Naver’s Proposed $10B AI Data Center Funding Explained
Naver’s proposed $10 billion AI infrastructure project combines a planned $1 billion investment from NVIDIA, up to $9 billion from Brookfield and additional funding from Naver. The capital is intended to expand the NVIDIA DSX AI factory at Naver’s GAK Sejong hyperscale data center in South Korea from an initial 55 megawatts to 200 megawatts by 2028, according to NVIDIA’s July 25 announcement.
The important qualification is that the headline does not mean Naver has already received $10 billion in cash. NVIDIA says its investment remains subject to customary closing conditions and Naver finalizing at least $9 billion of committed project financing. Brookfield’s current arrangement is described as a nonbinding term sheet for up to $9 billion, so investors should treat this as a proposed financing and expansion plan moving toward execution, not as fully deployed project capital.
What the $10 billion package actually includes
The announced structure has three participants with different roles. Brookfield is expected to act as the exclusive capital partner and fund up to $9 billion, NVIDIA plans to invest $1 billion into Naver, and Naver will fund the remaining amounts required for the project. The same allocation was reported by Yonhap’s July 25 account of the announcement.
That wording matters because “up to” and “nonbinding” describe limits and intentions, not an unconditional transfer. The final economic effect will depend on the definitive financing documents, the form of NVIDIA’s investment, drawdown conditions, Naver’s contribution and the obligations attached to construction and equipment purchases.
Investors should separate three stages that are often compressed into one news headline:
- Announcement: the partners disclose the intended project size and funding structure.
- Commitment: financing and investment agreements create enforceable obligations after conditions are satisfied.
- Deployment: capital is drawn as power, buildings, cooling, networking and AI systems are delivered.
As of July 25, 2026, the public announcement clearly establishes the first stage. The next disclosures will determine how much of the proposed package becomes legally committed and when the money can be used.
Why GAK Sejong is the project’s starting point
GAK Sejong gives the expansion an operating base rather than making it a proposal for an unidentified greenfield site. Naver’s June announcement described the facility as its core hyperscale data center hub and set out an initial 55-megawatt operating phase targeted for the first half of 2027, followed by a path toward 200 megawatts by 2028.
The earlier roadmap also connected the Korean facility with a broader global AI infrastructure strategy. Naver said it would combine its experience operating large in-house GPU clusters and hyperscale data centers with NVIDIA’s DSX platform. The company presented GAK Sejong as a launchpad for infrastructure serving Korea and, over time, customers in other regions, as detailed in Naver’s June partnership release.
An existing hyperscale location can reduce uncertainty around site selection and operating expertise, but it does not remove the main project risks. Electricity delivery, grid interconnection, cooling, construction sequencing, network capacity, server procurement and utilization still determine whether the expanded capacity becomes commercially productive.
What changes between 55 MW and 200 MW
The move from 55 megawatts to 200 megawatts is a substantial expansion of the planned AI factory, but power capacity should not be confused with sellable compute. The figure describes the planned facility scale; it does not by itself show how much power will be available to IT equipment, how much will be used by cooling and other systems, or how much capacity will be contracted by customers.
NVIDIA says the 200-megawatt facility is expected to include its Vera Rubin and Blackwell platforms. It also describes DSX as an end-to-end architecture spanning chips, systems, software, facilities and partner technologies. Those statements identify the announced technology framework, but they do not establish a current installed GPU count or guarantee a specific level of revenue.
Naver’s June release outlined a longer-term ambition to reach gigawatt-scale infrastructure. That objective should be kept separate from the current 200-megawatt project. A gigawatt roadmap is a strategic direction, not capacity that has already been financed or built through the July package.
NVIDIA’s role is strategic as well as financial
NVIDIA is not only supplying capital. The companies plan to build the expanded facility with the NVIDIA DSX platform, and the project is expected to provide production-scale compute for Korean and U.S. AI developers working on models, agents and AI-powered services.
This creates a vertically coordinated arrangement: Naver contributes cloud and data center operating expertise, NVIDIA contributes accelerated computing technology and software, and Brookfield contributes infrastructure capital. Such a structure can help align the parties around deployment, but it also links the project’s economics to NVIDIA’s product roadmap, supply availability and platform pricing.
NVIDIA’s planned investment may align its interests with Naver’s success, yet it should not be interpreted as a guarantee of customer demand, utilization or profitability. The announcement describes the intended partnership and investment conditions; it does not promise revenue, returns or minimum occupancy for the data center.
The same distinction applies to the term “sovereign AI.” In this project, it refers to building AI capacity in South Korea for domestic and international users, with Naver operating the infrastructure and NVIDIA technology forming a core part of the stack. It does not mean the facility is independent of foreign technology suppliers or insulated from commercial competition.
