Agora, Inc. (API) Earnings Call Transcript
March 3, 2026
Earnings Call Speaker Segments
Thank you, operator. Hello everyone, and welcome to our earnings call. To kick things off, I'll walk through how we performed operationally during the quarter just ended. We're happy to share that Q4 marked our fifth quarter in a row of GAAP profitability, and it also represents our first complete year of GAAP profitability going back to 2018. This was made possible by continued double-digit top-line expansion, better operating leverage, and careful management of costs. Fourth-quarter total revenue came in at $38.2 million, which is 10.7% growth compared with the same period a year earlier. GAAP net profit for the quarter reached $4.9 million, translating to a GAAP net margin of 12.9%. Now I'd like to update you on recent business developments, which show both how solid our core real-time engagement business is and how our conversational AI efforts are gaining speed. A high-profile live streaming event during Super Bowl weekend recently put our platform's scalability and reliability to the test. MrBeast, who has the largest following of any content creator, hosted a forecast session on Whatnot, the top video-based shopping platform and a customer of Agora for a long time. We provided high-quality full HD video to almost 600,000 peak concurrent viewers around the world, while also supporting their interactions at sub-second latency. To use our customer's own words from their technical blog, "On event day, Agora's real-time media pipeline performed reliably at peak. Time to first stream stayed under one second. Latency remained consistently low, and video quality held stable throughout the stream, even as we pushed systems to their limits at extreme load." In our view, this ranks as the biggest live video shopping event ever held in the U.S. Happenings on this scale represent the toughest possible stress test for real-time infrastructure. Being able to serve stable, high-quality video with ultra-low latency on a global scale shows we lead the way in network resilience, distributed architecture, and real-time routing. Only our platform powered this event, since no rival can keep up with our performance and scale. That's the reason industry leaders across e-commerce, social entertainment, and education keep putting their trust in our infrastructure for their most critical moments. Meanwhile, adoption of our Conversational AI Engine product is happening quickly. Usage has more than doubled quarter after quarter since the March 2025 launch. It's also encouraging to see, you know, early experimentation by our customers rapidly turning into real-world deployments spanning several verticals, such as customer service, smart devices, education, and AI-powered consumer applications. Companionship toys running on our solution, like Fuzozo, are seeing faster shipments along with strong user stickiness. As proof of this momentum, a major consumer hardware company recently rolled out a companionship toy that uses our technology. On top of that, our Conversational AI Device Kit, which combines a voice module with an emotion display screen, has established an industry trend and is now broadly adopted by manufacturers. We began the year with a warm response to our conversational AI solution for physical AI at CES 2026 in January. There, we unveiled the newest upgrade to our Conversational AI Device Kit, adding enhanced multimodal capabilities such as vision understanding and motion control. These fresh capabilities make it possible to build embodied AI hardware and robotics for a variety of use cases. Take Luwu Dynamics, one of our customers, for instance — they're building a desktop embodied AI robot that runs on this solution. At CES, a lot of our customers also showed off products built with what we offer, everything from AI companion devices and robotics to next-generation physical AI products. The strong interest from the market and the media attention coming out of CES gives further proof that demand keeps growing for real-time, human-like re-interaction embedded right into smart devices. Beyond one-on-one interaction between people and AI agents, we're also moving into module agent collaboration scenarios. During the quarter, we helped Agnes AI launch its next-generation AI group chat and multi-agent collaboration platform. Using our real-time engagement infrastructure along with conversational AI capabilities, Agnes AI lets multiple AI agents and human participants interact without friction. We think multiple agent observation is the next frontier when it comes to AI-driven productivity. At the same time, an agent can coordinate tasks, share information, and collaborate with humans in real time. With this development, one clear theme is emerging. As AI grows more interactive and multimodal, the technology complexity involved in delivering a seamless interaction experience between a human and an AI agent goes up significantly. Real-time conversational AI calls for not just powerful foundation models, but also advanced audio processing, ultra-low latency networking, global scalability, interruption handling, turn-taking management, and device-level optimization. Those are areas where we've invested substantially and have built a strong competitive edge. Our deep expertise in real-time infrastructure puts us in a unique position to bridge the gap between AI model capabilities and production-grade user experiences. Going forward, we stay focused on driving revenue growth and advancing conversational AI innovation throughout 2026; we head into the new year with strong momentum, backed by an expanding customer pipeline, growing production deployments, and increasing ecosystem partnership; we believe we're well-positioned to capture this transformation and create long-term value for our shareholders. Before I wrap up, I'd like to thank our customers, developers, partners, and shareholders for their continued trust and support. Our global team for their dedication and innovation. With that, let me hand things over to Jingbo, who will reveal our financial results.
