Super Micro Computer(NASDAQ: SMCI), known as Supermicro, has experienced a meteoric 2,220% surge in its stock over the past three years. This remarkable ascent was propelled by the explosive growth of the artificial intelligence (AI) market, driving data center operators to acquire more of Supermicro’s high-performance AI servers.
A significant portion of Supermicro’s growth can be traced directly back to its partnership with Nvidia (NASDAQ: NVDA). Nvidia’s provision of high-end GPUs crucial for complex machine learning and AI tasks, collaborated closely with Supermicro to develop a new range of servers and workstations that fully supported its H100 GPUs. This collaboration allowed Supermicro to carve out a high-growth niche in the saturated market for pre-built servers.
Image source: Getty Images.
However, Supermicro’s reliance on Nvidia poses challenges, as evident during early 2023 when it struggled to secure a consistent supply of Nvidia’s GPUs. Additionally, its major competitors, Hewlett Packard Enterprise and Dell Technologies, have also collaborated with Nvidia to design new AI servers, causing Supermicro’s position to potentially weaken. In its latest 10-K filing, Supermicro disclosed its lack of long-term agreements with Nvidia or other suppliers, leaving no exclusivity.
AMD’s Potential to Challenge Nvidia in the AI Race
AMD had only 17% control of the discrete GPU market last year, lagging far behind Nvidia’s 80% share. While AMD’s share primarily consisted of gaming GPUs for PCs, it has been expanding into the data center market with its Instinct GPUs tailored for AI tasks.
AMD introduced its first set of Instinct GPUs (MI6, MI8, and MI25) in 2017. The latest MI300 Instinct GPUs, utilizing TSMC’s 5nm and 6nm process nodes, were launched towards the end of 2023. Industry benchmarks indicate that AMD’s high-end MI300X outperformed Nvidia’s H100 in raw processing power and memory usage.
This development is concerning for Nvidia, especially as the H100, with ongoing supply constraints, still costs nearly four times as much as AMD’s MI300. AMD’s MI300 gaining traction is evident as CEO Lisa Su anticipates it to have the fastest revenue ramp in the company’s history. Su also notes that AMD’s Epyc CPUs have already captured 25% of the server CPU market share from Intel (NASDAQ: INTC). The growth of these two segments, along with the expansion of AMD’s programmable chip business, positions AMD with more data center bundling options compared to Nvidia.
Embracing AMD at Super Micro Computer
Super Micro Computer has been collaborating closely with AMD to design servers incorporating Epyc CPUs and Instinct GPUs. CEO Charles Liang foresees broad adoption of AMD’s MI300 GPUs, Nvidia’s latest GPUs, and Intel’s Gaudi AI accelerator chips, expanding Supermicro’s presence in the accelerated compute market. Liang’s projections suggest that diversification will substantially increase the company’s AI portfolio.
If Supermicro increases sales of AMD-powered AI servers, it could lessen its reliance on Nvidia in the long run and safeguard itself from potential future supply chain disruptions. Competition from AMD might also incentivize Nvidia to lower GPU prices, enhancing Supermicro’s gross margins by reducing component costs.
An Alluring Investment in the Evolving AI Market
Approximately half of Super Micro Computer’s revenue is derived from AI servers, with Bank of America predicting an expansion in its dedicated AI server market share from 10% to 17% in the next three years. Analysts anticipate a remarkable 42% compound annual revenue growth from fiscal 2023 through fiscal 2026. Super Micro Computer’s stock, trading at a mere 3 times this year’s sales, offers an enticing investment opportunity in the AI sector.
Supermicro’s gradual shift away from Nvidia towards AMD-powered servers positions it as a more balanced play in the AI market compared to either chipmaker alone.
