Home / Transcripts / Agora, Inc. (API) · November 17, 2026

Agora, Inc. (API) Earnings Call Transcript

November 17, 2026

NASDAQ US Information Technology Software earnings 32 min

Earnings Call Speaker Segments

Bin Zhao executive
#1

Thank you, Operator, and good day to everyone joining our earnings call. I'd like to start by walking through the operating results we delivered this quarter. I'm happy to share that we achieved our third quarter in a row of revenue growth that is accelerating, along with our eighth straight quarter of GAAP profitability. For the third quarter of 2026, total revenues came in at $42.1 million, which works out to 19% growth compared with a year earlier. Our dollar-based net retention rate stood at 110%, up from 97% during the same quarter last year. Both revenue growth and net retention improved again versus Q2. That performance speaks to the strength of our core business in human-to-human real-time engagement, and also to a growing revenue contribution from customers who are building human-to-AI voice applications; now let me move on to the update on our business, product and technology for the quarter. During last quarter, we talked about conversational AI moving from proof of concept into real-world production; in this quarter, we observed even stronger evidence of that shift through actual deployments by customers. In areas such as outbound marketing, market research, customer service and other enterprise workflows, customers no longer ask whether an AI agent is capable of completing a conversation. Rather, they increasingly ask whether it can deliver a reliable business outcome at scale. The bar for these deployments is high. In many enterprise use cases, AI agents get evaluated side by side with human employees; they need to grasp customer intent, respond in a natural way, follow the required workflows and perform consistently. Low cost by itself won't persuade customers to swap out human representatives. Adoption only occurs once the quality of AI interaction is comparable to, or better than, that of human interactions. Through our ongoing work to push our technology forward, which I'll cover shortly, we have reached human level performance across a growing range of use cases. As of today, our conversational AI products are deployed in more than a dozen use cases, handling over 1 million agent calls each day. Looking at the wider real-time engagement industry, I believe 2026 represents an important turning point. Earlier this year, OpenAI published a technical report on leveraging WebRTC to enable high concurrency, low-latency interactions between human and AI. Around that same time, several other leading AI companies likewise adopted WebRTC in their products. This step has helped cement WebRTC as a key technology for real-time human-to-AI interactions. Consequently, Google search interest in WebRTC hit an all-time high, which has brought renewed attention and momentum to the broader real-time engagement industry. For Agora, this amounts to a significant opportunity. We have invested more than a decade in building real-time engagement infrastructure, including our global network and software capabilities that support large-scale full duplex interactions; we are now taking this infrastructure beyond human-to-human engagement so it can support the growing demand for human-to-AI interaction. In fact, we have been at the leading edge of conversational AI evolution from the start, from early cascaded architecture to end-to-end voice models and now to more advanced systems that feature full duplex interaction along with both fast and slow reasoning; we have collaborated closely with developers and partners to bring these technology advances into real-world production. During our flagship Intelligent RTE conference last month, we unveiled our newest technology, the Agora Conversational Agent System, which lets AI agents perceive, grasp intent, reason and respond continuously in a full duplex manner. It also brings in capabilities like memory evaluation and self-improvement, plus other tools required to build more capable agents. Inside this system, our global real-time network delivers the low latency and reliability that human-to-AI interaction demands. Our conversational AI engine handles coordination of the underlying models, audio processing, interruption handling and conversation logic. Our orchestration and harness tool lets customers integrate memory and domain-specific workflows while they keep enhancing agent performance through feedback. Combined, these components offer a complete technology stack that helps customers go from a model demo to a production-grade AI agent. We also keep lowering the barrier to adoption for developers. Not long ago, we launched Skills, CLI tools and recipes that let developers use AI coding agents to build voice agents. Developer experience has always been a priority for us, and we will keep making our APIs intuitive and easy to use for both human developers and AI coding agents. Looking forward, we remain focused on 3 priorities. First, we will keep expanding and replicating deployment of our conversational AI product across customers, industry and geographies, with a focus on use cases that deliver clear business outcomes and attractive unit economics. Second, we will keep improving the ease of use and self-serve capabilities of our products while further expanding developer awareness and adoption. Third, we will keep investing in our infrastructure, including deploying real-time inference nodes in key markets while maintaining financial discipline. The broader industry shift is clear. Real-time engagement is expanding past human-to-human to include human-to-AI interactions. As conversational AI evolves from basic turn-by-turn exchange toward richer contextual understanding, more complex collaboration and increasingly personalized experience, reliable real-time infrastructure will grow even more important. We believe Agora is uniquely positioned at the intersection of conversational AI, real-time engagement and global infrastructure to help customers capture this opportunity. Before I conclude, I want to thank our customers, developers, partners and shareholders for their continued trust and support, and our global Agora and Shengwang teams for their dedication and innovation; with that, let me hand it over to Jingbo, who will review our financial results.

