Kakao Corp. (A035720) Earnings Call Transcript & Summary

[Interpreted] We will begin Kakao's 2026 Second Quarter Earnings Conference Call. This conference will start with a presentation followed by a divisional Q&A session. [Operator Instructions] I will now turn it over to Kakao.

[Interpreted] Hello. I'm Jeffrey from Kakao's IR. We will now begin the conference call. Today, I am joined by Shina Chung, the CEO; and Jayden Shin, the CFO. Please be reminded that the earnings results are consolidated estimates under the K-IFRS basis and are subject to change upon the auditor's review. Also, forward-looking estimates are based on assumptions, so actual results may differ from figures included in today's presentation. Now we will have a presentation from the CEO, Shina Chung.

[Interpreted] Good morning. This is Shina, CEO of Kakao. In the Internet and mobile areas, the tension economy focused on capturing users' attention primarily through advertising was a key theme. In the coming era of agentic AI, however, we expect the full scale shift towards intention economy, where new value is created in the process of understanding users' intent and completing actions. Amid this major transition, Kakao's aims to make agentic AI user experience, a habit. Kakao plans to provide a new form of agentic AI experience that identifies users' intent within the context of their conversations, recommends various types of actions based on the identified intent and enables agents to carry out those actions on their behalf. In particular, our goal is for this use of AI, not to remain a onetime experience but to become a new user habit that is repeated in everyday life. As the first step in delivering a new agentic AI experience, we continue to actively enhance the on-device AI models that we have consistently emphasized, enabling them to capture not only explicit intent, but also intend to embed within conversational context, while further deepening their overall understanding of the users. Our recently introduced Kanana 2 1.3B model delivers superior performance compared with global open source models in areas that are essential for operating real-world AI services, including conversations, knowledge, instruction following and tool calling despite it's significantly smaller parameter size. This lightweight model is already being used across Kakao's AI services, including Kanana and KakaoTalk, Conversation Summary, Call Summary and the AI National Secretary. We have also open sourced the model on Hugging Face, making it available for anyone to download and run. At the same time, we are continuing to enhance the long-term memory capabilities of our own device AI models. With this enhanced long-term memory capabilities, Kanana and KakaoTalk will move beyond understanding only fragments of conversations. Instead, it will understand user schedule and interest scattered across multiple conversations within a single context, enabling it to identify users' intent with a greater accuracy. Furthermore, the longer users engage with Kanana and KakaoTalk, the more their behavior patterns and preference accumulate, enabling us to deliver increasingly personalized services. As we continue to enhance our ability to understand users' intent, we believe that the key factor in making the new AI experience a habit is how far we can expand agentic actions that lead all the way to real-world actions and transactions across users' everyday life. Kakao has prioritized partnerships with top tier players in verticals, the users engage with frequently in their daily lives, and where AI agent suggestions and recommendations can naturally lead to real actions and transactions. Among those verticals, the first to be integrated into Kakao's Agentic AI ecosystem is food delivery. And today, we are announcing our A2A partnership with Coupang. Once Coupang is connected to Kanana and KakaoTalk through an agent-to-agent integration, the on-device AI model with KakaoTalk will identify users' intent to order food based on the conversational context, recommend menu items based on the preference it has accumulated over time and enable users to complete the entire ordering and payment process naturally within the chat room. The 2 companies are currently engaged in detailed discussions across multiple areas to ensure that agentic AI experience within KakaoTalk can lead all the way to actual transactions, and we plan to bring the services to market in the near future. At first glance, this may appear to be simply additional food ordering feature within KakaoTalk. However, we believe it is meaningful because it represents the first step towards establishing a new user habit, where AI understands the user intent connects it to real-world actions and helps users complete the recurring everyday tasks through AI. Building on this, Kakao's AI ecosystem