Home / Transcripts / Agora, Inc. (API) · August 14, 2026

Agora, Inc. (API) Earnings Call Transcript

August 14, 2026

NASDAQ US Information Technology Software earnings 34 min

Earnings Call Speaker Segments

Bin Zhao executive
#1

Thanks, operator, and hello everybody — good to have you on this earnings call. I'd like to start by walking through how our operations performed during the quarter. It's my pleasure to share that we posted yet another period of quickening revenue expansion, along with our seventh quarter in a row of GAAP profitability. For Q2 of 2026, total revenues came in at $40.4 million, which marks an 18% gain compared with the same period a year earlier. That result mirrors both the ongoing vigor of our main real-time engagement segment and the rising input from our conversational AI offerings as an expanding set of clients shifts from proof of concept into commercial production. GAAP net profit during the quarter stood at $2.2 million, a 50% rise versus the prior-year period, reflecting better operating leverage and careful cost discipline throughout the company. The biggest step forward we made this quarter took place in call centers around the world. Adoption of our voice AI agent is showing real momentum. Built on the finest sales and customer service playbooks, this agent turned in reliably high-quality results in each conversation; it never gets tired, never loses concentration, and never wavers in performance because of workload or the hour of the day; it also stays calm and steady even when calls get difficult. On top of that, clients are now realizing major cost reductions by putting our voice AI agent to work. In fact, we're starting to see it equal or even exceed what humans achieve on a growing number of tasks tied to hitting target business outcomes. Take outbound marketing and capturing buyer interest as the first illustration. Our voice AI agents are now used to place calls, qualify leads, gather information, and book meetings with prospective customers at conversion rates comparable to those of human reps. Two other key dimensions also saw our voice AI agent come out ahead: how many calls it can process and the unit economics it produces. Market surveys are the second illustration. Our voice AI agent is capable of running in-depth interviews for consumer insights and product feedback, all while gathering structured data over the course of each conversation. Because our solution handles such high concurrency, it shrinks what traditionally requires weeks or days for a large-scale survey down to merely a few hours. Marketing and surveying represent just 2 instances of how the voice AI agent can transform call centers globally. We notice comparable openings in financial services outreach, gaming user acquisition and retention, debt collection, and numerous other fields. We're already engaged with clients in these sectors, and we anticipate several of them will shift from proof of concept to large-scale deployment over the next few quarters. Meanwhile, we keep putting resources into our developer ecosystem; during this quarter, we rolled out Agora Skills and Agora CLI. Agora Skills bundles our platform knowledge for AI coding assistants, such as Claude Code, Cursor or Codex, and works with our newest SDKs and best practices when creating real-time engagement or conversational AI applications. The Agora CLI adds to this by providing one command-line interface so coding agents can carry out their work. Put together, these capabilities allow human coders as well as AI programming agents to more easily create and launch real-time engagement apps alongside us. We're likewise persisting in reinforcing our tech ecosystem via strategic alliances; during this quarter, we revealed an alliance with Gradium, a top voice AI platform started not long ago by the research group responsible for WuXi and TBK, a pair of speech models widely recognized within the open source world. Thanks to this collaboration, coders can activate Gradium TTS inside our conversational AI engine via an easy setup, and no extra latency expenses are introduced. Going forward, our first priority stays sharply on speeding up the shift of our conversational AI offerings from pilot stage into production across various use cases. Every use case brings its own distinct hurdles, yet each one also assists us in honing our technology. We think ongoing enhancements to our offerings will open up further demand and push an industry-wide move toward AI-driven workflows in call centers. Second, we'll keep putting money into our real-time infrastructure. Our software-defined real-time network, or SD-RTN, has served as a core edge for us over a long period. As we grow into human to AI interactions, the value of this infrastructure doesn't shrink; quite the opposite, it grows even more essential, since seamless conversations demand ultralow latency inference along with transmission. We feel certain that our spending on real-time inference and the communication infrastructure will prove to be a pivotal element in our capacity to compete and thrive within the conversational AI space; and third, we'll keep fortifying our partner ecosystem and growing developer venture. On October 23 and 24, we'll hold our flagship yearly gathering, IRTE, meaning Intelligent Real-Time Engagement, in Beijing; we can't wait to gather coders, partners, companies and industry leaders to dig into the coming stage of real-time engagement and conversational AI. To sum up, we think the AI industry's center of gravity is progressively moving away from model capabilities and toward the infrastructure and hardness layer needed to run voice AI agents dependably at scale. Sitting where real-time engagement, AI and global infrastructure meet, we think Agora holds a distinct position to assist the enterprise market to assist enterprises in making this shift and generating durable long-term value for our customers as well as our shareholders. Before wrapping up, I want to express gratitude to our customers, developers, partners, and shareholders for their ongoing confidence and backing, and to our worldwide Agora and Shenghong teams for their commitment and creativity. Having said that, I'll pass things to Jingbo, who is going to walk through our financial results.