Jingbo Wang executive
#2

Thank you, Tony, and hello everyone. I'll start by going over the financial results we reported for the third quarter of 2026, and after that I'll discuss our outlook for the fourth quarter. In the third quarter, total revenues reached $42.1 million, which was above the high end of the guidance range we had given and represented 19% growth versus a year earlier. The year-over-year rise in our revenue continued to accelerate when compared with both the first quarter and the second quarter. That growth was driven by strong performance in our real-time engagement services, particularly in verticals such as live shopping and social live streaming, along with a growing revenue contribution from our conversational AI products. Our dollar-based net retention rate climbed to 110%, compared with 97% in the third quarter of 2025. Gross profit for the quarter was $26.7 million, up 14.4% year-over-year. Gross margin was 63.4%, versus 66% in the same period a year earlier and 63.7% in this year's second quarter. As we've mentioned on prior earnings calls, the year-over-year decline in gross margin was primarily due to a shift in product mix. During the quarter, usage of our conversational AI products continued to rise, since these products have not yet reached sufficient scale and their gross margin remains at suboptimal levels. We believe that as we continue expanding our customer base and optimizing our real-time inference infrastructure, gross margins on these products can improve gradually. Gross margin was broadly stable on a sequential basis. Now on to expenses. R&D expenses in Q3 were $15.6 million, an increase of 14.6% year-over-year. R&D represented 37.6% of total revenues in the quarter, versus 39.1% in the same period last year. As we said on previous earnings calls, the increase in R&D expenses was mainly due to our ongoing investment in conversational AI products. Sales and marketing expenses in Q3 were $6.7 million, up 3.2% year-over-year. Sales and marketing accounted for 15.9% of total revenues in the quarter, versus 18.3% in the same period last year. General and administrative expenses in Q3 were $5.8 million, up 15.9% year-over-year. G&A was 13.8% of total revenues in the quarter, versus 14.2% in the same period last year. That increase was driven primarily by higher share-based compensation and a larger provision for current expected credit losses as the business grew. Now on to operating results. We recorded a GAAP operating loss of $1.5 million in the third quarter, compared with $1.6 million in the same period last year. On the bottom line, we generated GAAP net income of $3 million in Q3, up 10.4% year-over-year, which translates to a net income margin of 7.2%. Now on to cash flow. Operating cash flow in Q3 was $2.2 million, compared with $0.7 million in the same period last year. Now on to the balance sheet. We ended Q3 with $353 million in cash, cash equivalents, bank deposits and financial products issued by banks, which is up $1.3 million from the close of the second quarter. That rise was driven primarily by positive operating cash flow, with share repurchases partially offsetting it. During the quarter, we repurchased approximately 0.5 million ADS for around $1.9 million. As of September 30, 2026, under the current share repurchase program, we have bought back about 45.1 million ADS for approximately $161.8 million. That program concludes in February 2027. In addition, as noted on the prior earnings call and publicly disclosed, our CEO Tony has been executing his share purchase plan since September. Turning to guidance: based on the information available today, we expect fourth-quarter 2026 total revenues of $44 million to $45 million, representing year-over-year growth of 15.3% to 17.9%; this outlook reflects our current preliminary assessment of market and operational conditions, which may change. We also held $363 million in cash equivalents, bank deposits, and financial products issued by banks. In closing, we are pleased with the business's continued momentum. Revenue growth accelerated further in the third quarter, and profitability remained healthy. We will continue investing in conversational AI with discipline to drive long-term value for shareholders. Thank you all for joining today. We are ready to take questions.

