Naver Corp Ord

Good morning. We will now begin NABIRS 2026 Q2 Earnings Conference Call for the benefit of our investors. Joining from home and abroad, we will provide simultaneous interpretation service for the presentation and switch to consecutive interpretation for the Q&As. Analysts, investors, good morning. I am Paul Choi from the Capital Markets Office. I would like to thank the analysts on this call, we are joined by CEO Soo-yeon Choi and CFO Hee-cheol Kim, and they will walk you through Naver's business highlights and strategies and financial results after we will entertain your questions. Please note that the earnings results are KIFRS-based, provided for timely communications, and have not yet been audited by an independent auditor, and hence are subject to change after such review. With that, I will turn it over to our CEO to present on our business highlights. Good morning. I am Soo-yeon Choi, the CEO. Before I begin our Q2 business update, I'd like to talk about our strategic partnership with NVIDIA and our AI factory business. As announced in June, Aver is working with NVIDIA to jointly build and operate AI factories on a global scale. We expect to begin generating revenue with the launch of our first 55 megawatt AI factory in the first half of 2027. We then plan to expand capacity to 100 megawatts by the end of 2027 and 200 megawatts in 2028. On our long term, it is to build AI infrastructure at the gigawatt scale. We see this as an investment in a structurally growing market, not a one-time business opportunity. As generative AI adoption accelerates, AI transformation expands across industries, and inference in AI agents become more widespread, we expect global demand for AI computing to continue growing over the medium to long term. Some have raised concerns that low-cost, high-efficiency models could reduce computing demand. However, we see the opposite, as the cost per unit of compute declines as AI adoption, inference, and AI agents will become more widespread, driving further growth in the overall computing demand. Today, coding is the most proven AI use case. Demand for computing across broader industries and physical AI has yet to fully emerge. As AI expands across more industries and services, we expect additional inflection points where overall computing demand accelerates again. We are already seeing early signs of this trend. Leading global technology companies continue to report strong cloud performance and growing older backlog. Their continued investments in AI also demonstrate that AI demand for AI infrastructure is already becoming a reality. Meanwhile, despite growing demand, expanding supply takes time. It requires large-scale power, suitable sites, cooling infrastructure, skilled talent, regulatory approvals, and broad community support. Many companies are now beginning to build new data centers. They continue to face bottlenecks, including labor shortages, power constraints, limited semiconductor supply, and local opposition to New Construction. Nibiru is in a different position. Through our partnership with NVIDIA, we have secure technology and supply support. We also have ready access to power and sites, strong capital partnerships, and yet years of experience operating a full-stack platform. In addition, we were the first in Korea to secure customers and launch GPU services. We will build on that experience as we scale the business. To further strengthen this partnership, NVIDIA has decided to participate for the first time in Asia in a private placement of approximately US$1 billion of newly issued neighbor shares. This goes beyond the technology partnership. It reflects a long-term commitment in which both companies will share the opportunities and the risks of this business. At the same time, we are in discussions with Brookfield, a leading company Global Asset Managers to finance $9 billion of computing infrastructure required for the AI factory. We have selected Brookfield as our exclusive preferred partner and plan to move toward a definitive agreement. Under this partnership, NVIDIA will provide the technology and ecosystem, our capital partner will finance the infrastructure, and Naver will lead the platform operations and customer service. This structure allows us to scale the business quickly while keeping our upfront capital requirements to a minimum. We are also confident in our ability to secure customers. Through our global off-takers, NVIDIA's ecosystem, and our existing enterprise and public sector customer base, we expect to secure customers on their initial capacity ahead of launch. We have many years of experience operating data centers reliably. Our experience together with our sovereign AI track record demonstrates our full-stack operational capacities. to customers that require security and reliability. The AI needs of each industry are different. Manufacturing requires AI for production innovation. Defense requires AI infrastructure in air-gapped environments. The public administration sector is focused on AI transformation, while financial institutions prioritize data security. Yet across all these industries, demand is growing for trusted infrastructure and technology partners. NABIRD has both strong track record in areas where security and data sovereignty are critical. This includes GPU services for leading Korean enterprises, a nuclear power plant AI platform for Korea Hydro and Nuclear Power, and a dedicated AI platform for the Bank of Korea. More recently, we have also expanded into the defense sector. Through partnerships with Korea Aerospace Industries and Hanwha, we are continuing discussions on AI models and dedicated private AI cloud infrastructure for defense applications. The reason Naver has been able to continue building these references is that we have expanded our capabilities beyond data center and cloud operations and sovereign AI model development by adding the AI factory businesses. This allows us to flexibly provide the services the market needs. With our full stack capacities, expanding infrastructure model services, and together with our proven sovereign AI experience in Korea and overseas, we are approaching this market with a high value added offering that combines GPU computing cloud models and managed services, rather than simply leasing computing capacity. Supported by long-term contracts and high utilization rates, we aim to deliver meaningful revenue and profit early on. Through our AI factory business with NVIDIA, we will secure a new growth driver in the global AI computing market. Next, let me walk you through Neaver's platform performance for the second quarter. Neaver's core strategy is actionable AI, We are delivering a new search experience that seamlessly connects search, commerce, and payments. At the same time, we are actively applying AI across our services to improve content creation and quality, enhance the effectiveness of our advertising products, and deliver a better user experience. These efforts began to translate into stronger user engagement and improved monetization metrics in the second quarter. First, let me highlight the performance of AI Tab, our flagship actionable AI service. AI Tab officially launched in June 25th. As of early August, it has surpassed 10 million monthly active users, establishing itself quickly as a new search engine. Weekly return rates, a key indicator of user loyalty, have also more than doubled since the early testing phase and continue to grow solid momentum. At the same time, the CTR for shopping and local content has exceeded 40%, showing that users' exploration is translating into actual purchases and reservations. Powered by