Thank you, Tony. Greetings, everyone. First, I'd like to walk through how we performed financially during the fourth quarter of 2025, and after that I'll share what we expect for the first quarter of 2026. Revenue across the company hit $38.2 million in Q4, which is a 10.7% rise compared with the same period a year earlier and above the top end of the guidance we had provided. That gives us a fourth straight quarter of organic growth in the double digits. Looking at our two business divisions, Agora brought in $19.9 million during Q4, which works out to 14.4% growth year-over-year and 9.3% growth quarter-over-quarter. This robust performance shows how well we've penetrated the market and how adoption keeps rising in verticals including live shopping. Shengwang generated RMB 129.2 million in Q4, an increase of 5.7% year-over-year and 5.6% sequentially, fueled by ongoing business expansion and take-up in key verticals including social, entertainment, and IoT. On a dollar basis, our net retention rate stands at 109% for Agora and 89% for Shengwang. Gross margin came in at 65.1% this quarter, which is 1.5 percentage points lower year-over-year and 0.9 percentage points lower sequentially. That modest decrease stemmed mainly from the lower margin profile of our conversational AI-related products, since usage continues to ramp and is still at a sub-scale level. Now for expenses: R&D came to $13.6 million in Q4, a 7.7% decline year-over-year, which shows our ongoing cost discipline. R&D made up 35.8% of total revenues, versus 42.9% during the same period a year ago. Marketing expenses totaled $7.1 million in Q4, down 0.1% year-over-year. Marketing accounted for 18.7% of total revenue this quarter, versus 21.1% in Q4 of last year. G&A expenses were $5.4 million in Q4, which is 16.5% lower year-over-year. That was mostly because provisions for credit losses came down after customer collections improved. G&A made up 14.1% of total revenues, versus 18.7% in Q4 of last year. Now to the bottom line. We posted net income of $4.9 million in Q4, which translates to a 12.9% net income margin. As Tony said earlier, this is our fifth quarter in a row of GAAP profitability and our first full year of GAAP profitability since 2018. Given the momentum we're seeing in the business and our visibility into 2026, we anticipate net income will be higher than in 2025. Now on cash flow: operating cash flow came to $9.3 million in Q4, versus $4.5 million in Q4 of last year. Now to the balance sheet. We finished Q4 with $374.9 million in cash equivalents, bank deposits, and financial products issued by banks. The net cash outflow during the quarter was primarily attributable to $10.9 million of share repurchase; in Q4, we bought back 12 million ordinary shares, or 3 million ADSs, which equals 3.3% of our outstanding shares at the start of the quarter. Since our board approved the share repurchase program in February 2022, we have bought back $143.1 million worth of shares through December 31st, 2025, amounting to 71.6% of our $200 million share repurchase program. We're happy to share that the board has given the go-ahead for a 12-month extension of the share repurchase program, carrying it through February 28th, 2027, and nothing else in the terms changes. That shows how confident the board is about our growth prospects over the long term, and it also reflects our ongoing dedication to returning value to shareholders. Now on to guidance. Looking at the first quarter of 2026, our present expectation is that total revenues will land somewhere between $36 million and $37 million, up against $33.3 million in the first quarter of 2025, which works out to a year-over-year growth rate of 8.1%-11.1%. This outlook is based on our current, preliminary take on market and operational conditions, and those could shift. To wrap up, I'd like to express my heartfelt thanks to our outstanding teams in Shengwang and Agora. The double-digit revenue growth and double-digit net income margin we've sustained are directly attributable to their hard work and execution. Let's stay focused on driving revenue growth and pushing conversational AI innovation forward throughout 2026. And to our shareholders, thank you for your continued trust and partnership. Thanks to everyone for joining the call today. Let's open it up for questions.
