AiStaffo

Naver signs 2 trillion won AI computing deal with US inference firm

Stock photo: data center server racks
Photo: panumas nikhomkhai / Pexels

Naver, the South Korean internet company, said on 12 November 2026 that it had signed a computing services supply contract worth 2 trillion won (about $1.5 billion) with a US AI inference company. The deal, announced on its third-quarter earnings call, covers next-generation GPUs with initial supply scheduled for 2027. It is the first customer for Naver's AI factory business, which is being built with Nvidia and Brookfield. The customer has not been named. For businesses that automate work, the practical effect is indirect: it signals how much AI computing capacity will be sold in Korea and abroad from 2027, which can shape the price and availability of the AI tools they use. Nothing changes in day-to-day tools today.

In short

  • Naver signed a 2 trillion won (about $1.5 billion) computing supply contract with an unnamed US AI inference company, announced on 12 November 2026.
  • Supply starts in 2027, and the contract covers about 20 megawatts over roughly five years.
  • The contract is the first customer for the AI factory built with Nvidia and Brookfield, which plans to reach 200 megawatts by 2028.
  • For most businesses, the effect is on future AI pricing, availability and data location, not on tools they use today.
  • Ask AI vendors how compute costs, data location and model changes will be handled before signing.

What happened

According to the Korea Times, Naver signed a 2 trillion-won computing services supply contract with a U.S. artificial intelligence inference company, and it described the deal as securing an initial customer for its AI infrastructure business ahead of a commercial launch planned for the first half of next year. The announcement came on 12 November 2026 during the company's third-quarter earnings call, where CEO Choi Soo-yeon called the contract a sign of real market demand. The contract covers computing services using next-generation GPUs, and supply is scheduled to begin in 2027. The US customer is not named in the company's disclosure or in the coverage.

The currency figure needs a note. The Korea Times puts the value at about $1.5 billion, while the Seoul Economic Daily converts the same 2 trillion won to $1.4 billion. Both figures depend on the exchange rate used on the day. The won amount is the figure to use.

Naver's own reporting gives more shape to the deal. According to the Seoul Economic Daily, the contract covers computing services of about 20 megawatts and is worth 2 trillion won in total over five years, with revenue recognised once commercial service starts in the first half of 2027. Naver's chief financial officer, Kim Hee-cheol, said on the call that the 2 trillion won is total revenue over that period, not annual revenue, according to Digitaltoday.

The AI factory behind the contract

The AI factory is being built at Naver's GAK Sejong data center in Sejong, South Korea. Naver, Nvidia and Brookfield announced in July 2026 that the initial Nvidia DSX AI factory there would grow from 55 megawatts to 200 megawatts by 2028, according to a joint release published by Nvidia. The plan uses Nvidia's Vera Rubin and Blackwell platforms.

Naver's financing has moved forward since July. Naver said it had signed a computing infrastructure agreement with Brookfield worth $9 billion, according to The Herald Business. That agreement was the precondition for Nvidia's equity investment, which the Korean coverage describes as completed. Naver also said on the call that the capacity it has contracted so far exceeds the 2028 target it had disclosed, as reported by Digitaltoday.

Why it matters for businesses that automate work

Most businesses do not buy GPU capacity. They buy AI features through software, chat tools, document readers and automation platforms, and those tools run on compute like this. When a provider signs large multi-year capacity contracts, it is trying to secure the hardware that runs the models. That affects how much capacity is available for other buyers and how fast new models reach the market.

The deal also shows where the demand is coming from. Naver said it is in talks with more than 10 companies about further supply contracts, according to the Korea Times. The company's own release says the remaining capacity in its first phase is roughly 180 megawatts. If those contracts close, the Korean AI infrastructure market will have more large buyers competing for the same hardware in 2027 and 2028.

What changes in practice for a business that automates work

For most readers, nothing changes this quarter. The contract does not alter Naver's consumer services, and it does not create a new product for business customers. Its direct relevance is to suppliers and buyers of AI computing, not to a company running invoices or follow-ups on a standard software plan.

There are three practical checks worth making. First, if a provider's pricing rests on compute costs, ask how it passes on changes in those costs over the contract term. Second, if your data is processed in Korea or abroad, ask the vendor where the models run and where records are stored, and check that answer against your own data rules. Third, if you plan to move a workflow to a new AI model in 2027, keep the process documented so the change does not require rebuilding it.

Naver has said it will build on its ecosystem and on what it calls the private security of its AI data, according to an earnings call transcript published by Investing.com. Those are the company's own statements. Buyers should verify security claims in writing, from the provider, rather than rely on them.

Limits and risks

The contract is not yet revenue. Naver's operating profit fell 20.9% year on year to 451.3 billion won in the third quarter, as AI investment rose, according to the Korea JoongAng Daily. Investing.com also reports that the AI factory is expected to turn operating-profit positive in 2027, while near-term earnings remain under pressure. Infrastructure timelines can also slip. A capacity plan that is on paper in 2026 may arrive late or in a different shape.

What to watch next

  • Whether Naver names the US customer or the contract is confirmed in its filings.
  • Whether the commercial service starts in the first half of 2027, as planned.
  • Whether the 200-megawatt target by 2028 is kept as construction proceeds.
  • Whether Naver closes further contracts with the more than 10 companies it is talking to.
  • Whether Nvidia's next-generation platforms reach the Sejong site on the stated schedule.

How AiStaffo would automate this

For a business that runs invoices, payment follow-ups or reconciliations through an AI tool, the first step is a map of which steps depend on which model and which data location. AiStaffo would connect the accounts, inbox and document store that the workflow already uses, then set the AI worker to draft, match and chase items on a fixed schedule. The worker runs the routine steps automatically, such as extracting invoice fields or sending reminders from approved templates. The owner still approves exceptions, sets the payment terms and decides which model and vendor are acceptable. Start with a free automation audit, which can be booked through the Book a free automation audit button.

Questions people ask

How much is Naver's AI computing contract worth?
The contract is worth 2 trillion won, which the Korea Times puts at about $1.5 billion. The Seoul Economic Daily converts the same amount to $1.4 billion, so the exchange rate used affects the dollar figure. The Seoul Economic Daily also reports that the value is spread over five years.
Who is Naver's customer for the AI factory contract?
Naver has described the buyer only as a US AI inference company. The coverage does not name the company, and the customer has not been identified in Naver's disclosure. Any name would need to come from Naver or an official filing.
When does Naver's AI factory start supplying computing services?
The contract's initial supply is scheduled for 2027. The Korea Times reports that commercial services are planned to begin in the first half of 2027. Naver said revenue would be recognised once that service begins.
What is the AI factory at Naver's Sejong data center?
The AI factory is an Nvidia-based data centre run by Naver at its GAK Sejong site. Naver, Nvidia and Brookfield announced in July 2026 that the initial 55 megawatt build would grow to 200 megawatts by 2028. The plan uses Nvidia's Vera Rubin and Blackwell platforms.
Does this deal change the AI tools my business uses today?
Not directly. The contract is for computing capacity that supply starts in 2027, so it does not add a feature to any current product. It may affect future pricing and availability of AI services. Businesses should check with their vendors how they handle model changes and data location.

Book a free automation audit

Thirty minutes. We look at one process you run every week and tell you exactly what an AI worker would take off your desk, and what it would not.

We use these details only to reply to this request and prepare your audit, and keep them for up to two years. We send no marketing unless you tick the box above. Privacy policy

ai infrastructurenvidiaai investmentai chipsai models