personalized recommendations based on real purchase data and shopping reviews, AI Tab continues to shorten the journey from search to decision-making. AI Tab is quickly establishing itself as actionable AI that helps people in their everyday lives. AI Tab Q3 will enhance a real estate search function and whale browser agent. We also plan to launch a health agent later this year. Going forward, AITel will continue to evolve with a focus not only on quality of its recommendations, but also on its contribution to actual transaction conversions. Through this, we will further strengthen the virtuous flywheel that drives transactions across our platform. We will also continue to strengthen discovery and exploration by securing our content and data. Through NeighborMate, we have built a foundation for attracting high-quality creators and encouraging quality content. During the FIFA World Cup in June, we were also able to apply our media AI technology to automatically generate highlight clips during matches. In addition, we also introduced tag recommendations and anti-based features to further improve content quality. Monetization through generative AI advertising is also gaining traction. AI Briefing Ads completed testing during the second quarter and officially launched at the end of July. They are significant in two ways. First, generative AI services are beginning to emerge as a new source of advertising revenue. In the near term, we are primarily applying AI Briefing Ads to informational queries that were previously not monetized. This minimizes cannibalization of our existing advertising businesses. As a result, we expect AI briefing ads to expand to new advertising inventory rather than replacing existing search ads. Initial results have also been positive. AI briefing ads delivered a cost per click that was 30% higher than existing search ads, and CTR was also more than 30% higher. Purchase conversion rates improved by more than threefold, demonstrating strong effectiveness from advertisers' perspective. In the generative AI era, the key competitive advantage lies in understanding users' intent and context with greater precision and connecting them with the most relevant ads in real time. Going forward, we will apply the ranking and recommendation technologies validated through AI advertising across our advertising products, further strengthening the competitiveness of our advertising platform. Second, generative AI is becoming the foundation for increasing transaction value across the platform and expanding long-term monetization opportunities. In particular, the strong purchase conversion rates of AI briefing ads demonstrate more than just improved advertising efficiency. They show that AI can accurately understand users' intent and connect them to actual purchases and reservations. Naver is the only full-fledged platform in Korea that seamlessly connects search, shopping, place, and payments. As AI agents become more widely used for everything from exploration to purchases and reservations, we believe this full funnel structure will become a key growth driver for expanding transaction volume across our platform. AI briefing AI's are the first example of how actionable AI is creating new monetization opportunities. As AI agents play a greater role in connecting users with services throughout the transaction journey, Naver will continue to create new monetization opportunities beyond advertising, including commerce and payments. In doing so, we will further expand the value of our platform. Next, let me discuss our commerce business, which continues to drive growth across our services. In the second quarter, Smart Store GMB grew 15.5% year-on-year, with growth continuing to accelerate. This was driven by the three engines of the NeighborPlus Store app membership and delivery, working together to strengthen the structural competitiveness of the NeighborCommerce and deliver solid business results. First, the NeighborPlus Store app has firmly established itself as a primary transaction channel for our loyal customers. In the second quarter, GMB generated through the app grew more than 35% year-on-year, significantly outpacing our overall growth. The app's share of total GNV also continued to increase rapidly. In particular, traffic from returning users is growing at about four times the pace of new user traffic. App users also spend more per visit and make purchases more frequently than web users. This shows that the app's growth is being driven by repeat visits and purchasing habits among our existing customers. Membership is also serving as a strong driver of app growth. About 80% of customers who make purchases through the app are membership subscribers. Following the expansion of free shipping and free returns, the purchase frequency of end delivery membership customers increased by about 29% and their share of transactions exceeded 70%. As membership has proven to be a key driver of transaction growth, we plan to accelerate the expansion of our membership base. To support this, we will broaden our everyday lifestyle partnerships and expand offline payment benefits. Our goal is to make membership a core pillar that connects the entire neighbor ecosystem. The most notable achievement in commerce this quarter is that our stronger delivery capabilities are directly driving both user purchases and seller growth. As of June, end delivery GMV on Smart Store increased 76% year-on-year. Coverage also exceeded 20%, keeping us on track to reach our year-end target of 25%. In addition, the average GMV of end-delivery sellers grew significantly, more than that of other sellers in the second quarter. This demonstrates that delivery is not only increasing purchase conversions and repeat purchases, but also serving as a meaningful growth driver for sellers by helping them expand their transactional volume. Our three strategies for expanding end delivery are also progressing as planned. First, we are supporting sellers of core products with high delivery sensitivity. This is lowering the initial barriers to adopting end delivery and helping expand both participating products and sellers. Second, we are expanding direct contracts between neighbor and sellers. The scale of these direct contracts have increased by more than four-fold year-on-year, and we will remain on track to achieve our goal of raising the share of these direct contracts to more than 50 year-by-year end. As we expand these direct contracts, we are standardizing logistics terms and operating standards that were previously fragmented. This also provides the foundation for more competitive pricing and more consistent delivery quality. In addition, we are reducing sellers' operational burden by improving returns, settlement, and order management. This enables even small and medium-sized sellers to easily leverage high-quality delivery services. Third, we are strengthening delivery benefits through membership. Starting in October, we will support faster returns through a dedicated membership return center and officially launch a dedicated early morning delivery service. Just as our enhanced delivery benefits have led to higher purchase frequency and a greater share of transactions among membership customers, We expect that these new benefits in October to further strengthen the virtuous flywheel between the membership and end delivery, while accelerating end delivery growth in the second half of the year. NABR will continue to have the app drive discovery and exploration, membership drive customer engagement across the NABR ecosystem, and delivery drive purchase conversion and repeat purchases, further strengthening the