Hello, team. I appreciate you answering my question. To start, congratulations on those impressive Q4 results. There are two things I'd like to ask. First, can you bring us up to date on how demand for RTE is trending overall across China and international markets, and which sectors are driving that demand most significantly? Second, now that the ConvoAI Device Kit has launched, might you offer additional detail regarding conversational AI use cases, and which verticals and applications serve as the primary drivers? Also, I'm wondering whether you could provide some insight into the revenue goal you have set for conversational AI during this year. Thanks very much.
All right; when it comes to how the real-time engagement market is moving in China, we see social entertainment and education clients still expanding their demand at a moderate pace, and we stay hopeful about the enormous room for growth among IoT and digital transformation customers, which should fuel our revenue there. Over the past few months, competitive strain has eased even more, and in our view the sector will keep consolidating. Turning to the U.S. and other international markets, as I noted before, our achievement with one of the huge single channel live streaming events... cements where we stand and how well known our brand is with live shopping clients, and that should open up additional business possibilities. We feel certain we'll capture a larger share of that vertical. Regarding the Conversational AI Device Kit, our expectation is that Conversational AI revenue will keep rising. The scenarios involved, you know, aren't limited to companionship toys—as I said, Physical AI equipment is also taking off.
Now regarding revenue. As you're aware, our Conversational AI Engine came out in March of last year. Since launching, like Tony mentioned earlier, usage has grown by over two times each quarter. Right now its revenue share remains fairly small, since many customers are still in the POC phase, which means revenue growth trails behind the growth in usage. We are seeing a solid pipeline of customers. Given that, we anticipate revenue contribution from Conversational AI will build up over the course of this year. We're aiming for Conversational AI to get close to 5% of AR contribution by the end of this year. Tony, do you want to-
Yeah.
Talk more about the use cases?
Yeah, so we've discussed conversational AI use cases in the past. The emphasis remains on customer service, companionship devices, education, and interactive avatars; additionally, we're now working closely with global customers across the U.S., Europe, South America, Asia, the Pacific region, and within China to deploy our solution in a few customer service scenarios — things like outbound marketing, market polling, appointment scheduling, order confirmation, and so forth. On the companionship device side, device shipment and activation numbers look promising. Even more significant, our solution is turning into the de facto industry standard, or best practice. We anticipate more customers rolling out their products over the course of the year, among them some built on well-known IP that could potentially become a global hit.
I appreciate it, Jingbo Wang and Tony Zhao. Thanks so much.
Thanks to you.
hi, this is Han Tao from CICC. Can you hear me?
Yep.
Okay.
Yes, ma'am.
Appreciate you taking my question. Alright. Nice work on yet another strong quarter, particularly since sales landed above the top of our guidance range. First, I'd like to ask about gross margin. As Jingbo mentioned, we saw gross margin slip a bit compared with the same period last year, landing at 65%. Could you detail the primary drivers of that decrease? Is it fair to treat this as temporary and tied to mix, or is it something more fundamental, considering the costs associated with AI ramp-up? What's the right way to frame margin trajectory as we head into 2026? Second, I want to ask about profitability for 2026. Having reached four years of GAAP profitability in 2025, what's your outlook on operating income and operating margin for next year? Which main factors might enable additional margin expansion? Yeah, that's all. Thank you.
Okay, let me start with gross margin. Like I mentioned, the modest dip in gross margin came primarily from the effect of Conversational AI-related products, since a number of those customers remain at an early pilot stage, and we don't bill customers for pilot POC experimentations. Revenue ramp-ups trail behind growth in usage. On top of that, the Conversational AI infrastructure is right now operating on a very small scale, at sub-scale levels. So if we look just at that specific product's margin, it's extremely low at present, and that pulls the overall margin down a bit. We do anticipate improvement as usage and revenue ramp up, though it could take a couple of quarters to fully recover. When we guide our do our internal forecast and provide guidance on 2026 profitability, we're basically projecting flat gross margins versus Q4 2025. On operating income, we expect it to improve significantly, to further improve significantly compared with 2025. Again, that's driven by revenue growth, better operating leverage, and our goal is to reach GAAP operating profit in Q4 2026. Please note this comes after accounting for about $6 million of share-based compensation in 2026 and also nearly $4 million of amortization tied to the headquarters project. After those two items, we expect to significantly improve operating income.
All right, that's quite useful. Congratulations once more on a solid year. Best wishes for what's ahead.
Thanks.