Jingbo Wang executive
#2

Thanks, Tony. Hi, everyone. I'll begin by going over our financial results for Q2 of 2026, and after that I'll talk about what we expect for Q3. Revenue in total for the second quarter came in at $40.4 million, which is above the top end of our guidance range and works out to 18% growth compared to a year earlier; that makes it 3 quarters in a row where growth has sped up, fueled by ongoing expansion of our real-time engagement services in areas including e-commerce, along with rising adoption by customers of our conversational AI solutions. Our dollar-based net retention rate came to 104% this quarter, versus 94% in Q2 of 2025; that's a solid gain and puts us back over 100%. Gross profit for the quarter stood at $25.7 million, which is a 12.5% rise versus a year ago. Gross margin came in at 63.7%, compared with 66.8% during the same period a year earlier and 63.4% in Q1 of 2026. Compared with a year ago, the drop was mostly attributable to a shift in product mix, since conversational AI products kept seeing more usage during the quarter, though they remain at a subscale stage; looking sequentially, the gain came mostly from technical optimization. Now on to expenses. In Q2, R&D expenses totaled $15.4 million, an increase of 10.2% versus a year earlier. R&D came to 38.1% of total revenue for the quarter, versus 40.8% during the same period a year earlier. The rise was mostly because we kept investing in conversational AI products. In Q2, sales and marketing expenses amounted to $6.4 million, a decline of 1.5% versus a year earlier. Sales and marketing came to 15.9% of total revenue for the quarter, versus 19% during the same period a year earlier. The drop was mostly because of disciplined expense management. In Q2, general and administrative expenses amounted to $5.5 million, a decline of 9.5% versus a year earlier. G&A came to 13.5% of total revenue for the quarter, versus 17.6% during the same period a year earlier. The drop was mostly because of a reduced allowance for current expected credit loss, as customer credit conditions and collection outcomes got better. Now on to operating results. We posted a GAAP operating loss of $1 million in the second quarter, versus a loss of $3.1 million during the same period a year earlier, owing to ongoing gains in operating leverage. Given the momentum in our business right now, we aim to reach quarterly GAAP operating profitability before this year ends. Now on to the bottom line. We generated net income of $2.2 million in Q2, an increase of 50.3% versus a year earlier and equal to a net income margin of 5.4%; now on to cash flow. Operating cash flow came in at negative $2.1 million in Q2, versus negative $0.4 million in Q2 of 2025. Now on to the balance sheet. We closed Q2 with $361.7 million in cash, cash equivalents, bank deposits and financial products issued by banks. The drop in our cash balance was mostly attributable to annual bonus payments, as well as share repurchase during the quarter. During Q2, we bought back roughly 1 million ADS for roughly $3.7 million. Through June 30, 2026, under the share repurchase program currently in place, we've bought back roughly 44.6 million ADS in aggregate, spending about $159.9 million. Our ADS count stood at 83.8 million as of June 30, 2026, down from 87.3 million at the close of 2025. February 2027 is when the existing repurchase program runs out. Moving on to guidance now. Drawing on what we know at present, total revenue for Q3 2026 should land somewhere between $41 million and $42 million, which works out to 15.8% to 18.6% growth versus the same period last year. That forecast captures our preliminary, current thinking on market and operational conditions, and those could shift. To wrap up, we delivered another robust quarter, on both the top line and profitability. Meanwhile, usage momentum and commercial potential in conversational AI keep making us more optimistic, and we'll keep investing prudently to underpin our long-term growth. Thanks, everyone, for being on today's call. We're ready to take your questions.

Harry Zhuang analyst
#3

Nice work on the solid numbers and outlook. Three things from me. First, on demand: what's the pattern looking like for overseas versus domestic markets, and which sectors are mainly fueling expansion? Second, on AI: it's great to see the call center application levers expanding so quickly — so what other data scenarios might generate significant conversational AI demand growth? Also, how much revenue will conversational AI contribute by year-end, and how are gross margins trending? Finally, on competition: can management walk us through the current competitive landscape overseas versus our main rivals?

