"Only 7% of corporations worldwide that adopted artificial intelligence (AI) are using graphics processing units (GPUs) properly without waste. Utilization is low because they have not properly built AI infrastructure, but NHN Cloud will raise corporations' GPU efficiency through the next-generation AI full-stack platform 'NHN Factory X,' and based on this, we will lift the share of AI in total revenue to around 50% next year."
Chief Executive Kim Dong-hoon of NHN Cloud said on the 26th at a press briefing at The Plaza Seoul in Jung-gu, Seoul, "The core of the AI hegemony race is shifting from AI models to AI infrastructure." Kim said, "We are now in an era where it is crucial to produce and use tokens (the minimum unit of AI computation) as cheaply as possible," adding, "In other words, whether AI infrastructure is run solidly and efficiently determines the success or failure of AI."
NHN Cloud on the day unveiled 'NHN Factory X,' a new platform that integrates AI infrastructure, platforms, and services to support corporations' AX (AI transformation), nurturing its AI business as a core growth engine for the future. Based on an execution environment that spans three stages—from securing GPUs, to efficient platform operation, to running AI agents—it connects corporations' AI projects, which are stuck at the proof-of-concept (PoC) stage, to real businesses.
Kim said, "'Factory X' will be the engine for the next growth leap," adding, "The AI business accounted for 13% of total revenue last year, but it is expected to rise to 38% this year and reach 50% next year." The plan is to aggressively target the AI infrastructure market, which is expected to grow rapidly by an average of 20% annually to about $394.4 billion (about 594 trillion won) by 2030.
NHN Cloud cited its own infrastructure capabilities as the strength of 'Factory X.' NHN Cloud is currently running Nvidia H100 GPUs and a domestic Neural Processing Unit (NPU) at the Gwangju National AI Data Center, and at 'Factory X Seoul,' an AI-dedicated data center equipped with 7,656 ultra-high-performance B200s, it operates Korea's first exascale AI cluster with a total scale of 27.4 EF (exaflops).
Chief Operating Officer Kang Min-su (CIO) said, "AI platforms and services ultimately cannot exist without GPU infrastructure," adding, "What differentiates Factory X is NHN Cloud's 100% liquid-cooling GPU cooling system, GPU clustering technology, and internalized operating capabilities secured through 10 years of AI data center operations."
NHN Cloud commercialized the nation's largest single GPU clustering of 4,080 units and said it cut the annual GPU failure rate by about threefold compared with air cooling through a 100% liquid-cooling GPU system that controls a high-density environment of 75 kW per rack. It also internalized operating capabilities by directly handling the entire process from data center design, GPU cluster build-out, and network and storage optimization to incident response.
At the next platform stage, it supports domestic corporations in maximizing the use of their secured GPU resources. Kim said, "About 70% of corporations that have secured GPUs show peak-time GPU utilization below 70%," and introduced 'GPU Live,' an integrated management platform that helps maximize the use of GPU asset without waste.
Chief Technology Officer Kim Tae-hyung (CTO) said, "$2.1 per hour is wasted in idle mode for each H100," adding, "If we assume a customer operating an on-premises cluster is using only 50% of GPUs, about 14 billion won is wasted every year, but using 'GPU Live' can cut average fees by about half and process 1.4 times more workloads in the same time."
In the final service stage, it plans to help apply AI across corporations' actual operations through the AI agent execution service 'Project X.' 'Project X,' launching in the second half of this year, supports even non-developers in easily designing custom AI agents based on natural language. It features 24/7 stable task automation by linking with in-house systems and data while minimizing security and expense burdens.
NHN Cloud said it will use the launch of 'Factory X' as a starting point to grow beyond a simple cloud provider into a "national AI infrastructure corporation" that reliably supports corporations' AI transformation. Its strategy is to maintain double-digit growth by responding to diverse market demand, from general-purpose GPUaaS (GPU as a service) to custom AI infrastructure.
Kim said, "Factory X will fully back the great AI transformation of the public and private sectors," adding, "This year we will turn to the black on an annual basis."