NAVER's Operating Profit Falls 20.9% as AI Spending Bites

NAVER's operating profit fell 20.9% year over year to KRW 451.3 billion in the third quarter, even as revenue rose 10.7% to KRW 3.4724 trillion — a gap the Korean internet company attributes directly to its spending push into artificial intelligence. Announced on November 12, 2026, the results frame a company trading current margin for infrastructure it says will pay off from 2027. The operating margin narrowed to 13.0% from 18.2% a year earlier, according to the earnings presentation. That compression is the clearest measure of what NAVER's expansion is actually costing.

The announcement carried more than numbers. NAVER used the same date to unveil two structural bets: a logistics subsidiary and a standalone personal AI agent app scheduled for release by the end of 2026. Both share one logic — turning NAVER's existing traffic and search behavior into completed transactions rather than page views. Whether that conversion works while margins shrink is the operating question the quarter leaves open.

What did the quarter's numbers show?

The topline grew at a healthy clip, but the profitability picture moved in the opposite direction. Revenue of KRW 3.4724 trillion was up 10.7% year over year. Operating profit of KRW 451.3 billion was down 20.9%. NAVER attributed the profit decline to increased investment in infrastructure, content and offline data — three spending categories it says are intended to strengthen AI competitiveness.

Within the platform business, the numbers were steadier. NAVER Platform services revenue rose 13.1% to KRW 473.6 billion. Smartstore gross merchandise value — the value of transactions flowing through the platform — increased 6.4%, a figure the earnings presentation noted came despite the Chuseok holiday period. Repeat purchases through the NAVER Plus Store app and membership program supported that growth. In plain words: the commerce engine is still growing, just more slowly than the company's spending.

Why build a logistics subsidiary now?

The logistics plan rests on the argument that NAVER already owns the transactions a delivery network needs to serve. Chief executive Choi Soo-yeon said the logistics subsidiary will gradually strengthen delivery infrastructure around leased logistics centers. She described a self-reinforcing cycle: better delivery increases transactions, and the resulting volume improves efficiency for sellers, users and partners.

That is a promise with a built-in caveat. The cycle she describes depends on volume arriving fast enough to make leased logistics centers efficient, and NAVER did not disclose the subsidiary's capital budget, launch date or headcount. What the company did commit to is an expansion of N-Delivery coverage, its existing delivery service. The takeaway: NAVER wants to own more of the commerce stack, but it has so far shown the strategy, not the cost.

What is the personal AI agent supposed to do?

The agent app follows AI Briefing and AI Tab, NAVER's earlier AI service rollouts, and is scheduled to arrive by the end of 2026. NAVER said it is intended to extend services beyond recommendations and information delivery to actual purchases and visits, linking the agent's task handling to existing search and commerce activity. In other words, the agent is a transaction machine, not a chatbot.

Choi described an agent that can anticipate tasks users need handled or perform cumbersome work on their behalf. She framed the goals as attracting new user groups and creating monetization opportunities. The product plan sits alongside separate efforts to improve AI Tab revisit rates and conversion rates and to expand AI monetization within search itself. Notably absent from the announcement: any revenue target, pricing model or user projections for the agent. The agent's release date is firm; its business case is not yet on the record.

How much is AI investment hurting margins?

The margin numbers quantify the strain. Operating margin fell to 13.0% from 18.2% a year earlier — a drop of more than five percentage points in twelve months. Infrastructure expense rose 29.1% to KRW 282.3 billion, driven by computing assets for internal use and data-center rents, according to the presentation.

But AI is not the only line item growing. Marketing costs also increased, on shopping promotions and on Npay Connect terminal installations — point-of-sale hardware for NAVER's payment network. That detail matters for anyone reading the profit decline as a pure AI story: the quarterly pressure was not confined to AI infrastructure. Content, offline data and payment-terminal rollout all contributed. The honest reading is that NAVER is compressing margins on several fronts at once, with AI the largest and least immediately monetized.

When does the AI Factory start earning?

The biggest bet sits furthest out. NAVER said it has signed approximately KRW 2 trillion of computing-service supply contracts for the AI Factory business it is pursuing with Brookfield and NVIDIA. The contracts secure initial customers for capacity coming online in early 2027, and NAVER expects related revenue to begin entering results in the first half of 2027.

Choi said NAVER plans to pursue medium- and long-term contracts with large customers and expand its customer base among global AI companies, targeting both stability and profitability. There is an existing enterprise business to build on: enterprise revenue within NAVER's Global Opportunities segment grew 33.0% in the third quarter. The takeaway: the AI Factory has customers signed, but not a single won of recorded revenue until 2027.

What is at stake for NAVER?

The quarter sets up a timing problem common to every large internet company betting on AI infrastructure. NAVER must fund data centers, agent development and a logistics build-out out of a business whose profit is already shrinking — operating income down 20.9%, margin down more than five points — while the largest new revenue stream, the AI Factory, remains a 2027 item.

The company's own framing acknowledges the gap. The transaction-focused agent and the logistics subsidiary are designed to generate nearer-term monetization from users NAVER already has, bridging the period before AI Factory contracts convert into revenue. Whether that bridge holds depends on execution NAVER has not yet demonstrated: the agent has no disclosed monetization model, the logistics subsidiary has no disclosed budget, and the KRW 2 trillion in AI Factory contracts depends on capacity that does not come online until early 2027. NAVER has committed to the spending; the returns remain promises with dates attached.