Agnes AI in launching its next-generation AI group chat and multi-agent collaboration platform. By leveraging our real-time engagement infrastructure and conversational AI capabilities, Agnes AI enables multiple AI agents and human participants to interact seamlessly. We believe multi-agent orchestration represents the next frontier of AI-driven productivity while agents can coordinate tasks, share information and collaborate with humans in real time. Across these developments, a clear theme is emerging. As AI becomes more interactive and multimodal, the technical complexity behind delivering a seamless interaction experience between a human and an AI agent increases significantly. Real-time conversational AI requires not only powerful foundation models, but also advanced audio processing, ultra-low latency networking, global scalability, interruption handling, turn-taking management and device-level optimization. These are areas where we have made substantial investments and have built a strong competitive edge. Our deep expertise in real-time infrastructure uniquely positions us to bridge the gap between AI model capability and production-grade user experiences. Looking ahead, we remain focused on driving revenue growth and advancing conversational AI innovation throughout 2026. We enter the new year with strong momentum, supported by an expanding customer pipeline, growing production deployments and increasing ecosystem partnerships. We believe we are well positioned to capture this transformation and create long-term value for our shareholders. Before I conclude, I would like to thank our customers, developers, partners and shareholders for their continued trust and support and our global teams for their dedication and innovation. With that, let me turn things over to Jingbo, who will review our financial results.

Thank you, Tony. Hello everyone. Let me start by first reviewing financial results for the fourth quarter of 2025 and then I will discuss outlook for the first quarter of 2026. Total revenues for the fourth quarter reached $38.2 million, representing a 10.7% year-over-year increase and exceeding the high end of our guidance. This marks our fourth consecutive quarter of double-digit organic growth. If we look at the two business divisions, Agora revenues reached $19.9 million in Q4, representing 14.4% year-over-year growth and 9.3% quarter-over-quarter growth. The strong growth reflects our successful market penetration and growing adoption in verticals such as live shopping. Shengwang revenues reached RMB 129.2 million in Q4, up 5.7% year-over-year and 5.6% sequentially, driven by continued business expansion and adoption in key verticals such as social and entertainment and IoT. Dollar-Based Net Retention Rate is 109% for Agora and 89% for Shengwang. Gross margin for the quarter was 65.1%, down 1.5 percentage points year-over-year and 0.9 percentage points sequentially. The slight decline was primarily driven by the lower margin profile of our conversational AI-related products, as usage is still ramping and remains at a subscale level. Turning to expenses. R&D expenses were $13.6 million in Q4, down 7.7% year-over-year, reflecting our continued cost discipline. R&D expenses accounted for 35.8% of total revenues compared to 42.9% in the same period last year. Sales and marketing expenses were $7.1 million in Q4, down 2.1% year-over-year. Sales and marketing expenses represented 18.7% of total revenues in the quarter compared to 21.1% in Q4 last year. G&A expenses were $5.4 million in Q4, a decrease of 16.5% year over year, primarily due to lower provisions for credit losses following improved customer collections. G&A expenses represented 14.1% of total revenues compared to 18.7% in Q4 last year. Moving on to the bottom line. We delivered net income of $4.9 million in Q4, representing a 12.9% net income margin. As Tony just mentioned, this marks our fifth consecutive quarter of GAAP profitability and first full year of GAAP profitability since 2018. Based on our current business momentum and visibility into 2026, we expect net income to grow compared to 2025. Now turning to cash flow. Operating cash flow was $9.3 million in Q4 compared to $4.5 million in Q4 last year. Moving onto balance sheet. We ended Q4 with $374.9 million in cash, cash equivalents, bank deposits and financial products issued by banks. Net cash outflow in the quarter was mainly due to share repurchase of $10.9 million. In the fourth quarter, we repurchased 12 million ordinary shares, or 3 million ADSs, representing 3.3% of our outstanding shares at the beginning of the quarter. Since our Board approved the share repurchase program in February 2022, we have repurchased $143.1 million worth of shares through December 31, 2025, which represented 71.6% of our $200 million share repurchase program. We are pleased to announce that our Board has authorized a 12 month extension of our share repurchase program through February 28, 2027, with all other terms unchanged. This reflects the Board's confidence in our long-term growth prospects and our continued commitment to delivering shareholder value. Now turning to guidance. For the first quarter of 2026, we currently expect total revenues to be between $36 million and $37 million, compared to $33.3 million in the first quarter of 2025, representing year-over-year growth rate of 8.1% to 11.1%. This outlook reflects our current and preliminary views on the market and operational conditions, which are subject to change. In closing, I want to extend my sincere gratitude to our exceptional teams in Agora and Shengwang. Our sustained double-digit revenue growth and double-digit net income margin are a direct result of your dedication and execution. Let's remain focused on driving revenue growth and advancing conversational AI innovation throughout 2026. To our shareholders, thank you for your continued trust and partnership. Thank you all for joining today's call. Let's open it up for questions.

