NAVER, NVIDIA and Brookfield plan $10bn Korea AI factory expansion

NAVER Corporation, NVIDIA Corporation and Brookfield Asset Management Ltd. are planning a $10 billion expansion of South Korea’s sovereign artificial intelligence infrastructure, transforming an initial 55-megawatt deployment into a 200-megawatt artificial intelligence factory by 2028. Brookfield Asset Management plans to provide up to $9 billion as the exclusive capital partner, NVIDIA Corporation intends to invest $1 billion, and NAVER Corporation will contribute the remaining financing required for the project. The expanded facility at NAVER Corporation’s GAK Sejong hyperscale data center is expected to contain approximately 100,000 NVIDIA graphics processing units and serve artificial intelligence developers, businesses and government customers in South Korea and the United States. The announcement gives NAVER Corporation a potentially powerful new growth platform, but it also commits the company and its partners to a capital-intensive infrastructure strategy whose returns will depend on compute utilization, power availability, customer demand and the pace of artificial intelligence model development.

The planned infrastructure will use NVIDIA Corporation’s DSX platform and is expected to incorporate the company’s Blackwell and Vera Rubin computing systems. NAVER Corporation intends to create a dedicated resource pool for emerging artificial intelligence companies, giving customers access to computing capacity, software and technical support needed to develop and deploy models, agents and artificial intelligence services at production scale.

The announcement significantly increases the scale of a 55-megawatt expansion disclosed in June 2026. The partners now plan to reach 200 megawatts by 2028, while NAVER Corporation continues to target a longer-term deployment of one gigawatt of sovereign artificial intelligence infrastructure. The jump demonstrates how rapidly artificial intelligence infrastructure plans can expand once technology suppliers, data center operators and large institutional investors agree on a financing structure.

NAVER Corporation shares closed 8.43% higher at KRW 225,000 on July 27 as investors responded to the NVIDIA Corporation investment and the potential expansion of the company’s artificial intelligence cloud business. Brookfield Asset Management shares gained approximately 2.6% during New York trading, while NVIDIA Corporation shares fell about 3.3%, suggesting the announcement had a much more direct valuation impact on NAVER Corporation than on its substantially larger partners.

How the $10 billion financing structure distributes risk among NAVER, NVIDIA and Brookfield

The financing structure places Brookfield Asset Management at the center of the infrastructure investment. NAVER Corporation said Brookfield Asset Management would provide up to $9 billion as the exclusive capital partner, while NVIDIA Corporation would contribute $1 billion and NAVER Corporation would finance the remaining amount required to complete the $10 billion project. NVIDIA Corporation’s investment remains subject to customary closing conditions.

Brookfield Asset Management’s participation reflects the growing convergence of data centers, power generation, digital infrastructure and private capital. The investment firm manages approximately $100 billion across the artificial intelligence infrastructure value chain, including data centers, computing systems, semiconductor manufacturing and dedicated power assets. It has also operated in South Korea since 2014 across infrastructure, property and energy investments.

For Brookfield Asset Management, the project could create long-duration infrastructure exposure supported by demand for artificial intelligence computing. Its opportunity is not limited to owning data center buildings. Artificial intelligence factories require power supply, cooling, networking, computing equipment and operational services, allowing an infrastructure investor to participate across several layers of the development.

The risk is that artificial intelligence computing equipment can become technologically outdated much faster than conventional infrastructure. A bridge or power transmission asset may remain economically useful for decades, while graphics processing units can lose competitiveness within a few product cycles. Brookfield Asset Management must therefore structure its investment so that long-lived data center and energy assets are not economically dependent on one generation of computing hardware.

NVIDIA Corporation’s planned $1 billion investment gives the semiconductor company a direct financial interest in NAVER Corporation’s artificial intelligence expansion. The project is also expected to deploy large quantities of NVIDIA Corporation hardware, making the investment both a strategic partnership and a mechanism for supporting future product demand.

That arrangement strengthens NVIDIA Corporation’s relationship with a major Asian cloud and internet platform while helping establish the DSX architecture as an operating standard for large artificial intelligence factories. NVIDIA Corporation’s financial return could therefore come through more than the value of its investment. It could also benefit from hardware sales, networking demand, software adoption and a deeper position in South Korea’s artificial intelligence ecosystem.

NAVER Corporation contributes the operating platform, customer relationships, local infrastructure knowledge and artificial intelligence models. The company reported 2025 revenue of KRW 12.04 trillion, equivalent to approximately $8.18 billion, across search, commerce, financial technology, cloud computing, artificial intelligence and robotics. The scale of the $10 billion project is therefore substantial even for South Korea’s largest internet company.