will expand into verticals that users engage with frequently in their daily lives, including commerce, reservations, travel and payments. Across these verticals, we aim to build compelling reference cases through direct partnerships with players that occupy the top of consumers' minds in Korea. At the same time, we also plan to actively leverage PlayMCP to expand the ecosystem more rapidly. Today, PlayMCP already hosts more than 3,400 MCP tools, generating over 100,000 calls per day. Just as App Store grew in the early days of mobile era, not only through major apps but also through a wide range of everyday apps created by individual developers. The MCP ecosystem is likewise expanding to include MCP tools provided by both large enterprises and independent developers. We plan to identify the capabilities that we believe can become part of users' daily lives and integrate them into Kanana and KakaoTalk, thereby expanding both the scope and the value of AI agent within KakaoTalk. At the same time, by connecting Kakao's established authentication system and secure payment infrastructure, we aim to build an ecosystem that is both stronger and faster to scale. In the second half of the year, we are preparing to gradually introduce new agentic AI experiences within KakaoTalk. In the mobile era, the dominant interface was the GUI, where users navigated 6 screens and manually test buttons. As we enter the AI era, however, the interface is evolving into the LUI, or Language User Interface, where users interact with agents through natural language such as text and voice. KakaoTalk already has an LUI-based communication channel in the form of chat rooms where users send messages and images to their friends. This makes it possible to transform the chat room into the prompt interface, where users can simply make requests to AI agents or easily create their own AI skills. In addition, we plan to expand AI interactions beyond text, so the users can also communicate with AI through voice, just as they would with a real colleague. Building on the reliability that KakaoTalk has maintained over the years, we will continue to evolve the platform so that it can seamless support everything from authentication and permissions to memory, AI tools, payments and connections between agents. In the second half of this year, while pursuing multiple initiatives in parallel, and we expect the number of monthly active users who begin engaging with AI services within KakaoTalk to exceed 10 million by end of the year. Looking at our second quarter results. The strength of Kakao's platform competitors drove growth across the platform business from advertising, e-commerce and mobility and pay. As a result, profit growth outpaced the revenue growth with both consolidated revenue and operating profit reaching record quarterly highs. This demonstrates that the stable growth and profitability of our existing platform business are providing us with a solid financial foundation to continue making efficient investments in AI. In the era of agentic AI, the key competitive advantage will no longer be who builds a better model, but who can most naturally embed AI into users everyday lives and connect it through real-world actions. In this new competitive landscape, Kakao is well positioned to deliver the best agentic AI experience, having already established a strong user touch points and the broad service ecosystem. Considering both the direction of our AI strategy and the speed of execution, we expect to establish the foundation for the mass adoption of AI by the end of this year and begin meaningful monetization starting next year. As we have previously shared, in addition to transaction fees from agentic e-commerce and subscription revenue from AI services, we also expect the new category of AI advertising revenue, something that Kakao has not had before, to begin contributing in earnings. By 2028, we expect AI-related revenue within Kakao Talk Biz to expand rapidly into double-digit percentage of Talk Biz revenue, making AI both the key driver of renewed Talk Biz growth and one of our major revenue streams. Kakao transforms everyday life in the mobile era. Kakao is now pioneered in the AI era as the first mover, leading the mass adoption of AI so that everyone can have their own AI agents. Throughout this transformation, AI is not only a onetime initiative for Kakao but the foundation of our sustainable growth. That is why we are designing the future with safety as our top priority. Building on the many initiatives and investments we have made over years, we will continue to introduce innovative AI services step by step, establish AI as a new growth engine and ultimately enhance shareholder value. We sincerely appreciate your continued support. Next, Jayden, our CFO, will present the 2026 second quarter financial results.

