跳到正文
The Elec· Hyun-Seon, Park·· 5 小時前AI 評分42

Clunix 與 NeoAI Cloud 合作,為浦項 AI 數據中心建設整合 AI 與 HPC 的運作環境

Clunix, NeoAI Cloud Partner on Integrated AI and HPC Operations at Pohang Data Center

AI 導讀

Clunix 與 NeoAI Cloud 簽署合作備忘錄,將在浦項 AI 數據中心建立整合 AI 與 HPC 的運作環境。Clunix 將以 NovaTier 配置裸機 GPU 叢集、統籌 AI 與 HPC 工作負載排程,並支援 NVL72 等機架級系統;中心首期規劃 40 MW,可容納約 10,000 張 Nvidia 新一代 Vera Rubin GPU。

正文

NeoAI Cloud CEO Cho Yong-wan (left) and Clunix CEO Seo Jin-woo pose for a photograph after signing the agreement.

NeoAI Cloud CEO Cho Yong-wan (left) and Clunix CEO Seo Jin-woo pose for a photograph after signing the agreement.

NeoAI Cloud CEO Cho Yong-wan (left) and Clunix CEO Seo Jin-woo pose for a photograph after signing the agreement.
NeoAI Cloud CEO Cho Yong-wan (left) and Clunix CEO Seo Jin-woo pose for a photograph after signing the agreement.

Clunix, a provider of artificial intelligence (AI) and high-performance computing (HPC) infrastructure solutions, has signed a memorandum of understanding (MOU) with AI cloud company NeoAI Cloud to develop an integrated AI and HPC operating environment at the Pohang AI Data Center.

Under the agreement, Clunix will integrate its AI and HPC management technology with the next-generation GPU infrastructure being built by NeoAI Cloud. The companies said they will collaborate from the infrastructure design stage to ensure that GPU resources are fully utilized for AI training and inference as well as HPC workloads in manufacturing and research.

NeoAI Cloud’s Pohang AI Data Center is being developed as a fully liquid-cooled, AI-dedicated facility with a total project cost of 600 billion won. The first phase targets a facility capacity of 40 megawatts (MW) and accommodation for approximately 10,000 Nvidia next-generation Vera Rubin GPUs, using a high-density design exceeding 150 kilowatts per rack. Completion is scheduled for November 2027. NHN Cloud has pre-leased half of the available capacity, or 16 MW, and will jointly operate the facility.

NeoAI Cloud will oversee data center design and construction, infrastructure operations including power and cooling, and the provision of computing resources to tenants. Clunix will use its AI orchestration platform, NovaTier, to establish an operating environment for bare-metal GPU cluster provisioning, integrated AI and HPC workload scheduling, and rack-scale systems such as NVL72. It will also provide technical support for the operating environment and develop technologies that meet agreed requirements.

The partnership will focus particularly on efficiently allocating large-scale GPU resources and minimizing idle capacity. Clunix emphasized that it will draw on its expertise in operating large GPU clusters to improve computing efficiency and operational stability.

The companies also plan to leverage Clunix’s HPC customer base in manufacturing and research to attract computing demand from Pohang’s steel, materials and secondary battery industries, as well as local research institutions, to the data center. They will also cooperate on government-led national AI strategic initiatives and large-scale projects.

來源:The Elec · thelec.net