virtuous flywheel. Next, I will talk about the financial platform. In the second quarter, Naver rapidly expanded the installation of NPAY Connect payment devices, marking our full-scale entry into business of AI platform for business owners by connecting offline data with the Naver ecosystem. The purpose of Connect is to extend the success formula Naver has built online through search, data, advertising, shopping, and payments into the offline world while establishing Naver's physical touchpoints there. Starting this year, we plan to rapidly increase market penetration and establish a leading position. Through this, we aim to proactively secure a competitive advantage in data based on offline business owners that will be difficult for competitors to replicate. The offline commerce payments market is more than three times larger than the online market, yet it remains insufficiently digitized, its data is not yet well-structured, and it remains disconnected from the online ecosystem. Naver therefore sees this as a business opportunity with significant growth potential and the strategic area that must be secured early. Based on this view, we have accelerated our rollout schedule which was originally planned to be phased out throughout next year. As a result, just seven months after the launch, NPAY Connect has already secured more than active business owners than expected and continues to expand rapidly. There are three main reasons why Naver is investing in NPAY Connect. First, offline data will become a key asset that strengthens the competitiveness of the entire neighbor ecosystem. As AI evolves beyond providing information and recommendations to understanding users' intent and carrying out reservations, orders, and payments, offline behavior data will become an essential asset rather than an optional one. Nationwide touchpoints with users' everyday activities, including store visits, order history, responses to promotions, and payment methods, combine with neighbors' unique Online data will become a key asset that serves as a moat in the AAI era. NAVER plans to combine place search and reservation data, user's ID, payment, and membership data with the offline data collected through M-Pay Connect. This will connect the entire online-offline journey from search to reservations, orders, payments, and repeat visits under a single ID. And through this, AI agents including Naver AI tab will be able to deliver more sophisticated personalization and more complete execution. Our advertising capabilities will also become more sophisticated. Second, it will create a new monetization opportunity based on the business owner's touchpoints we are establishing, including business owner solutions and financial services for our business providers. We plan to evolve Npay Connect beyond the simple payment device into an agent for business owners that analyzes sales, visitor traffic, changes in the commercial district, while recommending the next actions such as coupons and promotions. Just as SmartStore became the foundation for online business owners, our goal is to provide offline business owners with a new AI platform centered on place and Npay Connect, integrating reservations, orders, payments, customer management, and marketing. Monetization will be rolled out in phases. In the initial stage, we will expand our business owner touchpoints by integrating NPAY Connect with Place. We will enhance CRM, customer analytics, commercial area analytics, coupon and promotional management, and store operation tools to create new monetization opportunities including business owner solutions. We will further expand our revenue streams by adding financial intermediation services such as low-interest insurance, Powered by business owner sales and customer data as well as new advertising products based on offline visit and purchase conversion data Third, it would allow us to significantly expand the scale of our offline payments and build a new revenue base By establishing one of the largest business owners networks in Korea We aim to become the clear market leader in domestic B2C payment volume combining both online and offline transaction as early as next year We also aim to expand the offline payment volume based on neighbor IDs to 130 trillion KRW within the next five years. This will broaden our payment fee revenue base while also expanding the foundation that connects user IDs, memberships, reservations and orders, advertising and financial services. On top of this, we will introduce a wide range of monetization models linked to Connect. Once the initial foundation is in place, our cost burden will gradually decline, while monetization opportunities across payments Advertising business owner solutions and financial services will expand rapidly. The AI platform businesses for business owners, which connects offline data with the neighbor ecosystem, is an area where neighbor strengths can be fully leveraged. Going forward, we will secure business owner touchpoints and a data foundation early and quickly demonstrate profitability by connecting them to a wide range of monetization models. Our global expansion business, One of NAVER's new growth engines delivers strong growth in 24.4% year-on-year in the second quarter. In particular, our C2C business continues to deliver strong GMB and revenue growth, driven by the strengthening competitiveness of the overseas platform we have invested in, including Poshmark, Soda, and Wallapop. In our enterprise business, the digital twin platform developed by our joint venture in Saudi Arabia has been selected as the Ministry of Municipality and Housing's single national standard platform. Development is also underway with the goal of launching a map-based super app later this year. Naver will continue to strengthen the AI competitors of this core business while securing new growth engines through AI Factory to create even greater growth opportunities. Now, CFO Hee-cheol Kim will discuss about the financial performance. Good morning. This is Hee-cheol Kim, the CFO. I will now walk you through Q2 financial performance. In Q2, revenue reached 3.3888 trillion KRW, up 16.2% year-on-year, driven by strong growth of the global C2C business and the performance of the shopping business. As a result of one-off expenses, including deploying and paying Konect devices and recognizing strategic IPs, such as broadcasting rights for the World Cup, as well as planned infrastructure investments such as GPU purchases, operating profits for Q2 recorded $523 and Seon-joo Chae. Based on our long-term data analysis, the actual average useful life of computing assets was longer than the existing depreciation schedule, which led to its extension. Moreover, the useful life schedules that differed across asset types were unified, and it is expected to generate an expense deferral benefit of around 100 billion KRW. For reference, excluding one-off factors such as the impact of consolidating Wallapop and Pi, Connect, broadcasting rights, and changing of the useful life period, Second quarter operating profit increased 0.9% year-on-year. Next, I will explain about the revenue by business segment. In Q2, neighbor platform revenue increased 12.3% year-on-year, driven by strong performance in commerce and consistent efforts to optimize and enhance advertising efficiency. Advertising revenue recorded 1.4472 trillion KRW A7.5 growth year-on-year. Naver continues to refine its ad performance prediction in our targeting models across various industries and placements while expanding AI-driven ad optimization. As a result, AI serves as a key growth driver for the advertising business, contributing more