[Interpreted] Hello, this is Jayden. I will now present Kakao's consolidated financial results for the second quarter. Consolidated revenue for the second quarter was KRW 2.099 trillion, up 9% Y-o-Y and 8% Q-o-Q. Platform revenue reached KRW 1.230 trillion, growing 17% Y-o-Y and 7% Q-o-Q. First, revenue for Talk Biz, Kakao's core business, was KRW 643 billion, up 12% Y-o-Y and 2% Q-o-Q. Second quarter Talk Biz advertising and subscription revenue reached a quarterly record high of KRW 400 billion growing 14% and 11% Q-o-Q. Starting with Business Message. Revenue grew 20% Y-o-Y supported by solid demand from financial advertisers. Advertisers that previously use messages mainly to deliver information are increasingly expanding their use of messages for marketing purposes. As a result, total message volume continues to rise, and we expect revenue growth to remain solid throughout the second half. Moving on to display ads. Following a revenue classification this quarter, I will discuss Talk DA and network ads separately. Talk DA, which covers display ads served within KakaoTalk grew 28% Y-o-Y. Growth was driven by higher user activity centered on feed-based services and solid performance from new ad products. For reference, on the same basis, Talk DA grew 19% Y-o-Y in the first quarter, indicating that growth accelerated further Q-o-Q. In the second half, we plan to further expand commerce-related ads and diversify our ad products to capture demand from a broader range of advertisers. Due to these efforts, we expect to continue attracting new advertisers while encouraging existing advertisers to increase their budgets. By contrast, network ads revenue, which covers inventory outside KakaoTalk declined 45% Y-o-Y. This was due to lower traffic on certain platforms and changes in advertising strategy. In the second half, we plan to strengthen the revenue base by expanding our product portfolio and diversifying our media partners. Beginning next year, we expect revenue to gradually recover and return to a stable growth trajectory. Moving on to commerce. Second total commerce GMV was KRW 2.7 billion up 9% Y-o-Y but down 5% Q-o-Q. Talk Gift maintained solid growth by strengthening its product lineup in luxury beauty, food and baby and kids categories, while expanding promotions to capture demand during Korea's family month in May. As a result, GMV for all key gifting occasions reached record highs. As mentioned last quarter, expanding self-purchases is one of Talk Gift's key growth initiatives this year. In the second quarter, self-purchase GMV grew 39% Y-o-Y, along with an inflow of new users purchase frequency among existing users also increased confirming the effectiveness of our strategy. Self purchases currently account for around 20% of total Talk Gift GMV. Given the significant room for further growth, we expect self-purchases to remain a key driver of Talk Gift's growing going forward. For Talk Store, we made promotions more regular and expanded collaborations with major brands to increase user visits and purchase frequency. As the number of buyers, payment transactions and average order value all increased GMV grew 7% Y-o-Y. Commerce revenue recorded KRW 243 billion, up 10% Y-o-Y. Growth was driven by higher GMV at Talk Gift and Talk Store, strong 1P sales and the recognition this quarter of a portion of GMV from the Shopping Festa held in March. Commerce revenue declined 10% Q-o-Q, reflecting the high base from Lunar New Year demand in the first quarter. Platform and other revenue reached KRW 587 billion, up 22% Y-o-Y and 12% Q-o-Q. Starting with mobility, solid performance from existing business was complemented by strong growth in last-mile logistics. As a result, both revenue and operating profit continued to grow at a healthy pace. For Kakao Pay, financial service revenue grew significantly, supported by increased trading activity in both domestic and overseas equity. Payment and Platform Services also maintained double-digit growth, enabling Pay to achieve record high quarterly revenue and operating profit. Moving on to the Content segment. Second quarter content revenue was KRW 868 billion, up 1% Y-o-Y and 10% Q-o-Q. First, Story revenue was KRW 211 billion, down 16% Y-o-Y and broadly flat Q-o-Q. Piccoma's revenue in yen declined 9% Y-o-Y as a slowdown in Japan's manga market continued. The second quarter includes Golden Week, Japan's major extended holiday and the manga market's biggest signaled seasonal peak. During this quarter, Piccoma carried out active marketing activities to mark its tenth anniversary and introduce new short-form animated video content. As a result, Piccoma maintained its #1 position in gross revenue across Japan's overall ad market, both in the second quarter and for the first half of 2026. In the second half, Piccoma will continue sourcing