Korean AI firms leverage ultra-high-performance AI computing resources with edge cloud provided by Gcore – set to transform the rapidly growing Korean AI market.
Gcore has announced the launch of its first AI public cloud service in Korea – powered by NVIDIA’s H100.
The Gcore data center will be equipped with 40 NVIDIA H100 servers, which are recognized as the most effective GPUs available for AI training – that’s 320 GPUs.
Until now, Korean AI companies have faced significant challenges in acquiring high-performance GPUs, which are essential for developing and training AI models.
Now, Korean AI specialists no longer must wait weeks to acquire large-scale GPUs or utilize resources in other countries but can easily access ultra-high-performance AI computing resources through Gcore’s edge cloud.
Gcore’s H100-based AI public cloud service, which is the first of its kind in Korea, is to provide powerful computing power by organizing servers equipped with eight SMX5-type H100 GPUs into a large-scale cluster and by configuring all connections between GPUs, which have the greatest impact on AI learning performance.
With InfiniBand NDR (400Gbps), the bandwidth per server reaches 3.2Tbps.
In addition, the non-blocking cluster managed by Unified Fabric Manager (UFM) is configured with hot and cold spares to ensure that the quality of service meets the standards of GDPR and 2TB of memory and 112 CPU cores for each server provide a large data processing and high-performance computing work environment.
The power of 12KW per server, fast network and secure center design enable the best service implementation.
In preparing for the service, Gcore worked closely with NHN Cloud, a leading cloud company in Korea, from selecting a data center to configuring the service to reflect the needs of Korean customers.
“Korea has tremendous growth potential in AI technologies, including AI semiconductors, systems, and services, and we look forward to helping Korean companies take a leading position in the global AI market by leveraging Gcore’s data centers located within the country’s high-speed telecommunications network, as well as Gcore’s global edge network, which is interconnected with ultra-low latency networks around the world,” said HyunYong Jung, Country Manager, Gcore Korea.
DongHoon Kim, CEO, NHN Cloud, said: “As the paradigm of the AI infrastructure market in Korea and globally is rapidly changing, NHN Cloud is building the largest GPU cluster in Korea with more than 1,000 NVIDIA H100s in our data centers, including the Gwangju National AI Data center and Pangyo Data center (NCC1).
“Gcore is a company with strong and secure cloud and edge AI know-how based on more than 160 PoPs around the world and by quickly introducing NVIDIA’s latest GPU, the H100, to Korea, we are setting the stage for a new phase in the Korean AI market.”
“Through our partnership with Gcore, NHN Cloud will strengthen the competitiveness of the Korean AI infrastructure market and actively respond to global expansion and market changes.”
Click below to share this article