First question comes from Daley Li from Bank of America Securities.

Firstly, congrats on the strong Q4 results. And I have two questions here. Firstly, could you update us on the overall RTE demand trend in China and overseas? And what industries are the key demand drivers? Secondly, you have released the ConvoAI Device Kit. And could you please share more color on the conversational AI applications and what industries and applications are the key drivers? And besides I'm not sure, could you share some color about your targeted revenue for the conversational AI this year?

Okay. For the real-time engagement market trend, in China, demand from social entertainment and education customers continues to grow at a modest rate while we remain optimistic on the vast growth potential of IoT and digital transformation customers to drive our China revenue. In recent months, competitive pressure has further abated, and we believe the industry will continue to consolidate. In U.S. and international markets, as I mentioned earlier, our success in one of the massive single-channel live streaming events solidifies our position and brand awareness among live shopping customers, which will bring more business opportunities for us. We are confident that we will gain more market share in this vertical. And for ConvoAI Device Kit, we do expect our conversational AI revenue to continue to grow. The use cases are not just companionship toys, as I mentioned, but also physical AI equipment.

For the revenue, as you know, we released our conversational AI engine in March last year. And since its release, as Tony just talked about, its usage has more than doubled every single quarter. Its revenue contribution is still relatively low at the moment because a lot of customers are in POC stage. So while the revenue growth lags behind usage growth, we do see a healthy pipeline of customers. So based on that, we expect to see revenue contribution from conversational AI ramp up throughout this year. Our goal is for conversational AI to approach 5% of ARR contribution towards the end of this year. Tony, do you want to talk more about the use cases?

Sure. We've been talking about the conversational AI use cases before. It's still focused on customer service, companionship devices, education and interactive. We're now also focusing closely with global customers from U.S., Europe, South America, Asia Pacific region and inside China to implement our solution in a couple of customer service scenarios such as outbound marketing, call center management, appointment scheduling, order confirmation and so on. For companionship devices, a number of device shipments and activations are promising. More importantly, our solution is becoming the de facto industry standard or best practice, and we expect to see more customers launch their products throughout the year, including some based on well-known IP with the potential to become a global hit.

Next, we have Ri Han from CICC.

This is Ri Han from CICC. Can you hear me?

Yes.

Congrats on another solid quarter, especially with revenue coming above the high end of guidance. My first question is on gross margin. We noticed that gross margin declined slightly year-over-year to 65%, just as Jingbo said. Can you walk us through the key factors behind that decline? Should we view this as mix-driven and temporary or more structural given AI ramp-up costs? How should we think about margin trends into 2026? My second question is on profitability for 2026. After achieving full year GAAP profitability in 2025, how do you think about operating income and operating margin next year? What are the main drivers that could support further margin expansion? Yes, that's it.

Sure. So I will talk about gross margin first. As I said, the slight decrease in gross margin was mainly due to the impact of conversational AI-related products because some of the customers are still in the early pilot stage, and we don't charge customers for pilot POC experimentations. So revenue ramp-up lags behind usage growth. Also, the ConvoAI infrastructure is currently running at a very small scale, at subscale levels. So that's why if we only look at the margin of that particular product, it's very, very low at the moment, and that drags down the overall margin slightly. We do expect this to improve as usage and revenue ramp up, but it might take a couple of quarters to fully recover. So when we kind of do our internal forecast and give guidance on 2026 profitability, we are essentially forecasting flat gross margins compared to Q4 2025. So in terms of operating income, we expect operating income to improve significantly compared to 2025. It's driven by revenue growth and improved operating leverage. Our goal is to achieve GAAP operating profit in Q4 2026. Please note, this is after taking into consideration about $6 million of share-based compensation in 2026 and also nearly $4 million of amortization related to the headquarters project. So after these two items, we expect to significantly improve the operating income.