than 60% of incremental ad revenue in the second quarter. In particular, powered by product enhancements, the paying advertiser base for ad boot shopping grew more than 2.5-fold year-on-year, Going forward, we will continue to improve ad efficiency and targeting based on our integrated recommendation model, while expanding new ad opportunities across generative AI services. Q2 service revenue increased 31.3% year-on-year to 455 billion KRW, powered by strong growth in commerce. Smart store GMB in Q2 rose 15.5% year-on-year, owing to the virtuous cycle among our app membership and delivery services. In June, the synergies from Samsung's customer appreciation and the Neighbor Plus sale promotion have materialized, and Neighbor's exclusive new products and brand-collaborated products received strong positive responses, contributing to the quarterly GMB growth. Membership revenue also recorded solid growth, supported by strength in delivery benefits and inflow of new members driven by the World Cup. Next, Q2 financial platform revenue increased 16% year-on-year to $770,000. to 470.7 billion KRW. Supported by smart store growth and expansion of external ecosystems, the total payment volume grew 21% year-on-year to 25.2 trillion KRW in the second quarter. Of the total, off-platform payment volume increased by 26.2% year-on-year to 14.1 trillion KRW and continues to account for 56% of the total payment volume. In the Q2, revenue from global growth areas increased 24.4% year-on-year to 1.159 trillion KRW. In the C2C segment, revenue grew 74.9% year-on-year with ongoing efforts to strengthen core platform competitiveness translated into solid performance. Poshmark saw significant improvements in purchase conversion and purchase frequency driven by quality improvements in AI-powered search and recommendation and recorded over 40% in revenue growth year-on-year. SORA delivered record high earnings with both revenue and GMV growing more than two-fold year-on-year driven by ongoing boom in trading card transactions. Wallapop continues to achieve GMV growth that outpaces the growth in the Spanish retail market supported by the growth of the European C2C market as well as efforts to stimulate platform transaction activities such as launching local delivery services and personalizing the home feeds of their app. At the same time, Wallapop is expanding its market leadership centering around the used car category, which holds top position in the number one registered listings. Content revenue increased by 0.5% year-on-year to $464.3 billion in Korean won in Q2. Within this segment, web team revenue increased by 0.9% year-on-year on a Korean won reported consolidated basis. For more details, please refer to the web team entertainment earnings announcement scheduled for August 10th local time. In 2026, Naver Web Team plans to focus on content diversification and strengthen personalized recommendations while also expanding its user base through new services and initiatives. Snow will continue to expand AI-powered product offerings and work to improve profitability. Enterprise revenue increased by 21.3% year-on-year to 153.7 billion KRW in Q2, driven by sustained B2B revenue from AI initiatives, including the GPU as a service deal secured in the second half of last year. In Q2, revenue posted solid growth as new cloud-based AI businesses with Korea Hydro and Nuclear Power and Bank of Korea transitioned from implementation to the operation and expansion phase. And discussions regarding new adoption within the financial sectors are also continuing. Technological capabilities for robotics have accumulated in the 1784 building was applied to the Tokyo Midtown Yayasu Building in Japan in partnership with NTT East and Mitsui Fudo-san, creating a first external commercialization use of Team Neighbors Technology. In the chain of line works, the number of new paid user IDs greatly increased during Q2, which was driven by securing large-scale deals, further solidifying its number one position in the business messenger market. In April, Global Care Call was officially launched in Japan by entering into an agreement with the city of Izumo and is garnering attention from many local governments and public institutions in the Taiwanese market where we entered in late last year. It was successfully taking roots with more than 100 companies confirming to adopt its services. Next is the detailed cost items. Development and operations expenses increased 15.2% year-on-year with the impact of consolidating WAPLAPOP. Partner expenses rose 22% year-on-year driven by higher commission expenses in line with revenue growth, recognition of content, rights cost, including World Cup and expanded deployment of NPACONNECT devices. Infrastructure expenses increased 11.2% year-on-year, as the effect of a change to useful life was reflected, even amidst continued expansion in computing asset investments. Marketing expenses rose 25.4% year-on-year, driven by strategic marketing investments in the commerce segment, as well as higher promotional spending for NPAY Connect. Amid a rapidly evolving market environment, Naver expects to continue expanding strategic investments in the near term to strengthen competitive positioning in the targeting the offline market through NPAY Connect and to strengthen capabilities for core businesses including commerce and AI-powered advertisements. Ultimately, the company aims for these investments to support revenue growth in its core business and serve as a foundation for long-term growth drivers while contributing to the enhanced shareholder value. Next, I'll explain about Naver's operating profit by business segment. Naver platform segment saw a 4.2 percentage and Ji-hye Kim. In the last two years, we have seen a significant year-on-year decline in operating margin despite solid revenue growth in advertising and commerce, primarily due to increased marketing costs as well as costs related to recognition of IP costs such as the World Cup processing rights. Financial platform segments continue to deliver solid revenue growth in Q2. However, operating margin declined by 4.9% percentage as investment costs necessary for expanding the deployment of connect devices was reflected. In the global growth segment, losses narrowed supported by accelerated growth in the C2C business. Due to consolidated net income increase by 41.6% year-on-year to 704.3 billion KRW, driven by an increase in equity method gains and valuation gains on financial instruments, due to free cash flow decreased by 150 and Seon-joo Kim. In the second half, we expect continued investment for growth, including computing assets like GPUs and CPUs, as well as the rollout of NP-Connect. However, as this year marks peak concentration of market growth, the growth rate-related expense is expected to moderate gradually starting next year. At the same time, as already demonstrated through initiatives like AI briefing advertisements, These investments are expected to generate new revenue streams and drive up top-line growth, progressively contributing to overall financial performance. Lastly, on August 3, Naver retired 4,901,094 Treasury shares, excluding bare minimum needed for granted stock compensation, representing 3.1% of total shares, approximately 1 trillion KRW in volume, to enhance shareholder value. Moving forward, Naver will continue to explore various measures to enhance shareholder value in line with our shareholder return plan and will promptly share any updates with shareholders via public disclosures as decisions are finalized. This concludes the overview of our Q2 financial results and we will now move on to the Q&A session.

