What happened
Data Center Dynamics reports that South Korea’s Naver Cloud and Vietnam’s FPT signed a memorandum of understanding to expand their AI data-center businesses and regional market presence. The companies plan to combine FPT’s enterprise-focused “AI factory” model with access to Naver Cloud’s HyperClova X model and technical exchanges focused on GPU infrastructure and performance.
Data Center Dynamics reports that Naver Cloud and FPT signed an MoU to expand their respective AI data-center businesses and market presence across the Asia-Pacific region. The article says FPT will bring its enterprise “AI factory” model, while Naver Cloud will provide access to its HyperClova X AI model. The companies also plan regular technical exchange programs aimed at optimizing GPU infrastructure and performance.
According to the report, FPT operates AI data centers in Japan and Vietnam. DCD says the Vietnam facility uses Nvidia H100 GPUs and has a reported Linpack performance of 46.56 petaflops, while the Japan facility uses Nvidia H200 GPUs and has a reported performance of 49.85 petaflops. DCD also reports that FPT has four facilities in Hanoi and Ho Chi Minh City following the launch of the Fornix HCM02 Data Center in August 2025.
DCD reports that Naver Cloud operates public Ncloud regions in South Korea, Japan, Germany, Singapore, and the US West Coast, with additional regions planned in the US East Coast, Vietnam, Taiwan, and Thailand. The article also says Naver plans to expand its Sejong data center to 200MW. These infrastructure and performance details are reported by DCD and are not independently confirmed in the supplied source.
Source details: datacenterdynamics.com ↗
Why it matters
The agreement could broaden enterprise access to AI infrastructure and models across Asia-Pacific, while giving Naver Cloud and FPT a route to combine cloud, data-center, and implementation capabilities. Its practical significance remains prospective: the source describes an MoU, not a completed deployment, and provides no customer commitments, commercial terms, pricing, launch schedule, or independently verified performance results.
The partnership places an AI model, cloud infrastructure, and enterprise implementation capability within one regional expansion effort. If executed, that combination could reduce the work required for businesses in APAC to access model services and supporting through a local technology partner.
The announcement is consequential mainly as a strategic commitment rather than as proof of a new product or operating facility. The source does not identify signed customers, new data-center capacity attributable to the MoU, deployment dates, service-level commitments, pricing, or technical benchmarks from the planned optimization work.
The arrangement may also support regional alternatives for organizations seeking AI services outside the largest global cloud providers, but that implication is not a confirmed outcome. DCD’s report does not independently test HyperClova X, verify the cited Top500 rankings, or establish how widely the model will be offered.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
A route planner searches possible journeys using explicit rules. What does this illustrate about AI?
What to watch next
Watch for specific country launches, named customers, available services, pricing, and evidence that the partnership produces new capacity or measurable improvements. The companies’ next disclosures should also clarify whether HyperClova X access will be generally available, limited to selected enterprise projects, or bundled with FPT’s services.
The next material evidence would be a named production deployment, a new cloud region or AI data-center build tied to the agreement, or documented enterprise availability for HyperClova X. It is currently unknown whether access will be public, private, or restricted to FPT-led projects.
Commercial details are also unknown. DCD does not report pricing, minimum commitments, geographic restrictions, hardware allocation, performance guarantees, or a timetable for the technical exchange programs.
Readers should distinguish the companies’ stated plans from demonstrated results. Further reporting should establish whether the MoU creates incremental capacity and whether any claimed or infrastructure improvements are measured against a disclosed baseline.