Presented by Supermicro
Three decades ago, CERN introduced the world wide web to the public, and Supermicro CEO Charles Liang founded a motherboard and power supply company in Silicon Valley. Initially focused on PCs, the company quickly shifted its technological expertise to the burgeoning server market, experiencing steady growth ever since. This convergence of timing, technology, and location has propelled Supermicro to new heights, achieving record revenue and 50% year-over-year growth in fiscal year 2023, driven by the surging demand for generative AI across various industries.
Recently, Liang spoke with Editor-in-Chief Matt Marshall at Supermicro's San Jose headquarters about the factors fueling the company's remarkable trajectory. “Every year, we see growth of 40% or more, with profitability more than doubling thanks to generative AI opportunities,” Liang stated. “We provide system designs and rack plug-and-play solutions that allow customers to simply connect a power and data cable to get started.”
Supermicro is a leading provider of end-to-end, application-optimized building-block solutions, delivering comprehensive rack-level system integration and plug-and-play options that enhance value for organizations of all sizes.
The Building Blocks of Success
“Thirty years ago, when I designed our product using building-block architecture, I could not have predicted its current power,” Liang reflects. “It’s been 30 years of continuous improvement. Our standardized building-block solution allows us to easily customize components for different customer applications, simplifying optimization for cost, performance, maintenance, inventory, and time to market.”
This modular approach is crucial, especially during a time when supply chain disruptions have impacted many technology companies. The ability to replace hard-to-find components with readily available parts ensures Supermicro can swiftly adapt to technological advances and meet market demands.
For example, the company is enhancing the performance, scalability, and reliability of its rack-scale AI solutions to support the computational needs of generative AI, large language model (LLM) training, and high-performance computing (HPC) applications. Recently, Supermicro announced support for the NVIDIA HGX H200, which features H200 Tensor Core GPUs, significantly improving capacity and bandwidth compared to its predecessor, the H100.
“Our only limitation is market size,” Liang asserts. “As generative AI and intelligent edge solutions evolve, we anticipate continued growth.”
Innovative Green Technology
Data centers currently consume up to 3% of global energy, projected to rise to as much as 8% by 2030. Supermicro prioritizes energy efficiency in its design philosophy. Servers are engineered for lower power consumption without sacrificing performance, and the modular architecture allows for upgrades of individual subsystems without replacing the entire unit. With many businesses opting for free air cooling, servers are optimized for maximum airflow while also accommodating liquid cooling systems, reducing fan speeds and the reliance on high-powered HVAC systems.
Supermicro recently launched the industry’s most compact high-performance GPU server: the NVIDIA HGX H100 8-GPU system, featuring a proprietary liquid cooling solution. This innovation allows data center operators to minimize footprint and energy costs while maximizing AI training capabilities.
“This liquid-cooled server doubles the density of our rack-scale AI solutions and enhances energy efficiency,” Liang explains. “It can save customers between 10% and 30% in power consumption.”
Strategic Location Advantage
Supermicro's building-block architecture and rack plug-and-play strategy provide a competitive edge. While others can produce similar systems, Supermicro's proximity to industry leaders like NVIDIA, Intel, and AMD enhances collaboration throughout the design, validation, and deployment phases.
“We are uniquely positioned close to key partners and clients, fostering collaboration from design to deployment, which accelerates innovation and time to market,” Liang says. Supermicro’s manufacturing capabilities allow for rapid delivery of fully populated server racks to global customers within a week.
“We’re in a hundred-meter dash,” Liang concludes. “While anyone can run, we have trained longer and run faster.”
Don’t miss the video interview between Matt Marshall and Supermicro CEO Charles Liang here.
Ranking