Jingbo Wang executive
#4

Sure; on demand, I'll cover the RTE part and Tony can address AI; regarding RTE, honestly not much has shifted broadly from what we're seeing this quarter across China, the U.S., and international markets — mostly consistent with the prior quarter. In China, with a steadier operating environment, demand from social entertainment and education customers keeps recovering; I'd say those look fairly stable. For the U.S. and international, demand across large shopping, financial services, and gaming use cases keeps rising. So on the RTE side, demand appears healthy.

Bin Zhao executive
#5

Right. Now regarding artificial intelligence, I believe it's reasonable to state that our vision has received confirmation and been strengthened through what we do every day, thanks to swift progress and ongoing gains we've delivered in real-world settings. At call centers, like I noted during my prepared comments, AI agents are equaling and in some cases exceeding what human representatives do on particular jobs, delivering dependable, provable commercial results for enterprise clients. We're only at the start of an extremely, extremely lengthy journey, and I think we'll see a shift happen globally from human call center agents to AI agents, in addition to the way large language models have transformed software engineering. For individuals and for society, this is genuinely a positive development, since it will liberate people from work that is highly monotonous and pressure-filled. Spending hours speaking with various individuals about an identical assignment while keeping every exchange strictly businesslike is deeply tiring and takes an emotional toll.

Jingbo Wang executive
#6

Yeah. So when it comes to where conversational AI actually gets used, I can break it into 2 verticals: call centers and companionship devices. Tony already went into a lot of detail — he covered quite a bit about call centers. But there's 1 thing I want to emphasize. A cost center isn't just one use case. It's a whole collection of many, many use cases, and each one has its own features and its own knowledge. And when you look at these use cases, the difficulty of having those AI agents replace them really forms a spectrum — easy ones on the left-hand side, hard ones on the right-hand side. Right now we're only starting to explore a handful of use cases on the license side, things like outbound marketing surveys, but those are low hanging fruits. As we keep refining what we offer and build up more experience with our customers, we'll move more and more use cases from impossible to proof of concept to real-world production. So there's a very long run rate here; it isn't one single use case we can counter in 1 quarter or in 1 year; this is going to be a multiyear process. And secondly, on the companionship device, we've talked with our customers before; this quarter, we actually expanded into new markets in Japan and other countries, and the initial feedback has been quite encouraging. We also partnered with several chip makers so our solution works with more chips, because these aren't mobile phone chips — they're very specialized LTE chips — and that will make our solution available on a wider range of smart devices, including robots. So overall, we're still targeting a 5% revenue contribution from conversational AI by the end of this year. Competition. Tony, do you want to talk about competition?

Bin Zhao executive
#7

Okay. You want to talk about gross margin or not?

Jingbo Wang executive
#8

No, not right now.

Bin Zhao executive
#9

All right. Turning to the competitive landscape, particularly in conversational AI — that space is made up of a few separate technology layers. You have the agent layer, which handles orchestration and optimization for a call center or the calling experience. Then there's the model layer, covering large language models plus voice models like ASR or TTS. And lastly, the infra layer, which includes telecom APIs and cloud. Each layer draws different competitors into this market from their own angle. Take Twilio, for instance — they'd play to what they do well in telecom API and phone numbers through their CCaaS operation. Our own focus is on voice models, audio preprocessing and post-processing, low-latency cloud infrastructure and the agent layer, so that we can deliver the best call experience possible. With the conversational AI market carrying such enormous potential, competition showing up is only natural. In truth, much of the technology behind conversational AI has to do with audio processing — things like dealing with noise echos or packet loss. Clearly, we bring plenty of experience in those areas, and that can greatly enhance the conversational AI experience. That's why we stay confident in our position within this market.

Jingbo Wang executive
#10

Harry, is that a satisfactory response to what you asked?

Harry Zhuang analyst
#11

It is. Those details helped. Once more, well done on the results.

Yue Xu analyst
#12

Nice work on yet another solid quarter. Regarding conversational AI, that's my opening question. So might you provide an update on how conversational AI revenue is progressing, plus what portion of the full-year figure it is expected to represent? The second thing I'd like to ask concerns how long it takes for AI use cases to convert. Would you be able to share what backlog currently exists for these AI use cases, and perhaps just the customer accounts? Also, how do you expect AI revenue to trend across the quarters ahead, possibly into next year?

