Investors in leading cloud and semiconductor stocks have probably seen big gains this year, as the artificial-intelligence (AI) revolution sparked by the introduction of ChatGPT last November began translating into blockbuster revenue and profit gains for well-positioned companies this spring.
While many have homed in on GPU leader Nvidia (NVDA 6.18%), cloud giant Microsoft, or more speculative software bets like C3.ai, one company in the less well-regarded server market has also been a huge winner.
Super Micro Computer (SMCI 8.89%) has carved out a very attractive niche for itself within the AI industry, and the stock has climbed 163% in 2023. But what's even more impressive is that, unlike some other tech stocks that were down in 2022, Super Micro is achieving these gains after rising nearly 87% last year -- and remember, that was in a year in which the Nasdaq Composite (^IXIC 2.02%) fell more than 30%!
Although the stock has nearly quintupled in a mere 18 months, one Wall Street analyst recently upgraded Super Micro while raising her price target to $325 -- about 50% higher than shares sit today.
That may seem laughable to some, but unlike some other AI plays that seem extremely expensive, there's a perfectly reasonable case for Super Micro to reach that target and even surpass it.
Loop Capital is all in on SMCI's "Legos"
On June 14, research firm Loop Capital analyst Ananda Baruah raised her price target on Super Micro Computer from $200 per share to $325, indicating a roughly 50% gain from the stock's current level.
In her note, Baruah said Loop's work suggests that "the Gen AI compute build has both legs and commercial commitment, with Gen AI going to 30%-40% of all Hyperscale applications from less than 5% today." Within the generative AI megatrend, Baruah especially likes Super Micro Computer because of its unique "complexity at scale" building block model, which Baruah believes has at least a 12- to 24-month lead over the competition.
What is the building block business model? At a recent conference hosted by Rosenblatt Securities, Super Micro founder Charles Liang describes the strategy of building servers as if by "Legos," with each component optimized for quality, functionality, cost, and integration, which can be formed into a myriad of customized server designs.
As opposed to other server companies that mass-produce standard models, customers can optimize Super Micro's server systems based on the end application. And with the proliferation of not just artificial intelligence but also 5G, edge computing, and the Internet of Things, customers appear to be hungry for customized solutions that can be delivered quickly at scale.
In fact, Loop had mentioned this large-scale customization advantage in its previous note on May 24, when it had initially raised its price target on Super Micro to $200 from $150. In that note, Baruah said that while AI is the most important and prominent catalyst, "Super Micro's core value-add is deep server customization providing a differentiated impact for increasingly strategic applications, including AI/ML [machine learning], Data Analytics, Strategic Video Streaming, etc."
Super Micro can save AI data centers lots of money. Image source: Getty Images.
And don't forget the advantages of green computing
Loop isn't the only research firm that's bullish on Super Micro's prospects from here, either. Earlier in June, Rosenblatt Securities' Hans Mosesmann initiated coverage on the company with a $300 price target, implying about 40% upside from current levels.
Mosesmann also noted the mass-customized building block model but is also bullish on Super Micro's "green computing" ethos, whereby the company touts its servers as the most energy-efficient in the industry. While "green computing" used to be a nice-to-have feature, in the world of AI, it's much more important. Leading-edge chips require tremendous amounts of electricity to run and also generate large amounts of heat, which requires some way to cool the chips. Since cooling systems require space and electricity in and of themselves, this increases the electricity and square footage needs of AI data centers even further.
Mosesmann wrote in his June note that Super Micro's new liquid cooling technology can increase rack compute power by over two times, which Mosesmann describes as "a disruptive dynamic in a power constrained data center."
So while Super Micro isn't in the business of designing highly specialized proprietary semiconductors, its novel server architecture also contributes to increasing the computing power of AI data centers, albeit by more indirect means.
A plausible path to $325
Even after quintupling, Super Micro trades at only around 20 times earnings, and those earnings incorporate a weak non-AI server market, as well as supply constraints for AI chips that limited Super Micro's growth last quarter.
Analysts have consistently underestimated Super Micro's growth over the past few years and are still only forecasting modest earnings growth in the year ahead for some reason. Yet Charles Liang sees the company growing "at least" 20% in the upcoming fiscal year, starting July 1.
If Super Micro can maintain a 20% earnings growth rate over the next few years, that would amount to a PEG ratio of just 1. That's actually fair or even cheap for a technology stock, which means Loop's $325 target is certainly attainable in the year ahead.