Yes. I appreciate you fielding my question, and congratulations on how the previous quarter went. I've got just one thing to ask, building on that earlier question from Bank of America about AI. We're seeing share prices, particularly among U.S. software firms, swing around lately. It's like the market carries a great deal of worry when it comes to AI software. What I'd like to understand is your take on where infrastructure providers and security companies stand in this AI period. And perhaps also how leaders at other firms view this matter, plus where those other businesses sit within the AI landscape. Thank you.
Right. SaaS offerings got stronger owing to the sharp drop in expense for constructing UI/UX and a prediction layer logical software through web coding or AI coding. Yet at the system level, or the infrastructure level, core services — which include the PaaS and API services that we offer — are in fact seeing rising demand coming from web coding. The requirement for API services of still greater quality that scale well is, you know, far more necessary than it used to be. It's tough to picture that hardcore low level, or system level, infrastructure, you know, technology being so readily, you know, upended by web coding or, you know, simply AI coding. With demand, you know, for real-time multimodal interactions involving AI engines expanding, particularly in this sector, we stand to gain substantially from the worldwide direction of AI advancement. You know, who isn't an AI company nowadays? If you aren't one, you're behind the times. Our company was the earliest to bring AI technology into the RT sector, doing so even ahead of the generative AI period. We were first to roll out AIGC RT SDK. First to demonstrate full-duplex conversational AI; we deliver the finest AI turn-taking and AI waiting models anywhere in the world. We're among the handful that launched the real-time API alongside OpenAI. We've put a great deal into AI development and the AI infra side; we've likewise established the company as a top innovator during the generative AI era, and we're dedicated to, you know, remaining one across the decades ahead.
Right, so let me put this in plain terms, from someone who isn't an engineer. Say you tell a coding agent — Claude Code or OpenCode — to build an app with live engagement capabilities, something like a meeting app for your own team or company. Odds are, if you give it a shot, you'll watch that agent reach for an API to put it together, rather than attempting to reconstruct all of the real-time communication infrastructure and that core code from scratch. The truth is, coding agents are going to consume it, not take its place.
All right, understood. That's all I had to ask. Thanks. I'll rejoin the line.
Thanks.
All right. Thank you. I appreciate the management team for fielding my question, and congratulations on the strong results. There are two things I'd like to ask. My first one concerns gross profit margin. This quarter, we can see that the gross margin has been somewhat pulled lower because of AI investment. What's your projection for how AI product margins will trend going forward? My second one is about the growth path you foresee for AI, for AI toys in the home quarters, and for customer service. In those verticals, has the inflection point for adoption arrived yet?
Sure. So, on gross margin once more, our honest view is that the Conversational AI product carries excellent margin upside, according to our own internal estimate. Operating at normal levels, with a good utilization rate and decent scale, the AI product's gross margin ought to be at minimum comparable to, and possibly above, what the current core RTE products deliver. What explains the roughly low margin right now is really suboptimal scale plus a great deal of POC still ongoing. We hold no fundamental worry about margin. Ramping up to those product levels simply requires some time. That covers gross margin; now, on AI product adoption, as Tony mentioned, correct? Our expectation is that adoption will keep growing across this year. You brought up performance and costs, right? I'd say this applies not only to us and our internal performance, but to every player worldwide in conversational AI — plenty of new startups are concentrating on this space. When we confront a problem — the same problem — the technology as such is essentially ready. From an engineering standpoint, plenty of corner cases and use case adaptation work is still left to be done; that takes time, yet it is purely a matter of time, not a matter of whether it will succeed. We believe we've already made substantial progress in 2025. That's the reason that, in a number of use cases such as companionship clients, and outbound calling, in a number of use cases it's already functioning, and we'll resolve additional problems this year; our view is that this isn't a single turning point, but rather we'll resolve things use case by use case and step by step penetrate additional verticals. Regarding cost, as everyone knows, right, cost keeps declining steadily, across all the models, so our view is that cost won't act as a blocking factor.
Thank you. Bye-bye.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Agora, Inc. transcript - plus 131,000+ transcripts from 7,000+ companies, speaker segments and full-text search - through the EarningsAPI REST API or hosted MCP server.
Get an API key View API docs →For developers and AI pipelines
Programmatic access to Agora, Inc. earnings transcripts and 131,000+ others is available through the
EarningsAPI REST API and the hosted MCP server.
Quarterly plans from $162 - full transcripts, speaker segments, full-text search,
and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.