Jingbo Wang executive
#13

Appreciate it. Thanks for asking. We've already covered the use cases extensively. Let me begin by clarifying something: getting a use case to truly scale up from where the POC begins — the point at which AI agents are able to produce steady results and be rolled out broadly — genuinely requires a fair amount of time; the deployment itself is only a matter of several months. I've mentioned that there are a great many use cases, and each one has to be tackled individually. We're able to handle a handful simultaneously, yet even so, this will take — this will be a considerable undertaking. As I said before, our aim is a 5% revenue contribution by year-end; put simply, our objective is that in Q4, or measured by the run rate — the AI run rate — we intend to reach 5% by the time the year closes. Clearly, that implies the total for 2026 won't come to 5%. It will be less than that. However, if we do hit 5% by year-end, and considering the robust pipeline already in our hands, which is only going to expand by year-end, we think substantial growth potential remains for next year. Indeed; so our outlook on how we're positioned in this market stays quite positive.

Zongxuan Yang analyst
#14

All right. There's just a single item I want to ask about, and it concerns the AI side of our operations. Has guidance ever been provided by you covering the long run — roughly 3 to 5 years out — for the rate at which AI penetrates our overall revenue, and likewise for gross margin expectations?

Bin Zhao executive
#15

I'll handle that one. Ultimately, the market for call centers is going to break into 3 segments. The first covers the simplest work, things like basic notifications. IVR-based technology will take care of that — it isn't genuinely intelligent, yet it can pick out easy keywords spoken by a person. The second segment covers the toughest jobs, or the ones that matter most, like resolving complaints from customers who are high-value, or an emergency situation; people will keep handling those. Even when an AI agent could technically do the job, there are situations where responsibility can only rest with a human, as with a 911 call. The third segment covers all the rest that sits between those earlier tools, and voice AI agents possessing genuine intelligence will handle those; pinning down precisely how large this portion becomes is difficult, though it will represent a significant share of the whole market.

Jingbo Wang executive
#16

Alright, so Tony covered the 3- to 5-year view for the call center sector, and that's an enormous space. Right now, roughly 20 million individuals across the globe work in call centers. As Tony mentioned, going forward there will be 3 categories, and the middle one is likely to end up being the largest, and voice AI agents will take that on. So even capturing a modest slice of that market would mean a dramatic change for us. At this moment, though, providing a precise figure is difficult since this market remains at a very, very early stage. On top of that, adoption of AI agents overall is still quite minimal. That makes it tough to predict precisely how things will unfold, yet it will definitely be a transformation for us. Now regarding gross margin: currently, our gross margin isn't high. We discussed this last quarter — it's crossing the negative positive line, but it's still around there. That isn't due to any fundamental reason; rather, first, we are subscale. There's more volume, and volume occurred across different geographies, so within each single geography, or even smaller. So subscale. Second, we haven't truly concentrated much on technical optimization. At present, our optimization centers more on the experience about the cost. Once we have a larger schedule and also place more emphasis on the technical activation on cost, we believe ultimately the gross margin will be similar to, if not higher than, what we currently have in the RTE business.

Analyst analyst
#17

Good quarter. Tony, about the plan you announced to buy an extra $20 million of stock in the open market — how many shares have you actually acquired so far? Also, what valuation do you think properly captures Agora's intrinsic worth? And with the stake you already hold, have you looked into taking the company private, or giving more capital back to shareholders via a special dividend or an accelerated buyback program?

Jingbo Wang executive
#18

I'll take that one, since it's fairly technical. To be clear, Tony hasn't begun, because of some blackout and legal restrictions; once those clear, he'll begin repurchasing — he's obviously an insider, and he made a prior purchase earlier this year. So legal restrictions do apply; as for the $20 million, that hasn't started yet. Through share repurchase, we've returned roughly $160 million of capital to shareholders so far. That's against a market cap of about $400 million, $350 million for the company, which is substantial — likely among the largest of any public company worldwide; so we'll keep doing that. That said, a special dividend hasn't been considered at this point — we may look at it down the road, but not right now.

Bin Zhao executive
#19

I think my purchase will start in like 2, 3 weeks, right? September, October.

Jingbo Wang executive
#20

Correct. That's possible, though some requirements would apply.

Bin Zhao executive
#21

Understood. Taking the company private isn't something I expect us to pursue. Our intention is to stay engaged with the capital market while we concentrate internally on running the business and strengthening our technical and operational capabilities overall. Our confidence in where we're headed remains strong — the potential there is substantial. We believe that by zeroing in on business operations and technical progress, we can create significant value for our customers, our shareholders and our employees.

Operator operator
#22

That brings us to the end of the question-and-answer portion — there are no more questions coming in. We appreciate all of you joining us on today's call for the company. Just a note before we go: you'll be able to find a replay along with the earnings release on the company's site, investor.agora.io. Should anything else come up, don't hesitate to reach out to the company by e-mail. Thanks so much. You're welcome to hang up now. Have a wonderful day, everyone.

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.