To ask a question, please press star and number 1 on your phone. To withdraw your question, please press star and number 2. In consideration of all participants, we kindly request that you limit your questions to 2 per person.

The first question will be provided by Stanley Yang from J.P. Morgan. Please go ahead with your question.

Thank you for the question. I would like to ask you about the AI factory business. We have Naver 100% self-employed company, Opco, and Brookfield FI, which is a major shareholder, SPB. There are two entities, and the profits generated from the AIDC business look like a double structure on both sides. What happens to the economic structure of OPCO and SPV, and I think you mentioned the yield expected by OPCO in the past, and I would appreciate it if you could tell us that again. And I have one more additional question here. I'm curious about the risks or variables when you do business with AIDC. For example, when the price of the token, the price of the computing power is variable, how does Okku and SPB hedge this price variable risk?

Good morning. I would like to ask you a question about your AI factory initiative. My understanding is that Naver will have 100% ownership of the OpCo, the operating company, and there is going to be also SPV where the Brookfield is going to be the majority shareholder. And I would assume that the profit that's generated through your AIDC business is going to be allocated or shared by these two entities. So we'd like to gain some more color as to what the economic structure is between Opco versus the SPV. And also, I think you've mentioned this previously, but I would like to know as to what the projected return or profitability that you are looking for from the operating company is. And also going forward, we'd like to know as to what are some of the risk factors or variables that could actually impact your AIDC business. For instance, from changes in the token price or fluctuations in the compute price. We'd like to know how the OPCO as well as the SPV is going to hedge against that fluctuation in pricing.

Let me answer the first question first. First of all, I think it would be good to refer to the published data on Monday. Basically, instead of the structure that Brookfield, SPV, and Opco are biased, you can think of it as a structure where all the revenue and cost of the computing business, including the margin, are earned by the factory users.

So this is the CFO responding to your first question. I think it would be helpful if you could also refer to the disclosed material which we disclosed on Monday. You've mentioned how the profit or earnings that's generated could actually go under Brookfield SPV as well as the operating company. But I think it's more correct to say that all the revenue, the cost, as well as the margin that arises from this business is going to come under AI factory operating entity.

To run this business, to run a computing business, you need a large amount of computing assets. It's not a structure that Naver does directly, but it's a structure that Brookfield does through SPV and supplies it to us, and it's a structure where Brookfield gets the margin from there.

When you're engaging in this type of a computing business, there's going to be a massive amount of computing asset that needs to be sourced. and Naver, rather than directly sourcing those computing assets, Brookfield, through its SPV, is going to be financing and also sourcing those computing assets, and they will be supplying that asset to us, Naver, and in that process, there will be margin that's going to be generated, which will come under Brookfield.

In the case of Opco's expected yield, I think there will be some changes depending on maturity. In the beginning, the margin rate may start a little low, but depending on the maturity of the business,

In terms of the profitability or the margin levels that we're expecting from the OPCO is that it will depend on the maturity level of the market. Initially, it may start off with a lower level of margin, but as time goes by, basically we are looking forward to at least double-digit margin and eventually around 20% level.