Next question comes from Zongxuan Yang from CITIC Securities.

Also congrats on the last quarter's performance. So I just have one question following the first question from Bank of America regarding AI. So we can see that the stock price for especially for those U.S. software companies has fluctuated recently. So like the market has a lot of concern about AI software. I just want to know how do you think of the infrastructure and the cybersecurity company's position in this AI era? And also maybe like other company's leaders on this issue and the other company's position in AI.

Yes. So SaaS services have strengthened because of the drastic cost reduction in building UIs/UX and application layer logic of software by web coding or AI coding. However, the system level or infrastructure level core services, including the PaaS and API services we provide, are actually facing increasing demand from web coding. The need for even higher quality and scalable API services is much needed than before. It's hard to imagine those hardcore low-level or system-level infrastructure technologies would be easily disrupted by web coding or just AI coding. So as the demand for real-time multimodal interactions with AI engines grows, especially in this sector, we will largely benefit from the global trend of AI development. Plus, who is not an AI company these days? If you're not, you're outdated. We, as a company, were the first to introduce AI technology into the RTE sector even before the generative AI era, and we are the first to launch AIGC RTE SDK. We are the first to demo full duplex conversational AI. We provide the best AI turnkey and AI models in the world, and we are one of the few to launch the real-time API with OpenAI. So we are heavily invested in AI development and AI infrastructure front. We have also positioned the company as a leading innovator in the generative AI era, and we are committed to being one in the coming decades.

Yes. Actually, I want to add like a layman's perspective from a non-technical person. So now if you ask a coding agent, a cloud code, or open cloud to write an app with real-time engagement features like write a meeting app for your own company or your team, it's most likely if you try, you'll see that the agent will call API to build this instead of trying to rebuild the entire real-time communication infrastructure and the fundamental code again. So actually, it will be used by the coding agents rather than be replaced by coding agents.

Our last question comes from Xu Yue from China Securities Co.

Congrats on the solid results. So I have two questions. The first question is regarding the gross profit margin. We see that this quarter, the gross profit margin is kind of dragged down by AI investment. So how do we forecast for future AI product margin trend? And the second question is, how do you view the growth trajectory for the coming quarters for AI toys and customer service? And have we seen the inflection point of adoption in these verticals?

Sure. So again, in terms of gross margin, we actually think the conversational AI product has great margin potential based on our own internal estimate. If we operate at normal levels at a good utilization rate and a decent scale, the gross margin of the AI product should be at least similar, if not higher, than the current core RTE products. So the current relatively low margin is really due to the suboptimal scale and also a lot of POC ongoing. We don't have a fundamental concern on the margin. It will just take some time to ramp up to the target levels. So that's on gross margin. In terms of the AI product adoption, as Tony talked about, right, we do expect the adoption to grow throughout this year. And you talked about the performance and cost, right? I think it's not just us in terms of performance, but for all the players globally in conversational AI, there are a lot of new start-ups focused on this area. I think we face the same problem; the technology itself is fundamentally ready. But from an engineering perspective, there remain a lot of corner cases and use case adoption to be done. This will take time, but it's really just a question of time, not a question of whether it will work or not. We think we made a lot of progress already in 2025. That's why in several use cases like companion products and outbound calling, there are already working use cases, and we'll solve more problems this year. We do think it's not one single turning point, but we will solve use case by use case and gradually penetrate into more verticals. On the cost side, as we all know, costs are coming down steadily on all the models, so we do not think the cost will be a blocking factor.

Thank you. There are no further questions. That concludes today's Q&A session. Thank you, everybody, for attending the company's call today. As a reminder, the recording in the earnings release will be available on the company's website at investor.agora.io. And if there are any other questions, please feel free to email the company. Thank you.

Thank you. Bye-bye.