high-quality titles and strengthening its lineup to enhance the competitiveness of its core business. At the same time, it will work to build traction for its newly launched short-form animation and merchandise businesses. Through these efforts, Piccoma will expand its business across the full IP life cycle in and distribution to downstream IP commercialization while developing new growth drivers. Meanwhile, Kakao Entertainment Story business saw slight Q-o-Q increases in both domestic platform GMV and revenue. This was supported by visible results from its strategy of expanding popular web novel IPs into webtoons. Next, Music revenue reached KRW 558 billion, up 8% Y-o-Y and 15% Q-o-Q. Y-o-Y growth was driven by expanded concert activities by major artists together with higher MD and licensing revenue. Q-o-Q growth was driven by increased album sales and MD revenue following comeback by anchor artists, including aespa and NCT WISH. Lastly, media revenue was KRW 99 billion up 5% Y-o-Y and 7% Q-o-Q, reflecting an increase in the number of titles under production. Next, regarding operating expenses. Second quarter operating expenses were KRW 1.821 trillion, up 6% Y-o-Y and 5% Q-o-Q. Labor costs increased only 2% Y-o-Y as we maintained a conservative hiring step. They remain broadly flat Q-o-Q at KRW 448 billion. Marketing expenses reached KRW 97 billion, up 22% Y-o-Y and 33% Q-o-Q, reflecting expanded marketing activities at Piccoma and Kakao Pay. For reference, marketing expenses accounted for 5% of consolidated revenue in the second quarter. Cost of revenue was KRW 746 billion, up 10% Y-o-Y and 6% Q-o-Q. This was due to higher production costs following expanded activities by major music artists. Despite continued growth in infrastructure demand, we are effectively managing cost increases through efficient investment and operations. As a result, outsourcing and infrastructure costs increased only 2% Y-o-Y and rose 7% Q-o-Q to KRW 223 billion. Depreciation and amortization expenses were KRW 192 billion, down 8% Y-o-Y and 3% Q-o-Q. The decline reflected the base effects from content license amortization and one-off bad debt expenses. Consequently, second quarter consolidated operating profit was KRW 277 billion, up 36% Y-o-Y and 31% Q-o-Q. The operating margin was 13%, improving by approximately 3 percentage points Y-o-Y. For reference, stand-alone operating profit for the second quarter was KRW 126 billion, up 17% Y-o-Y. The stand-alone operating margin improved by approximately 1 percentage point to 18%. Moving on to nonoperating items. Second quarter nonoperating income was KRW 87 billion increasing by KRW 51 billion Y-o-Y, reflecting a gain on disposal from the sale of VX. On a Q-o-Q basis, nonoperating income decreased by KRW 34 billion due to lower equity method income and lower dividend income following the disposal of investment stakes. Second quarter consolidated net income was KRW 18 billion. The gap between operating profit and net income this quarter was attributable to a one-off loss from discontinued operations arising from the process of simplifying our governance structure as well as income tax expense. These items should be viewed separately from the operating performance of continuing operations. Profit before income tax from continuing operations was KRW 364 billion. After reflecting the one-off income tax impact, consolidated net income from continuing operations was KRW 199 billion. Lastly, total CapEx for the second quarter was KRW 188 billion. This consisted of KRW 165 billion in PP&E and KRW 24 billion in intangible assets. CapEx increased by KRW 89 billion Y-o-Y and KRW 70 billion Q-o-Q as server purchases required for the full year were concentrated in the second quarter. This concludes the presentation on the second quarter earnings of 2026. We will now proceed to the Q&A session. [Operator Instructions]

[Interpreted] [Operator Instructions] The first question will be provided by Eric Cha from Goldman Sachs.

[Interpreted] I would like to submit 2 questions this morning. I understand, and also you've mentioned this in your opening presentation, in order for you to actually create an AI agentic-based ecosystem, you were planning on partnering up with external vertical service players. But compared to the schedule that you shared with us and during the previous earnings call, it seems like there has been a bit of a delay. I would like to know as to why that is the case? And can you also update us on where the status is at this point? And if you could lay a possible launching schedule going forward, that would also be very helpful. My second question relates to your Kanana in KakaoTalk and ChatGPT in Kakao. So I would like you to share with us some of the user metrics and engagement metrics and what your strategies are going forward to improve those metrics.