When it comes to AIDC business-related risk, we internally made thorough review and we can identify around four different risks.

The first potential factor or risk factor could be from securing the demand from the customers. But unlike other companies, Naver itself also requires significant amount of AI chip and compute resources. And hence, we will be able to make and strike a good balance between the demand from our external customers as well as for Naver's own requirement and demand. Through that approach, we believe that we will be able to manage this in a quite steady manner.

Second, in the case of chip acquisition and other implementation risks, Naver has established a strategic legal structure that provides NVIDIA as a third party. As a result, Naver has been able to provide chips easily and provide a foundation for business performance.

And secondly, when it comes to sourcing of the AI chips, that's considered to be one of the key aspects of the execution risk. But by onboarding NVIDIA as our third largest shareholder, and by entering into strategic cooperation with NVIDIA, we believe that we have the basis to be able to secure AI chips in a timely manner.

Thirdly, there may be financial risks such as capital collection, interest rates, etc. And third potential risk has to do with the financing risk in regards to the interest cost. We not only have a one single contract arrangement of five to six years of contract term,

We have multiple types of customers where there are various different types of contract durations and terms and conditions. So we will be able to flexibly respond to potential changes in the supply prices and also adequately respond to and minimize potential financing risk as much as possible.

Lastly, there may be a risk of technological change in speed, such as the release of next-generation infrastructure at a faster rate than expected. As the growth rate increases,

The fourth potential risk factor is the speed at which the next generation technology actually transforms. So with the coming up of the next generation infrastructure as well as the speed accelerating, basically with a higher level of obsolescence, There could be the risk of the per unit computing price actually uptrending, but in step with the changes in the technology, we believe that the contractual arrangement with our customers would also undergo change.

Next question, please.

The following question will be presented by Min-joo Kang from Bernstein. Please go ahead with your question.

Hello, I'm Min-joo Kang from Bernstein. Thank you for the opportunity to ask a question. I have two questions. First of all, I was most curious about the question about the allocation of resources for the military. As Naver has recently announced the implementation of the Neocloud business and is actively working on it, I wonder if the priority ranking of commerce businesses can be adjusted in terms of the distribution of growth resources in the future. I wonder if the large data and traffic of Naver users and traffic collected through the commerce business will be able to act as a differentiated competition in terms of AI-related business in the long term. Secondly, it is related to the Neocloud business. I understand that the rate of success of the Naver platform margin has been affected by the price-to-value ratio recently due to the recent increase in GPU investment. Is there a possibility that some of the existing GPU investments will be postponed in the process of establishing SPC for the future Naver Cloud business? If so, I wonder if this will have an impact on the platform margin rate improvement. If not, I wonder if we can maintain the current level. And for the third question, I'm curious about the competitiveness of Naver compared to other AIDC peers in Korea. I'm curious about the competitiveness of Naver compared to other AIDC peers in Korea. And for the third question, I'm curious about the competitiveness of Naver compared to other AIDC peers in Korea. And for the third question, I'm curious about the competitiveness of Naver compared to other AIDC peers in Korea. And for the third question, I'm curious about the competitiveness of Naver compared to other AIDC peers in Korea. And for the third question, I'm curious about the competitiveness of Naver compared to other AIDC peers in Korea. Thank you for taking my question.

I am Kang Min-joo from Bernstein. So my first question has to do with your NEOcloud related business initiative as well as the asset allocation priority, whether it could undergo certain changes. Since you would have to focus quite a bit of your growth capital towards this now cloud business, would like to know whether that would change your priority that you've placed on your commerce business. And also, this extensive amount of data and traffic that you are able to drive under your commerce initiative, how would that work as a competitive edge supporting your AI business moving forward? Second question has to do with your now cloud business. I see that because of the investment and the depreciation that follows, The margin rate from Naver Platform had declined somewhat. With the setup of the SPV, would you be planning to carve out your and transfer your GPU to that SPV entity? And would that have an impact of improving the margin for Naver Platform business? Or do you think that the current level will continue onwards? Third question is, compared to your domestic peers in AIDC, you seem to be moving more quickly in terms of the build-out of the AIDC infrastructure and those relevant endeavors. So would like to know as to what are some of the customer sourcing approach as well as long-term agreements, or in terms of the pricing negotiations, do you believe that you will have an upper hand? My last question is also related to NailCloud. For you to be operating now cloud business, it seems like you would need a specialized talent and organization. So what are your plans regarding this?

This is the CEO responding to your question.

The AI Factory is basically the Neo-Cloud business that Naver is engaging in, and we are aligned with the current trend when it comes to the Neo-Cloud business, and we believe that this is a specialized B2B domain where we, Naver, will be able to leverage all of our capabilities that we've been building over the years in infrastructure as well as cloud.

As you can see from our quarterly results, commerce business is a very high growth segment for us. It is our core fundamental business that really is driving the growth of advertisement and core services. But those areas are B2C domains and Naver operates its resources, organization, and structure independently.

And so although there could be areas where we will be able to drive synergies across those two domains, this will not be shifting any priority away from our previous focus on commerce. And also on the commerce side, we are really considering various different investments to further bolster our logistical competitiveness. And just like the approach that we're taking for a Neo-Cloud business, we will also be using an asset-light approach.