Why Brookfield’s participation matters to investors
Brookfield brings an infrastructure-investment perspective that differs from that of a semiconductor company or cloud operator. Data centers can be financed as long-duration assets when they have reliable power, disciplined construction, credible customers and contracts that support predictable cash flows.
The proposed structure could be attractive if Naver secures customers willing to pay for local production-scale AI capacity and if the facility reaches operation in stages that match capital deployment. The investment case becomes weaker if construction costs rise, power is delayed, hardware becomes uneconomic before it is fully utilized or customers can switch to competing capacity at lower prices.
Brookfield’s involvement is therefore evidence of institutional interest in the project, not proof that every commercial and technical hurdle has been cleared. Investors should look for details on whether the final funding consists of debt, equity, preferred instruments, structured financing or a combination. They should also examine which partner carries construction overruns, power-delivery risk and technology-related obligations.
This partnership model reflects a wider shift toward financing AI capacity through cooperation between operators, chip companies and infrastructure investors. A related example is the use of debt to expand AI cloud capacity without issuing new shares, as discussed in Nebius’s debt-funded expansion. The structures are different, but the investor question is similar: who provides the capital, who absorbs the risk and how quickly can capacity become revenue?
What the deal means for South Korea’s AI infrastructure
The announcement strengthens South Korea’s position in the competition to build domestic AI capacity. Naver and NVIDIA say the expanded infrastructure is intended to serve enterprises, industries, government organizations and AI cloud customers, while giving Korean and U.S. developers access to production-scale computing.
For businesses, local infrastructure can matter where latency, data governance, supply resilience or access to large compute clusters are important. It may also support the development of Korean-language models and services, although the announcement does not provide commercial forecasts or customer commitments that would quantify the opportunity.
Customers will still compare the facility with alternative cloud and data center capacity. Price, reliability, software compatibility, geographic reach, data-handling rules and access to the latest hardware will influence whether the project attracts external workloads rather than serving mainly Naver’s own demand.
The project therefore has both strategic and financial significance. Strategically, it connects national AI ambitions with a local operator and a global platform provider. Financially, it must demonstrate that the resulting capacity can earn an adequate return after power, equipment, financing and operating costs.
The risks investors should monitor
The first risk is execution. A 200-megawatt AI facility requires several systems to arrive in the correct sequence. A delay in electricity, cooling, networking, construction or server delivery can leave capital committed while productive capacity remains unavailable.
The second risk is financing certainty. NVIDIA’s planned investment depends on Naver securing the required committed financing, while Brookfield’s announced term sheet is nonbinding. Until final agreements are disclosed, the $10 billion figure should not be treated as an unconditional funding round.
The third risk is demand. A large facility can be strategically valuable and still generate weak returns if customer utilization trails the construction schedule. Relevant disclosures would include signed customer contracts, minimum usage commitments, the mix of internal and external workloads, pricing terms and the timing of revenue recognition.
The fourth risk is technology depreciation. The project is expected to use multiple NVIDIA platforms, including Vera Rubin and Blackwell. That may offer flexibility, but AI infrastructure can lose economic value when newer systems deliver materially better performance per watt or lower cost per token.
The fifth risk is Naver’s balance-sheet exposure. Investors need to understand Naver’s required contribution, guarantees, purchase commitments, lease obligations and any ownership or dilution effects from NVIDIA’s investment. The headline project size does not show how much financial risk remains with Naver.
A practical checklist for following the transaction
Investors can assess progress through public milestones rather than relying on the announced amount alone:
- Look for definitive agreements replacing Brookfield’s nonbinding term sheet.
- Check the instrument, closing conditions and ownership effect of NVIDIA’s planned $1 billion investment.
- Confirm whether Naver has secured at least $9 billion of committed project financing.
- Track construction and power-delivery milestones for the initial 55-megawatt phase.
- Watch for evidence of contracted utilization for the planned 200-megawatt facility.
- Review updated capital expenditure, funding and operating-cost guidance from Naver.
These milestones help distinguish a financed and operating project from a strategic announcement. They also show where the value creation must occur: not in the headline amount, but in the conversion of capital into reliable capacity and paying demand.
What to take from the announcement now
As of July 25, 2026, the Naver project is a proposed $10 billion AI infrastructure expansion with a clear partner structure and a defined capacity target. NVIDIA plans to invest $1 billion, Brookfield has announced a nonbinding term sheet for up to $9 billion, and Naver will fund remaining amounts. The facility is planned to grow from 55 megawatts to 200 megawatts by 2028 at GAK Sejong.
The sensible next step for investors is to follow final financing documents, power and construction delivery, and customer commitments in that order. If those milestones are achieved, GAK Sejong could become a significant regional AI compute hub. Until then, the project’s strategic importance is clearer than its final financing certainty or commercial return.
Also read:
Subscribe to our newsletter
Get the latest Web3, AI, and crypto news delivered straight to your inbox.