Supermicro CEO Charles Liang has predicted that “the AI revolution could be bigger than the industrial revolution,” and Supermicro’s phenomenal growth over the last year seems to support this prediction. Supermicro’s revenue grew 200% year over year in the third quarter and is expected to grow another 152% by the end of the year. At VB Transform 2024, Liang spoke about the surge in profitability the company has experienced, trends that are supporting investor confidence that the AI revolution and profitability for well-positioned companies will continue.
And Supermicro is in a better position than most: AI relies on GPUs made by companies like Nvidia (mostly Nvidia), but without Supermicro, you have no GPUs. Supermicro ships clusters and racks of networked, cooled, pooled AI infrastructure (including GPUs) to meet the ongoing demand for best-in-class building block solutions that let customers customize every component of the platform.
“As a Silicon Valley-based company, we have the ability to work with all the technology leaders, such as Nvidia, Intel, AMD, Broadcom, etc., so if a product needs to be developed, our engineering team will work with their teams from early morning to late night,” Liang said. “We also work closely with all the big software companies and all our key customers to brainstorm and design the most optimized platform for them based on our building block solutions, with the best time to market.”
Preparing for an AI future
“We believe this AI boom is just beginning,” Liang said, “so we’re building the capabilities to support entire industries around the world.”
In Silicon Valley alone, the company manufactures and ships 4,000 racks per month, about 1,000 of which offer liquid cooling solutions, and is currently ramping up production to ship about 2,000 liquid-cooled racks per month. Liang added that liquid cooling will be essential in the future as power constraints become more severe. Currently, data centers consume huge amounts of power, and power companies are struggling to keep up. But liquid cooling is the answer, Liang said. In fact, Elon Musk uses Supermicro’s liquid cooling solutions for xAI.
“Liquid cooling enables our data center solutions to reduce energy consumption by 30 to 40 percent, which not only saves our customers energy costs, but also allows them to deploy 30 percent more computing power with the same power budget,” he explained. Our goal is to have at least 20 to 30 percent of our total data center deployments liquid cooled within the next 12 months.”
