Revolution Ahead: Super Micro Computer Trails Blazing New Path with AMD Partnership

Super Micro Computer(NASDAQ: SMCI), popularly known as Supermicro, has catapulted its stock price by a staggering 2,220% over the last three years. This meteoric rise has been catalyzed by the booming artificial intelligence (AI) sector, propelling data center operators to flock to the company for its top-tier AI servers.

Supermicro’s blockbuster growth owes much to its reliance on Nvidia(NASDAQ: NVDA), the provider of high-end GPUs essential for processing complex AI and machine learning tasks. Collaborating closely, Nvidia and Supermicro launched a new line of servers and workstations optimized for Nvidia’s H100 GPUs. This kinship enabled Supermicro to carve out a profitable niche in the saturated market for pre-built servers.

Image source: Getty Images.

However, Supermicro’s reliance on Nvidia presents a quandary. It grappled to secure a steady supply of Nvidia’s GPUs in early 2023, while its major rivals, Hewlett Packard Enterprise and Dell Technologies, have been forging partnerships with Nvidia to innovate new AI servers. Acknowledging this vulnerability, Supermicro disclosed in its latest 10-K filing that it lacks any long-term agreements with Nvidia or other suppliers to cement exclusive partnerships.

With a mere 17% share in the discrete GPU market last year, according to Jon Peddie Research, AMD sat distantly in Nvidia’s shadow, commanding an 80% market dominance. Although AMD’s stronghold lay in gaming GPUs for PCs, it embarked on a journey to penetrate the data center market with its Instinct GPUs tailored for AI processing.

Introducing its initial batch of Instinct GPUs (MI6, MI8, and MI25) in 2017, AMD unveiled its latest MI300 Instinct GPUs, manufactured via TSMC‘s 5nm and 6nm process nodes in late 2023. By industry benchmarks, AMD’s cutting-edge MI300X outshone Nvidia’s H100 in raw processing power and memory usage.

Such a feat raises eyebrows at Nvidia, as the H100, plagued by ongoing supply chain hurdles, still carries a price tag roughly four times that of the MI300. While Nvidia contends the H100 triumphs when optimized software runs, this slight edge likely fails to justify its premium cost for budget-conscious data center operators.

Supermicro already collaborates closely with AMD to engineer servers powered by its Epyc CPUs and Instinct GPUs. In a show of optimism, Supermicro’s CEO Charles Liang prophesied that AMD’s MI300 GPUs, Nvidia’s latest offerings, and Intel’s Gaudi AI accelerator chips would all “garner widespread adoption and expand our presence in the accelerated compute market.” A further prediction in January posited diversification was set to “more than double the size” of the company’s AI portfolio.

Amidst already generating half its revenue from AI servers, Bank of America envisions Supermicro expanding its dedicated AI server market share from 10% to 17% in the next three years.

Analysts foresee revenue swelling at a compound annual rate of 42% from fiscal 2023 through fiscal 2026 for Supermicro, a stock trading at just three times this year’s sales. This modest valuation makes Supermicro an enticing long-term AI market play, with its gradual pivot from Nvidia toward AMD-powered servers promising a more balanced stance in the domain compared to the competing chipmakers.