Harry Zhuang analyst
#3

Congrats once more on the robust revenue growth in this quarter and the solid guidance for the coming quarter. I've got 3 questions. First, on the demand trend: could management elaborate on demand in the domestic market as well as overseas? And which key verticals are driving the growth? Second, about the conversational AI products: what revenue contribution should we expect in Q4 of this year? Management previously said it would be roughly 5% of total revenue. And how does that scale up next year? Also, on gross profit margin for the generative conversational AI business, how will it evolve over the coming quarters? And finally, operating profit: we see a somewhat higher OpEx base this quarter, and I'd like to know the timeline for breaking even on operating profit.

Jingbo Wang executive
#4

Okay. On the first question about demand, I'll cover the RTE side, and Tony can speak briefly on the AI side. The overall trend in China, the U.S. and the international market is actually largely unchanged versus prior quarters. In China, demand from social, entertainment, education — these more Internet verticals — customers keep recovering and stay stable. Beyond that, IoT, Internet of Things, smart devices, has been the bright spot over the past few years and has grown very fast in each of the past few years. That's China. In the U.S. and international market, revenue has grown faster than China, driven by use cases like live shopping, financial services, gaming, social live streaming, and the pipeline remains strong; so overall, RTE demand looks quite healthy. Tony?

Bin Zhao executive
#5

When it comes to AI, customer demand keeps rising across a wide range of use cases, from call centers — where clients actually serve plenty of different verticals and industries — all the way to companionship; right now we are powering more than 1 million agent conversations every day.

Jingbo Wang executive
#6

Right, so on the conversational AI side of things, as Tony mentioned during his opening remarks, we've seen a great deal of progress over the last few quarters, and revenue has grown pretty quickly since the start of this year. At this point we remain quite confident about driving that revenue contribution up toward 5% in Q4 of this year. Looking ahead to next year, there's definitely plenty of room to grow, though there's also considerable uncertainty around how much the technology itself will improve, along with customer adoption and our own market expansion. So we don't have clear guidance on the specific number right now; but one thing is certain — next year's figure will be significantly higher than what we deliver in Q4 this year, given the enormous market opportunity and given our customer pipeline as well. Tony?

Bin Zhao executive
#7

Sure. On the technology side, I'd say this remains a space that evolves quickly. If you look at everything happening around conversational AI, or the work on multiple multimodal AI models just this year from Thinking Machines Lab, and the interaction model moving toward the many releases of full duplex conversational models, there is a lot of improvement and development across the industry. Even so, each of them shows different strengths, and none is perfect yet. We expect the technology to keep advancing rapidly on every front. Our own emphasis is more on practical use cases and on getting real, practical conversational use cases to the point where they are mature enough for production.

