고성능 AI인프라 구축에 각 기업들이 사활을 걸고 있는 가운데, 에너지 절감·비용효율·유연성과 빠른 배포 등이 서버 인프라 구축에 핵심 과제로 떠오르고 있다.

Charles Liang, CEO of Supermicro, announces at Computex 2023 (Photo: Supermicro)
Supermicro Supports Data Center to Edge Rack Scale Design and Manufacturing
Charlie Liang, CEO: “Corporate customers build their own clouds and data centers”
As each company is risking its life to build high-performance AI infrastructure, energy saving, cost efficiency, flexibility, and rapid deployment are emerging as key tasks in building server infrastructure.
At the recent Computex 2023, Super Micro Computer (hereinafter referred to as Supermicro) presented a wide range of cutting-edge server and storage solutions for modern IT workloads, and explained the latest server infrastructure solutions and market trends.
Supermicro’s Building Block Server approach is powered by the latest technologies from Intel, AMD, and NVIDIA, and purpose-built servers provide solutions for a wide range of AI, cloud, and 5G workloads from data centers to edge devices.
Supermicro will showcase notable products at Computex 2023, including: Rack-scale liquid cooling solutions, GPU-optimized systems, SuperBlade, Hyper, CloudDC, GrandTwin, SuperEdge, Petascale storage, and liquid-cooled AI development platforms.He cited platforms, etc.
While the server market has been struggling and slowing down due to the overall economic downturn this year, it was mentioned that strong growth is expected in the remaining second half and the forecast through order demand in 2024. Charles Liang, CEO of Supermicro, added, “Many enterprises are deploying their own clouds and data centers,” and “We expect to expand our customer base as telecommunications companies also transform into part of the cloud infrastructure.”
“Supermicro has expanded its production capacity to meet the rapidly growing demand for large-scale, high-performance AI infrastructure and cloud data centers,” said Charles Liang, CEO. “The result is advanced systems integrated into a turnkey total rack solution.”
He highlighted “the broadest portfolio of solutions, from the most powerful AI servers with up to eight NVIDIA H100 HGX GPUs to small edge servers that must operate in harsh environments.” Supermicro’s product lineup includes a new liquid cooling solution that recently reduced data center power consumption and improved performance.
Additionally, Supermicro will soon be launching a new product line based on the NVIDIA Grace CPU superchip. The new servers will feature dual CPUs with 144 cores each, delivering up to 900GB/s of data per second, to support ultra-responsive AI applications and applications that require ultra-low latency response. The system is expected to reduce energy consumption of cloud-native workloads and next-generation AI applications with CPUs running at 500W TDP.
As AI applications proliferate, the demand for high-performance AI design servers is increasing, and system providers face new challenges to integrate the latest CPUs and GPUs. The top-of-the-line Supermicro GPU server integrates dual CPUs and up to eight NVIDIA HGX H100 GPUs, with a liquid-cooling option to support operational cost efficiencies.
“NVIDIA is working closely with Supermicro to innovate new server designs that meet the most demanding customer requirements,” said Ian Buck, vice president of Hyperscale and HPC at NVIDIA. “With Supermicro servers based on the upcoming NVIDIA Grace CPU Superchip and the globally acclaimed H100 GPU, we are collaborating to bring AI to a broad range of markets and applications.”