The first question is about the profitability of the Naver platform.

The second question on NeoCloud, I take it that there are three parts to that question. First, having to do with enabled platform margin. Now, basically, that is impacted because of the infrastructure that Naver is using to serve its Naver services and because of the depreciation that arises out of that.

I don't think that the existing GPU assets that Naver has will be abandoned due to the development of this new neocloud business. Basically, since the neocloud business is developing a new business under the S-Lite structure, I don't think it will have much to do with the cost structure of the existing Naver platform.

And basically, with the NEOcloud business that we are pushing forward, this is not going to entail any carve-out of the GPU asset that Naver currently has. It is a new business for us, based upon which we are implementing our efforts, and NEOcloud, once again, is going to also take on an asset-light approach.

In the second question, you talked about the construction schedule. We may be quick, but in fact,

So, because of that, just going back to the previous part of the question, it's not going to have any significant impact or impact on the cost structure of the Naver platform. Now, in terms of the build-out planning, yes, we are fast, but there are also other characteristics to that.

We can expect the entry of various suppliers. Our speed and specific execution

There may be a number of multiple other service providers entering into this domain, but we are moving fast, and also we have a very concrete execution plan in place, so that will help us gain an upper hand in this market.

Lastly, in terms of personnel, technical personnel, and organization building, we are basically implementing a neocloud-like business In terms of the required technology, the talent, and the organization, we've already been engaging in a NeoCloud business internally within Naver through our full stack capabilities through our subsidiary, Naver Cloud.

So rather than extensively newly building out organization and acquiring new talent, we will be able to leverage our current resources.

Next question is from Lee, Jun-ho Lee.

The following question will be presented by Jun-ho Lee from Hana Securities. Please go ahead with your question.

Hello. Thank you for the question. I'm Jun-ho Lee from Hana Securities. I have two questions regarding the ARDC. First, please update us on the current schedule and how it's going. Secondly, from last year, we've been seeing news about the meeting of the Middle East Day Center and this year, the South America Day Center. Please explain the roadmap and business structure of the business. Thank you.

Thank you for taking my question. I am Lee Joo-no from Hana Securities. I have two questions regarding AIDC business. Can you give us an update as to what your capacity addition schedule looks like for KAK Sejong Data Center? And second, we've seen from news articles that you're engaged in discussions regarding data centers in Middle East and in Latin America. Can you also give us an update on the roadmap and the structure?

Our exact schedule is 55MB in the first half of 2027, 100MB at the end of 2027, and 200MB at the end of 2028. In terms of the timeline for AIDC, we've already previously communicated that by the first step of 27, the capacity will be 55 megawatt.

and by end of 2700 and by 2028 on a cumulative basis 200 mega. That is going to be our target and up until the 200 meg milestone we are going to move quickly into entering into the market and we have completely sourced and have sourced the external, the server rack space as much as possible for rental purposes.

In the case of each of the three, each of them is divided into two sub-districts. Currently, one sub-district is a building, but in the case of the amount in it, we are consuming about one-third of the amount, and we are conducting a project to expand the remaining two-thirds. On Gak Sejong Data Center, there are two sites. Basically, for one site, the capacity addition has been more or less complete. So we are able to cater to one-third of the required volume, and right now what we're doing is expanding on the remaining two-thirds.

Afterward, we will continuously expand our capacity to one giga scale, and for the remainder of the two sites, there is going to be also some greenfield development taking place in parallel.

For the Middle Eastern market, right now we're in the process of implementing a build-up for the data center, and we are looking to gain cloud-related partnership.

And once we have that determined, we will come back to you and make that additional communication. In the case of South America, we are in the process of making a very initial development.

As I mentioned earlier, our strategy is to make a very initial development

For Latin America, our discussions are only in the initial and early phase, and as I previously mentioned, as we move from 200 meg to 1 giga scale during the process, the domestic demand and greenfield is not going to be ample enough, so we definitely need global partners.

In particular, we are securing global partners who can strategically assess the cost of building greenfields, such as low electricity costs and the cost of building greenfields. In addition, we are cooperating with various partners that can meet the demand for Soborin AI, which has a strong point from this perspective.

And when it comes to the demand for training purposes, basically it's quite important to think about the electricity or the power-related cost, reasonably priced power-related cost, as well as the greenfield of build-out to relevant expenses and where actually we have a certain level of strength. And so we are in the process of looking and discussing with potential global partners and also solving AI demand is an important aspect as well. So we are in talks with various different potential partners to carry out those efforts.

In the end, the domestic demand that we mentioned up to 200 megawatts is a starting point and reference for the very specific AI factory business that we are already predicting. The part that we are judging as the volume of growth in the future is the entire global AI computing market,

Up until the 200 megawatt scale, domestic demand is going to play an important aspect and we've made the appropriate forecasting based on that. But when it comes to AI factory, therefore it's going to be an important starting point for us and it will work as an important reference point. However, the true demand lies in the global market when it comes to the sovereign AI-related demand and requirements. So it is quite important for us to make endeavors in that direction to gain that global AI computing market positioning.

The last question will be presented by Joon-yoon Kim from HSBC. Please go ahead with your question.

