SMCI - Fair Value and Forward Expectations

Investment Analysis Clubs Saul’s Investing Discussions
1

I know SMCI is not a popular hold in these parts, but there are a few. Now that it’s popped 65% YTD after trading sideways for the past 6 months, I just wanted to see what others on the board think about the holding. It’s become my #1 holding even though I haven’t added since May 2023 as my previous #1 holding CELH has traded negative to sideways since August.

The preannouncement raised midpoint guidance revenue from 2.8B to 3.6B, and Non GAAP net income from 4.64 to 5.47 - so margins are decreasing this year - I suppose due to the contents of their solutions increasing in price. I’m not seeing any scenario where margin will really ever expand, this company is a pass through for NVDA and AMD chips - but revenue could theoretically continue to dramatically increase for the next several years depending on how long this AI cycle lasts. Even with margin compression, we’re still looking at a 60%+ EPS increase for this year (~$12 to $20). They’ve gone from 2.19 to 5.54 to $12.09 in the past 3 years. However, the only competitive advantage I can see is their relationship with NVDA (and maybe AMD) - seems very brittle/can turn instantly. I do have to admit that “technological moat” seems to be vastly overrated in some cases - AEHR for example supposedly has a monopoly and can only book $2m in a quarter. Maybe the fact that SMCI has such an in with NVDA is good enough.

So my question to holders on this board is this: despite the lack of a traditional competitive advantage, what is your plan for the shares you have? What is the exit strategy/what would make you sell the holding? We have no idea where in the cycle we are - if it’s still on the uptrend or if we’re nearing the top. The YTD pop is merely the stock multiple getting re-rated, and it seems fair if the cycle is just beginning, but overextended if companies are already close to finished with their purchases of GPUs. It feels like we’re only at the beginning, but those are famous last words.

33 Likes
2

Here’s my two cents for what it’s worth and of course caveat: I could be wrong.

I have owned SMCI since January 2023.

I started trimming once it topped 10% of my holdings as I am not as comfortable with extreme concentration as many members of the board, but it keeps popping back up past 10% and I hate to keep trimming because it’s been such a great winner. So I’ve stopped.

In thinking about SMCI’s future we must consider two things: the growth of the market and the competition.

Growth of the Market
Everyone says it can keep on growing so who am I to disagree? But seriously, I do agree. When I bought SMCI and NVDA (earlier, in 2021) it was not for AI per se and certainly not LLMs because LLMs hadn’t come out. I bought them for the data center market –the “hype” about LLMs has exploded this market but this market was already going to be big and growing forever.

Before LLMs there was ML (machine learning) and that is real and extremely useful and not going away, and ML sucks up a massive amount of data – think UPST and its 1600 or whatever parameters for every person for every loan payment. In addition to that, you have business intelligence etc. etc. Data never shrinks (except during periods of extreme optimization) – it only grows and grows.

So whatever happens to LLMs and how many players there are or aren’t, (and I do think the titans will invest what it takes for however long in this space to make it useful; they have the money and the desire and believe there is a positive outcome to all this, even if they don’t quite know where it is yet) business intelligence and ML aren’t going away.

That’s not to say that there might not be temporary over-purchases and pullbacks, this market will definitely be cyclical, but look out 5-10 years, and data will grow, which means stocks that support the data center will grow.

So nothing wrong with the growth of the market here. And certainly nothing wrong with the growth of the market for at least a year or so, as we know that’s that backlog.

Competition
Now let’s turn to the competition. This is what everyone is worried about. Racks are a commodity, anyone can build them, SMCI has no IP moat. That’s all true.

From what I understand, the biggest competitors for SMCI are HPE (HP Enterprise) and Dell. Other competitors come up in google searches but they appear tangential and many are performing very badly re: revenue growth.

Here’s what others on seeking alpha say about HPE and DELL:

HPE
HPE is divided into several business segments, starting with the server business. Rack-mounts and tower-based servers, HPE is focused on all-purpose servers as well as storage servers. Not driving the company as it once was, the so-called Compute segment remains safely in the 10% or greater source of revenue.

More of a growth industry is Intelligent Edge, where HPE is putting together solutions for companies in data-first modernization. This includes the edge-to-cloud strategy of making cloud storage as seamless as possible, and their own cloud platform, Green Lake.

