[AI Pick] Naver, with Nvidia's backing, to leap as an AI infrastructure company?

[AI Pick] Naver, with Nvidia's backing, to leap as an AI infrastructure company?

55MW operation in 2027…to expand to GW-class AI factory in the future

Leveraging Brookfield's capital to lower investment burden and target profitability

Naver has unveiled a detailed blueprint for its AI Factory business, which it is pursuing with Nvidia and Brookfield.

The plan is to start with a 55-megawatt (MW) AI data center in the first half of 2027, expand to 200MW in 2028, and ultimately to a gigawatt (GW) class infrastructure in the long term.

Furthermore, the strategy is to grow the global AI infrastructure business while minimizing asset burden, based on Nvidia's $1 billion equity investment and Brookfield's infrastructure investment.

Naver CEO Choi Soo-yeon explained the AI Factory construction schedule, investment structure, and revenue model during the Q2 earnings conference call on the 7th, stating, "We will approach the market with a high-value-added business that combines GPU computing and cloud model operation services."

◇ Nvidia's $1 billion investment…AI Factory pushed with asset light strategy

The most eye-catching aspect of this collaboration is Nvidia's decision to invest approximately $1 billion (about 1.4 trillion won) by participating in Naver's third-party capital increase.

Naver CEO Choi Soo-yeon explained during the conference call, "To further solidify this partnership, Nvidia has decided to participate in our third-party capital increase worth approximately $1 billion, the first such instance in Asia." She added, "This signifies that we have evolved beyond simple technical cooperation into a long-term partnership where both companies share business performance and risks."

Additionally, the $9 billion (about 12.6 trillion won) infrastructure construction fund will be secured through the formation of a Special Purpose Vehicle (SPV) with Brookfield, a global asset management firm.

Naver has selected Brookfield as the exclusive preferred bidder and is in the process of signing a formal contract.

CEO Choi said, "Nvidia provides equipment and ecosystem, capital partners provide infrastructure investment, and we are responsible for platform operation and customer services. This structure has laid the foundation for rapidly expanding the business while minimizing initial capital burden."

CEO Choi further emphasized, "Based on our full-stack capabilities, from infrastructure to model services, and our proven sovereign AI experience both domestically and internationally, we are approaching this market with a high-value-added business that combines GPU computing and cloud model operation services, going beyond simple computing rentals."

This approach is interpreted as minimizing initial risks by reducing Naver's significant financial burden.

◇ Increasing demand for AI computing…competing with full-stack and sovereign AI

Currently, the global AI data center market is experiencing severe bottlenecks for new entrants due to large-scale power restrictions, difficulties in securing land, and issues with cooling facilities and permits.

CEO Choi compared, "Many operators who are just starting data center construction are facing bottlenecks due to labor shortages, power supply constraints, semiconductor shortages, and local community opposition to new construction. In contrast, we already have cooperation with Nvidia, partnerships for prepared power and land capital, and full-stack operational capabilities accumulated over a long period."

CEO Choi also addressed recent concerns in some market segments about a potential decrease in AI computing demand.

CEO Choi emphasized, "While there are concerns that computing usage might decrease with the emergence of low-cost, high-efficiency models, we believe that as unit computing costs decrease, AI applications, inference, and agents will spread, thereby driving overall computing demand."

Naver is leveraging its existing data center infrastructure and its proven sovereign AI references for clients like Korea Hydro & Nuclear Power, the Bank of Korea, and Korea Aerospace Industries (KAI) as strengths.

CEO Choi added, "While the types of AI required by each industry differ – process innovation in manufacturing sites, closed-network AI infrastructure in defense, AI transformation in public administration, and data security in finance – the common demand for reliable infrastructure and technology partners is growing."

◇ "AI DC margin rate over 20%"…confident in profitability and risk management

The success or failure of the AI Data Center (DC) business depends on chip price volatility, technological obsolescence, and the payback period for initial investment funds.

Regarding this, Kim Hee-cheol, Naver CFO (Chief Financial Officer), stated, "We have conducted extensive internal reviews regarding AI data center business risks and examined about four types of risks: demand, execution, finance, and technological changes."

Naver plans to establish a wholly-owned subsidiary, an operating company (OpCo), to exclusively manage the AI Factory business jointly pursued with Nvidia and Brookfield.

CFO Kim said, "The expected profitability of OpCo will vary depending on its maturity. While the margin rate might start a bit lower initially, as the business matures, we fundamentally expect a margin rate of at least double digits, eventually exceeding 20%."

Regarding chip price declines, CFO Kim explained, "If chip prices halve, the computing infrastructure that can be purchased for the same amount in GPU financing doubles. Since it's not a structure where we contract the total quantity at current prices all at once with the secured $9 billion, the negative impact of chip price changes on AI data center profitability will be very limited."

As Naver's AI Factory business aims for its first operation in the first half of 2027, the key tasks for business success will be to build the planned infrastructure without setbacks and to secure stable clients based on the Nvidia ecosystem to prove profitability.