[Interpreted] This is Shina. I will be responding to the 2 questions that you've asked. First, in regards to Kanana in KakaoTalk and ChatGPT in Kakao. Agentic AI that is envisioned by Kakao is not simply connecting to external APIs, but it is one that creates new user experience. in which AI makes suggestions and recommend services within the context of the conversation and completing the flow all the way across authentication and up to ordering and payment process. And in order for this to happen, partner service strength and competitiveness has to stay intact. And that is why we need to redesign service architecture and user experience to allow AI to naturally intervene into the scene. That is why we also need to go through a process to align R&R roles and responsibility of each of the respective companies and also match baseline for business model and operations. Now this approach is yet to be proven even in the global market as the agent goes beyond the boundaries of the platform and has to be incorporated into the off-platform ordering and payment processes and make that connection to real-world activities. So through our initial partnership, we felt that it is very important to show how agentic AI actually works in connection to action thesis in everyday services. And we felt that delivery service was most appropriate because of high frequency of usage by the user base and relatively lower user intervention that is required along the process because experience and value is mostly placed on how fast and convenient the process is rather than having to conduct in-depth information searching. So in creating the most optimal experience, we wanted to be able to validate the convenience. And we felt that this would actually show the utility of agentic AI the best. So in creating the most optimal experience, we want to be the first to validate the scalability and business feasibility of agentic AI business. So user intent and connecting that to ordering and payment process, we are able to generate, I believe, through this process, new customer acquisition for our partner companies and also drive GMV growth. And this will become strong basis for scaling agentic AI ecosystem as well as our partner base. And based on such experience and operational know-how, we will develop A2A based integration structure supporting role of the agents and also the call method. And repeatable features like authentication, delegation and ordering and payment will be offered in the form of SDKs and agent builders. And for our subsequent partners, we believe that they can implement this optimal agentic AI experience very quickly with less cost and development resources, while the ecosystem scales much faster, leveraging PlayMCP. Now let me move on to your -- the second part of this question on the update on AI services and our plans to expand on user engagement. So looking at Kanana in KakaoTalk and ChatGPT for Kakao. Firstly, for Kanana in KakaoTalk, since its release, it's been -- we've been releasing enhancements on average of every 2 weeks based on user feedback and incorporating data usage as we further refine user experience. Because competitiveness, we believe it will be determined by how fast we incorporate actual usage data to our services. And the results show that of the users who left positive feedback on proactive message and the responses were above 80% and 90%, respectively. In particular, we see high ratings for intent-based recommendations within the context of the conversation for search, shopping and local play services. With users seeing how useful it is, we are also seeing increase in repeat and recurring usages on the back of better service experience. And so user retention has moved close to 80% and versus 70%, which was immediately after the release. So we plan to expand on intent to action, conversion, connecting all the way up to ordering and making payments with our endeavor with the delivery platform and make sure users intent are precisely captured within the conversational context so as to build on that experience. We will also start implementing marketing in full swing around key use cases when it becomes clear that we have full laid the basis to drive sustainable usage of Kanana in KakaoTalk as part of people's daily lives. We will quickly increase new customer acquisitions and user acquisitions and will enhance service experience, so their first try can lead to repeated usage. ChatGPT for Kakao in the second quarter had a reported 13 million cumulative subscribers. So we're now in a phase where activity and usability metrics is more important than subscriber numbers. Daily average outbound message per user now exceeds more than 6 messages and daily average time spend is close to 8 minutes as of the end of the quarter, showing very clear uptrend in activity as we move beyond subscriber expansion stage. And to further facilitate this trajectory in recent update, we enabled the feature where ChatGPT can be called into the chat room like a chatbot and people can ask questions and share the responses with others in that chatroom. And we expanded this to the PC platform as well, further scaling the touch point. This will help people to use ChatGPT seamlessly wherever they are within KakaoTalk whenever they have any questions. We will try to spread such usage as part of the conversational theme and through the sharing. So in the second half of the year, we will further drive agentic AI experience of Kanana in KakaoTalk and also enhance image generation and chatbot features of ChatGPT for Kakao as we further scale AI services embedded in the Talk platform. In light of user metrics that we are seeing at this point, and future plans on our service expansion, we think we can achieve 10 million MAU by the end of the year as people become accustomed to such services.

[Interpreted] We'll take the next question.