DELL
Today, Dell provides a comprehensive suite of IT solutions, including servers, storage devices, networking products, and software. Catering to a diverse clientele, from individual consumers to large enterprises and public institutions, Dell customizes its solutions to address specific customer needs.

SCMI + OTHERS
So immediately we see a big difference between SMCI and HPE and DELL: SMCI is focusing exclusively on the racks and HPE and DELL are not. In fact, HPE and DELL see bigger opportunity elsewhere and who can blame them? As everyone points out, anybody can build racks, there may be a price war, so the natural inclination is to put your effort into more upscale targeted solutions right?

But maybe SMCI’s greatest weakness is it strength – that’s all it builds, all day every day, and it’s not going anywhere else (at least that we can tell). They are opening a new manufacturing facility in Taiwan and later this year in Malaysia, where they can build these racks even cheaper.

We see SMCI’s margins going down in this recent earnings report and immediately think it’s a cause to worry (I know I worried). But what if that’s SMCI’s strategy? Let’s make these quality racks as cheaply as we can and sell them with low margins and let’s be the best (or at least in the top tier) racks for a price so low that nobody wants to compete. And yet still make a great profit on volume.

If you are a little guy breaking into the field, how are you going to raise the funds to build a plant to compete with let’s say for comparison – the Amazon of racks. Thin margins with massive volume. You are not. Sure you can build racks, they are a commodity, but there are so many places to go that will be more lucrative than a price war with the King of Racks, and you are going to have to convince the titans that your racks are better or they will have to be a lot cheaper.

So that leaves us to worry only about the big flop if volume drops precipitously. Certainly a possibility and something to keep an eye out for.

I tend to be a very long-term hold kind of person. A big mistake I’ve made in the past is to look out 5 years and ask myself, can I see this company really winning then and if I can’t see clearly, I either don’t go in or I sell. That kept me completely out of zoom, but look how foolish that was in the short term – I lost out on an obvious gain because I could see past that gain and could see there were no adjacencies. But I have learned from this board that thinking short term can work too.

Is SMCI a forever hold? Likely not (whereas I believe NVDA is, as long as Jensen Huang is there). But does it seem like a good bet for the near future? Yes, indeed. May I trim again? Yes, I may – depending especially on the next investor call with the full earnings report. Also, FWIW, I think the CEO is uninspiring and glassdoor ratings are borderline unacceptable. Two minuses I think about all the time. However, how inspiring must you be when all you do is build racks of excellent quality really cheaply?

I will always be looking for data that will change my thesis. But until I see it, I’m holding. It’s still a small company, it’s financials are great, and it’s up 550% in one year. Congrats to all of us who have held!

44 Likes
3

It’s my impression $SMCI is differentiated vs. its competitors though in that its liquid-cooled system dramatically reduces power consumption by dramatically decreasing heat output. I wonder if that’s a material differentiator or not?

If so, that makes $SMCI not just a AI/HPC play but also a datacenter-optimization play.

IMO datacenter optimization is basically a permanent “trend” that is relevant to several stocks discussed here including $SMCI, $ANET, $PSTG and arguably $NET

No position in $SMCI, long $ANET, $PSTG, $NET 2%-ish each

16 Likes
4

The moat for SMCI is their reference designs. They provide a proven PCB and chipset that has the US and international homologations and certifications. SMCI will identify the number of reference units shipped and their reliability. This is a big deal for appliance and rack equipment manufacturers as far as cost, risk, and schedules. This allows in house board design and manufacturing to be out sourced with a reduced head count. As a customer you can then have them customize the reference to your spec. They will recertify but at a lower risk as the reference design has already been approved.

I would argue that the services that SMCI provide are not a commodity that is easily replicated.

-zane

29 Likes
5

Thank you to both Iamnzane and intjudo for your commentary. I read everything I can get my hands on for SMCI and have seen the cooling differentiator before but not the reference design. It’s so amazing that we have people with such good knowledge on the board, and now I am even more convinced to just hold and ride for SMCI. If anybody has a solid reason not to, please post.

8 Likes
6

A lot to unpack here.

