Super microcomputer (NASDAQ: SMCI), better known as Supermicro, has seen its shares jump 2,220% over the past three years. This impressive rally was fueled by the rapid expansion of the artificial intelligence (AI) market, which prompted data center operators to purchase more of their powerful AI servers.
Much of Supermicro's growth is directly attributable to this Nvidia (NASDAQ: NVDA), which provides high-end GPUs that handle complex machine learning and AI tasks. Nvidia worked closely with Supermicro to develop a new line of servers and workstations that fully support its H100 GPUs. This close collaboration enabled Supermicro to carve out a high-growth niche in the highly commoditized prebuilt server market with its AI servers.
Image source: Getty Images.
But Supermicro's dependence on Nvidia is a double-edged sword. The company struggled to secure a stable supply of GPUs from Nvidia in early 2023, and its two biggest competitors – Hewlett Packard Enterprise And Dell Technologies – have also worked with Nvidia to design new AI servers. In its most recent 10-K filing, Supermicro admitted that it has no long-term agreements with Nvidia or its other suppliers that actually bind them as exclusive partners.
For this reason AMD's (NASDAQ: AMD) recent expansion into the data center GPU market could be good news for Supermicro.
Why AMD could catch up with Nvidia in the AI race
According to Jon Peddie Research, AMD only controlled about 17% of the discrete GPU market last year, putting it well behind Nvidia in second place with an 80% share. AMD's share consists primarily of gaming GPUs for PCs, but the company has also expanded its reach into the data center market with its Instinct GPUs for processing AI tasks.
AMD launched its first Instinct GPUs (MI6, MI8 and MI25) in 2017. The company launched its latest MI300 Instinct GPUs, which are manufactured with TSMC's 5nm and 6nm process nodes, late 2023. In several industry benchmarks, AMD's high-end MI300X actually outperformed Nvidia's H100 in terms of pure processing power and memory usage.
The story goes on
That's a glaring warning sign for Nvidia, as the H100, which faces ongoing supply chain constraints, still costs about four times as much as the MI300. Nvidia claims that the H100 still outperforms the MI300 when running optimized software, but that small difference probably won't justify its premium price for cost-conscious data center operators.
That's why it wasn't surprising when AMD CEO Lisa Su recently said the MI300 was on track to achieve the “fastest sales increase of any product” in the company's history. Su also estimates that AMD's Epyc CPUs have captured 25% of the server CPU market, at the expense of Intel(NASDAQ: INTC) market-leading Xeons. With the growth of these two companies and the expansion of its programmable chip business (from Xilinx), AMD has more data center aggregation capabilities than Nvidia.
Get comfortable with AMD
Supermicro is already working closely with AMD to develop servers for its Epyc CPUs and Instinct GPUs. In November, Supermicro CEO Charles Liang predicted that AMD's MI300 GPUs, Nvidia's latest GPUs and Intel's Gaudi AI accelerator chips would all “see widespread adoption and expand our share of the accelerated computing market.” In January, Liang predicted that continued diversification would “more than double” the size of the company's AI portfolio.
If Supermicro sells more AMD-based AI servers, it could reduce its long-term dependence on Nvidia and insulate itself from future supply chain constraints. Competitive pressures from AMD could also prompt Nvidia to lower its GPU prices, which would boost Supermicro's gross margins by reducing component costs.
A balanced play in the AI market
Supermicro already generates around half of its sales from its AI servers Bank of America expects it could expand its share of the dedicated AI server market from 10% today to 17% in the next three years.
Because of this, analysts expect revenue to grow at a compound annual rate of 42% from fiscal 2023 (which ended in June) through fiscal 2026. That's a staggering growth rate for a stock that trades at just three times this year's sales.
This low valuation already makes Supermicro an attractive company in the AI market over the long term, but its gradual diversification away from Nvidia with AMD-based servers could make it a more balanced company in the AI market than the two chipmakers.
Should you invest $1,000 in Super Micro Computer now?
Before you buy shares of Super Micro Computer, consider the following:
The analyst team at Motley Fool Stock Advisor just found out what they think they are The 10 best stocks so investors can buy it now… and Super Micro Computer wasn't one of them. The ten stocks that made the cut could deliver huge returns in the years to come.
Stock Advisor offers investors an easy-to-understand roadmap to success, including guidance on building a portfolio, regular updates from analysts, and two new stock picks per month. The Stock Advisor service has more than tripled the return of the S&P 500 since 2002*.
Check out the 10 stocks
*Stock Advisor returns as of February 20, 2024
Bank of America is an advertising partner of The Ascent, a Motley Fool company. Leo Sun has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Advanced Micro Devices, Bank of America, Nvidia, and Taiwan Semiconductor Manufacturing. The Motley Fool recommends Intel and Super Micro Computer and recommends the following options: long January 2023 calls on Intel for $57.50, long January 2025 calls for $45 on Intel, and short February 2024 calls for $47 on Intel. The Motley Fool has a disclosure policy.
Forget Nvidia: AMD Could Be Super Micro Computer's New Best Friend was originally published by The Motley Fool