NAVER Signs AI Computing Contract Worth Nearly 2 Trillion Won for 20 Megawatts of Capacity

Key Facts

NAVER signed a computing services supply contract worth nearly 2 trillion won with a U.S. company specializing in AI inference, the company said in an announcement on November 12, 2026. The agreement secures an initial customer for the AI factory it is developing and turns part of its plan into contracted revenue. Capacity tied to the contract is about 20 megawatts, with revenue to be recognized gradually over the supply period. The contract runs for 4 to 5 years depending on the chip type, while coverage estimated revenue at about 2 trillion won over five years. NAVER said the contract represents a "computing services supply contract worth nearly 2 trillion won."

The contracted capacity amounts to 20 megawatts, while NAVER said the total data-center capacity it has secured through contracts exceeded its announced target for 2028. This refers to total secured capacity, not the new contract alone. The company plans to begin operating the facility at 55 megawatts in the first half of 2027, then raise capacity to 100 megawatts by the end of that year and 200 megawatts in 2028. Over the longer term, NAVER aims to build AI computing infrastructure with capacity of 1 gigawatt. The company said its ultimate target is about four times the maximum capacity of the GAK Sejong center.

NAVER is in talks with more than 10 prospective customers outside Asia, including major AI model developers, for the remaining 180 megawatts of capacity, according to Korea Herald. It is also in discussions with NVIDIA over a server order the company described as the largest in the Asian region. NAVER has not disclosed the order's size or when it will be signed. These talks are tied to securing graphics processing units and the infrastructure needed to operate them, as the company said sites with the liquid cooling required to run Vera Rubin are rare. It added that having this infrastructure could give it an "unrivaled competitive advantage" in securing chip volumes and data centers.

The GPUs supplied under the contract will combine B300 and Vera Rubin, according to CHOSUNBIZ. The chief executive said revenue from the contract would be spread over a period of 4 to 5 years, and that the AI factory business is expected to turn operating profit positive in 2027. The chief financial officer said the company expects the factory business to generate operating profit starting that year, with margins potentially exceeding double digits after operations are fully ramped up the following year. NAVER linked these prospects to its ability to secure suitable sites for liquid cooling and the required chip volumes. The contract's financial impact will therefore be determined over its operating period, not solely when it is signed.

The deal comes as NAVER spends on infrastructure, content and offline data to strengthen its competitiveness in AI. Revenue rose 10.7% year over year to 3.4724 trillion won in the third quarter of 2026, while operating profit fell 20.9% to 451.3 billion won. The company attributed the profit decline to increased spending related to those investments, while saying growth in commerce, its financial platform and global business-to-consumer commerce supported revenue. NAVER platform revenue was 1.9484 trillion won, financial platform revenue was 487.7 billion won and global growth business revenue was 1.0364 trillion won in the same quarter. These results provide context for the company's computing expansion, as spending weighs on current profitability.

The expansion plan is based on financing and infrastructure arrangements the company announced in recent months. On October 27, NAVER said it had signed a $9 billion agreement with Brookfield to secure computing infrastructure, with Brookfield to invest up to that amount in assets including GPUs and data centers. NAVER and NVIDIA announced an alliance in June to build a large-scale AI factory, with a phased launch planned at 55 megawatts in 2027. The new contract is part of that effort, as the company continues negotiations with NVIDIA over the server order and with prospective customers outside Asia over contracts. The first phase is scheduled to begin operating in the first half of 2027.