The partnership allows NAVER Corporation to pursue infrastructure growth without funding the entire project from its own balance sheet. That reduces the immediate financial burden, but it does not eliminate commercial risk. NAVER Corporation must attract enough paying customers to keep the facility utilized and generate returns for infrastructure investors whose capital is not being supplied as a charitable contribution to the artificial intelligence gods.

Why scaling from 55 to 200 megawatts changes South Korea’s sovereign AI strategy

The expansion from 55 megawatts to 200 megawatts is more than a simple capacity increase. It moves NAVER Corporation’s artificial intelligence factory toward a scale capable of supporting model training, inference and agent-based services for multiple large customers rather than functioning primarily as an internal corporate computing environment.

NAVER Corporation said the planned 200-megawatt facility would contain approximately 100,000 NVIDIA Corporation graphics processing units. The exact number and configuration may change as Blackwell and Vera Rubin systems are deployed, but the estimate provides a sense of the project’s intended computing density.

South Korea’s sovereign artificial intelligence strategy is intended to ensure that domestic companies and public institutions can access advanced computing infrastructure without relying entirely on foreign hyperscale cloud platforms. Sovereign infrastructure can give governments and enterprises greater control over data location, security standards, model development and industrial policy.

NAVER Corporation is well positioned to serve that role because it already operates search, commerce, mapping, financial technology, cloud and content platforms. The company also has experience operating large-scale digital infrastructure and developing its HyperCLOVA X family of artificial intelligence models.

The expanded facility is expected to serve customers in both South Korea and the United States. This international ambition matters because sovereign artificial intelligence projects can struggle financially if they depend only on domestic demand. Opening capacity to foreign artificial intelligence developers could improve utilization and turn a national infrastructure program into an export-oriented cloud business.

However, serving overseas customers may introduce tension between sovereignty and commercial scale. Governments typically support sovereign artificial intelligence infrastructure because they want sensitive workloads, strategic models and domestic data governed locally. International customers may require different compliance, security and pricing arrangements, forcing NAVER Corporation to separate protected national computing resources from globally marketed cloud capacity.

Power supply will also determine whether the project can progress from 200 megawatts toward one gigawatt. Artificial intelligence factories consume electricity continuously and require extensive cooling, transmission capacity and backup systems. Securing graphics processing units attracts headlines, but securing dependable power is what keeps the headline from becoming a very expensive collection of silent metal.

Brookfield Asset Management’s energy and infrastructure capabilities could become as important as its capital. The firm’s experience across power generation, infrastructure and data centers may help coordinate the physical systems required to support NAVER Corporation’s expansion.

The one-gigawatt target should still be treated as a long-term ambition rather than committed capacity. Reaching that scale would require customer demand, additional financing, power agreements, construction approvals and repeated technology deployments. The 200-megawatt phase will function as a critical test of whether NAVER Corporation can convert sovereign artificial intelligence policy into a commercially sustainable infrastructure business.

Can NAVER turn sovereign artificial intelligence infrastructure into profitable cloud growth?

NAVER Corporation’s commercial opportunity extends beyond renting graphics processing units. The company plans to combine computing capacity with proprietary data, artificial intelligence models, software tools and cloud operating expertise. That integrated model could produce stronger customer relationships and higher value per contract than a business focused only on providing raw computing capacity.

NAVER Corporation is advancing HyperCLOVA X using NVIDIA Corporation’s Nemotron 3 Ultra open models alongside its own data and training capabilities. It is also the first South Korean company to join the NVIDIA Nemotron Coalition, contributing to model development across pretraining, post-training and reinforcement learning.

The company plans to launch an artificial intelligence agent platform in South Korea during the second half of 2026 using NVIDIA Corporation’s Agent Toolkit software. NAVER Corporation is also developing a Seoul World Model that combines its street-view and spatial data with NVIDIA Corporation’s Cosmos world foundation models.

These projects show how the infrastructure could support applications across search, advertising, commerce, robotics, urban planning, physical artificial intelligence and enterprise automation. NAVER Corporation can use its own services as anchor workloads while selling excess or dedicated capacity to outside customers.

The strategy could improve NAVER Corporation’s position against global cloud providers, but competition will remain intense. Amazon Web Services, Microsoft Corporation and Google Cloud have enormous capital budgets, established enterprise sales channels and access to multiple types of artificial intelligence hardware. South Korean telecommunications and technology groups are also investing aggressively in domestic artificial intelligence infrastructure.

NAVER Corporation’s differentiation must therefore come from local language capabilities, domestic data, regulatory familiarity, model customization and integrated services. Competing only on computing price would expose the company to rapidly changing hardware economics and the purchasing power of much larger cloud providers.

The economic challenge is utilization. Data center profitability depends on keeping expensive equipment working for paying customers. Underutilized graphics processing units still depreciate, consume facility capacity and require technical support, while excessive demand can frustrate customers and limit the development of new services.