Jingbo Wang executive
#8

Yes — regarding gross margin, as I mentioned earlier, the conversational AI products haven't yet hit an optimized utilization rate, and the technology deployment infrastructure itself isn't fully optimized either. That's what explains the gross margin being suboptimal at present. The good news, though, is that we're already putting a great deal of optimization in place. And as we keep growing revenue, we can already see a clear trend of gross margin improving. Reaching a target level might take a few quarters, but over the long run we do expect the conversational AI product to produce a gross margin similar to, if not higher than, the RTE products, because first, the pricing is higher. If you check the official pricing on our website, it's 1 to 2 order of magnitude higher than the RTE products. So the price is higher, and the value we create for customers is greater too — beyond just delivering the data, it also comes with intelligence and actual business outcomes. I believe your third question is about operating profit. From a strategic perspective, we believe it's important at this moment to make sufficient investment into the conversational AI business. For example, we need more forward deployed engineers who can work closely with customers to accelerate the transition from proof of concept to production grade deployment and also improve performance, and we also need to invest in R&D so that our in-house models or agentic systems are more competitive. That's why we need to keep investing. That said, we will invest with financial discipline in mind, so that we remain GAAP profitable on a net income basis. But in terms of operating income, right now we are looking at the second half of next year. I think that will be a reasonable target for achieving operating profit on a GAAP basis.

Analyst analyst
#9

Nice work on the strong results this quarter. My question also concerns the AI business; could you share some specifics on the current ARPU for our key customers in that AI business? And how does management expect ARPU to rise going forward? That's what I wanted to ask.

Jingbo Wang executive
#10

Much like our existing RTE business, we have 2 segments. One is the self-serve developer, and the other is enterprise customer. Right now, our large enterprise customers come from industries including automobile, health care, financial services, telecom, and these are bigger customers. We have a relatively small count of these customers, yet each one contributes a larger size of revenue. On the other side, we also recently launched our call Agent Studio product, which is a self-serve offering and lets any developer, any individual developer, very easily build powerful and very flexible conversational AI experiences; and we are seeing healthy adoption among these small developers as well. There, the number is bigger, but each customer will contribute only a small amount of revenue; so given the mix of the 2 types and also given we are still at a very early stage, the overall number of customers is quite small, but growing rapidly. We don't think we have a stable ARPU to show right now, right? ARPU will fluctuate quite significantly given the small base. So I think it's probably too early to answer that question. Maybe in a few quarters' time, we'll have a more stable pattern, and then we can talk about ARPU and how we plan to improve it.

Zongxuan Yang analyst
#11

Well done on how the business performed last quarter. I've got just one question about our AI call center business. What I'd like to understand is how much visibility this business has going forward, and what does the call center market structure look like? And, if so — which market opportunities do we think can be captured? And where does the current penetration rate stand?

Bin Zhao executive
#12

As we've mentioned in earlier quarters, the global call center industry is genuinely huge, with enormous potential. What we're seeing now is plenty of different efforts, naturally, from various players in the market aiming to grow across different use cases. From our perspective, all of those efforts remain at a fairly early stage, and the penetration rate overall is still a small percentage. The use cases being handled, particularly where the conversion rate beats a human being, are still relatively easy ones. Most companies, ours included, are still concentrating on steadily enhancing how well the AI agent performs, so that it can reliably outperform human reps across one use case after another. On the overall structure, at a high level, we can speak of call-out service versus call-in service. And that spans different industries, each with its own data and its own particular context. That data and context represent the most important asset for truly penetrating each of those industries. Continuing to build those data sets and that context is part of why use cases take time to mature. I hope that's helpful.

Yue Xu analyst
#13

I have a question about the AI business. Going forward, will that AI business focus mainly on the domestic market, or on overseas markets?

Bin Zhao executive
#14

You mean the AI market, correct?

Jingbo Wang executive
#15

Yeah. So at this point, it's a mix of both, correct? We've got customers in China as well as in the U.S. and other international markets. Still, we do think that over the long run the opportunity is likely larger in markets outside China, since the population there is bigger and, more broadly, acceptance of software products tends to be stronger.

Bin Zhao executive
#16

This also takes in a larger enterprise market in certain developed nations. And naturally, there are particular kinds of demand-supply problems, particularly for call center providers in high income markets.

Jingbo Wang executive
#17

Thank you. Bye-bye.

Bin Zhao executive
#18

Thank you.

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.