Thank you for the question. I have two questions. The first is, you explained a lot about AIDC. We recently had a lot of expectations about the AI infrastructure market, and at the same time, there is a concern that if the chip price drops, the GPU loan price will drop and the profitability of the AI data center may have a negative impact. I'm curious about what the business people think about this. Also, if the R&D direction grows like TPU, How do you see the risk of our business being tied to an expensive NVIDIA chip? When learning AI, even if the GPU is absolute, these days, when it is used as an AI agent, it becomes Churon, and Churon seems to be using other chips instead of GPU chips, so I'd like to ask. And the second question is about financial changes. AI data center business, Doonamu, Thank you for taking my question. My first one has to do with your AIDC business. There is building anticipation and expectation on AI infrastructure and business, but there's also at the same time some concern that

If the chip price starts to fall, whether that will have some negative impact on your AIDC business. And also, with the greater adoption of TPU chips based upon the ASICs architecture, if you are bogged down by the NVIDIA partnership, will that not have some negative impact? Because for training purposes, AI GPU is quite important, but we're moving into AI agentic phase. First of all,

In terms of price, through GPU financing, we will be able to supply the required GPU capacity with our investment in Brookfield. There will be a lead time gap between supply and demand for a certain period of time, but we expect that the actual purchase price will also be lowered if the house price drops at that time when continuous additional issuance occurs.

Regarding your question on chip pricing, as you know, we have taken out a GPU financing arrangement, and we will be able to source the needed GPU capacity on a timely basis through that arrangement with our investor, Brookfield. There will be a lead time gap between demand and supply. However, later on when there is a decline in the chip price, then that will also translate into a lower purchasing unit cost from our perspective.

If the chip price is lowered by half, we will double the computing infrastructure that can be purchased with the same amount of money in GPU financing. Since we do not have a structure to contract the total amount at the same time with the current price with 90 billion dollars already secured, I think the negative effect of the change in the chip price will be very limited.

What that means is if the chip price actually falls by 50%, basically under our GPU financing scheme, we will be able to purchase double the computing infrastructure with that amount of money. So that $9 billion is not a fixed basis upon which a volume purchase is fixed. So it's not going to be that because of the changes or the lowering or the decline in the chip price, That will not be having an impact on our bottom line. TPU 등 커스텀 매식 칩 관련해서는요, 저희가 출원 시장에서는 비엔비디아 칩 사용의 비중이 올라갈 수 있을 것으로는 예상을 하지만, 그렇다고 해서 모든 기업들이 다 다른 제조사 칩을 사용할 수 있는 것은 아니라고 생각합니다. Regarding the TPU, custom ASIC chips, in the inference market, it is possible that chips that is made by other companies other than NVIDIA, so non-NVIDIA chip usage may go up, but that does not mean that all of the companies can all resort to using a chip that's made by other manufacturers.

As already known, NVIDIA is easily accessible through the CUDA ecosystem, but if you use another chip, Because right now, all the enterprises can actually access NVIDIA because NVIDIA has put in place the CUDA ecosystem, so there's easy access.

But if an enterprise decides to use a chip from another supplier, then basically you have to, depending on the AI model, go through an optimization process. Only then would you be able to enjoy the best token output speed.

Most of all, in the phase 1 stage where we want to serve 200MW until 2028, it is difficult to predict that there will be a significant change in the competitiveness of the existing GPU.

And also, in particular, by 2028, where our objective is to serve that 20 meg, which is within the Phase 1, we believe that, and it's quite hard to believe, that there will be any significant changes to the competitive edge that current GPU actually has.

However, in the medium term, there may be changes in technology or trends, so we will continue to plan our business in a direction that will be flexible when there are changes in trends or ecological changes.

However, there may be sudden changes in the way the technology is deployed and the changes in the trends, so we will very closely monitor those changes. But I can assure you that we have put in place a business plan and design that will ensure that we can respond in a flexible manner.

Lastly, you gave us three examples related to cash flow. First, as I keep saying, the AIDC business is an SL-like business, so I don't think it will have much to do with cash flow.

You also asked about cash flow impact. AIDC, once again, because it's an asset-light business structure, will not have much of an impact on cash flow?

I think it's a plus factor that we can use cash flow as a connection point.

In terms of Toonamoo, the arrangement with Toonamoo, of course, we have to wait and see how it actually, you know, actually the end state of after it is closed. But rather than it having a negative impact on our cash flow, from a consolidation perspective, there will be an opportunity for us to make use of those cash and cash equivalent assets.

Yes, in terms of commerce, in the case of normal promotion, of course, we are expecting an increase in sales, so there is some link there. As for the logistics side, as our CEO has mentioned, our logistics investment approach is also based on asset-light concept. In terms of the margin guidance in the second half of the year, we will be further expanding investment support for NP Connect. So there may be some incremental expense and investment that will emerge.

However, so compared to the first half of the year, our operating profit and the profit per se may go down slightly. However, it will not be to the extent that it will impact our cash flow or our investment capabilities.

Yes, then due to time constraints, I would like to conclude the 2-minute performance announcement. If you have any other questions, please contact us at the Capital Market Room.

With that, we would like to now close the second quarter 2026 Neighbors Earnings Conference Call. If you have more questions, please feel free to call us at the Capital Markets Division. Once again, thank you for joining us and thank you always for your encouragement and support.