The dedicated resource pool for emerging artificial intelligence companies may help develop future customers, but smaller developers may not initially generate enough revenue to support infrastructure at this scale. NAVER Corporation will likely need a balanced mix of government contracts, large enterprise workloads, internal demand and international cloud customers.

The project’s success should therefore be measured by more than installed megawatts or graphics processing unit counts. Investors will need evidence of contracted capacity, customer retention, average utilization, cloud revenue growth and acceptable returns on invested capital. A larger artificial intelligence factory is strategically impressive, but a profitable one is much more useful.

What the NAVER, NVIDIA and Brookfield share reactions reveal about investor sentiment

NAVER Corporation shares rose 8.43% to KRW 225,000 on July 27, with trading volume above its recent average. The stock reached an intraday high of KRW 234,500 before surrendering part of the gain, indicating strong enthusiasm accompanied by some caution over financing and execution.

The rally suggests investors believe the partnership could change how NAVER Corporation is valued. The company has traditionally been viewed through its search, advertising, commerce, content and financial technology operations. A credible artificial intelligence infrastructure platform introduces the possibility of a new cloud growth engine supported by outside capital and NVIDIA Corporation technology.

The announcement also provides strategic validation. NVIDIA Corporation’s planned investment indicates that the semiconductor group sees NAVER Corporation as an important artificial intelligence platform and infrastructure partner. Brookfield Asset Management’s proposed funding adds institutional support for the physical development of the project.

Brookfield Asset Management shares rose approximately 2.6% to $47.56 during July 27 trading. The increase indicates that investors were comfortable with the firm’s latest artificial intelligence infrastructure commitment, although the movement cannot be attributed exclusively to the NAVER Corporation announcement.

NVIDIA Corporation shares traded around $200.10, down approximately 3.3% during the session. The decline should not be interpreted as a direct rejection of the NAVER Corporation investment because NVIDIA Corporation’s market capitalization, trading activity and investor sentiment are influenced by a much wider range of developments. The planned $1 billion investment is strategically meaningful but relatively modest compared with NVIDIA Corporation’s overall scale.

The difference in the stock reactions reflects the relative importance of the transaction to each company. For NAVER Corporation, the project could materially reshape future capital allocation and business positioning. For Brookfield Asset Management, it is another major infrastructure opportunity within a portfolio exceeding $1 trillion in assets under management. For NVIDIA Corporation, it is one component of a global strategy to encourage the construction of national and commercial artificial intelligence factories.

Sentiment toward NAVER Corporation is likely to remain constructive if the company discloses additional information about financing, ownership, customer commitments and deployment schedules. Investors may become more cautious if capital requirements rise, construction slips or artificial intelligence cloud demand fails to match the scale of the planned facility.

The announcement establishes an ambitious direction, but execution will determine whether the July 27 rally is sustained. NAVER Corporation has secured two powerful partners and a potential $10 billion financing framework. It must now prove that sovereign artificial intelligence infrastructure can become a durable source of revenue and returns rather than a technologically impressive national expense.

Key takeaways from the NAVER, NVIDIA and Brookfield artificial intelligence factory plan

• NAVER Corporation, NVIDIA Corporation and Brookfield Asset Management plan to finance a $10 billion sovereign artificial intelligence infrastructure project centered on NAVER Corporation’s GAK Sejong data center.
• Brookfield Asset Management is expected to provide up to $9 billion as the exclusive capital partner, while NVIDIA Corporation plans to invest $1 billion and NAVER Corporation will fund the remaining requirement.
• The initial deployment will expand from 55 megawatts to 200 megawatts by 2028, more than tripling the computing capacity announced in June 2026.
• NAVER Corporation said the expanded facility could contain approximately 100,000 NVIDIA Corporation graphics processing units using Blackwell and Vera Rubin systems.
• The companies intend to provide production-scale computing capacity to artificial intelligence developers, enterprises and government customers in South Korea and the United States.
• NAVER Corporation’s longer-term one-gigawatt goal could establish it as a major global artificial intelligence cloud operator, but that target remains dependent on power, financing and customer demand.
• NVIDIA Corporation’s investment supports future hardware and software adoption while giving the semiconductor company a deeper strategic position in South Korea’s sovereign artificial intelligence ecosystem.
• Brookfield Asset Management brings capital, data center and energy infrastructure expertise, although rapid hardware obsolescence creates a different risk profile from conventional long-life infrastructure.
• NAVER Corporation shares surged 8.43% after the announcement, reflecting investor expectations that artificial intelligence infrastructure could become a new growth engine beyond search, commerce and financial technology.
• Long-term value will depend on utilization, contracted customers, cloud revenue and returns on invested capital rather than the announced number of megawatts or graphics processing units alone.