First, I think the speculation about margin compression is premature at best. Remember, the preannouncement was about guidance, and it makes sense that management would temper guidance raises at least somewhat, if not very conservatively. Which would mirror what Nvidia itself has done with its guidance raises - as much as Nvidia raised guidance the past few quarters actual results exceeded that.

Second, I don’t think it’s fair to describe SCMI’s relationship with NVDA as “brittle.” The relationship of the two CEOs goes back decades. I think Super Micro was the first company to package up Nvidia chips into a data center product. The CEO of SuperMicro was on the stage with Jensen Huang recently as well.

A third point is that in the past Meta (aka Facebook) has been a large (over 10%) customer of SuperMicro. And Zuck just announced that Meta will seriously grow its installed base of H100-based servers this calendar year to 350k Nvidia H100s, plus another 250k of “equivalents.” So, Meta intents to purchase could be part of SuperMicro’s raised guidance.

One point I haven’t seen discussed is that with H100s being locked out of China now and Nvidia saying they will still be production constrained, that originally-for-China supply is going to other places. Being an US-based company, it’s not unlikely that opens up more H100 supply to SuperMicro, so it can expand its sales.

As to stock multiples and re-ratings, let’s remember that we’re talking serious better company performance being promised here. Even with the same multiples the stock price would have shot up a lot. I won’t argue that the run-up the past could of days (say over $450) doesn’t imply investors valuing SCMI more highly than in the past (we’ve had a short report and insider sales this past year or so), but certainly the preannouncement of guidance even at the same multiples as before would have produced a nice bump in the stock price.

Finally, I strongly believe the AI cycle is just getting underway. Companies are far from “close to finished with their purchases of GPUs,” as Meta’s announcement shows, plus commentary from Nvidia’s CEO on the last earnings call about all the new markets for AI, including governments of countries needing AI, etc. At some point the high growth aspects of the AI cycle will shift from the current chips & servers (picks and shovels), to platform software and applications, but there are lots of existing markets to which adding AI will transform, even disrupt those markets, as well as entirely new markets based on the new capabilities.

36 Likes
7

Lots of investors ask “where’s the moat?”… and “they have no advantage”. Heard that for years, often applied to failures :wink: like NFLX and TSLA and others. IMO, the best evidence of a strong business is continued growth in sales, profits and market share. Yes, there are nuances in that, especially if applied to the companies I mentioned. Don’t run off on a rabbit trail mentioning those companies… I won’t bother to respond.

I’ve been pondering the question of the original poster, which I’d summarize as “What’s SMCI worth?”…

That’s something I’ve struggled with for over a year, since I first dipped my toes in around $75. I don’t have THE DEFINITIVE ANSWER. lol. But let me share some thoughts:

  • Historically, this market space has been considered a commodity space. Yet SMCI apparently has some advantage. I’ve seen claims that their liquid cooling is an advantage. Maybe so, but that doesn’t seem to be a durable advantage, especially in the face of larger competitors. SO, I’m “suspecting” SMCI’s advantage is not just innovation… but that it’s coupled with nimbleness (something big established companies often can’t duplicate… hello, GM!) and an ability to customize and deliver quickly.

  • With that being said, I ask myself “What is a reasonable PE?” I HAD BEEN somewhat resigned to a PE around 20-22, but then I ask myself if the recent re-pricing indicates a growing “belief” in the market that perhaps a higher PE is warranted due to them having a durable advantage… at least for a few years. That involves some sub-thoughts:

— Declining margins. If true, and it may be, the bottom line is the bottom line… which is exploding. And a small nimble organization like SMCI can deal with lower prices/margins easier than a big clunky corporation known for being ponderous.

— So… finally… in trying to find some alignment between the CURRENT share price… and that the market wants to see share price gains from the current level (or else it isn’t worth buying at a high price if you don’t think it’ll continue to rise)… I’m tentatively thinking maybe a PE of 30 may be “reasonable”.

— AND… given at least $5.50 for 4Q, that gives us a current earnings run rate of $22/year. THEN, if you “assume” the coming four quarters have at least a 40% growth from there, you can do some crazy math and say: $22/year X 1.4 (growth) X 30PE = $924/share as a “reasonable” value a year from now.

— Disclaimer: We MAY be looking at “irrational exuberance” and that $924 is just crazy talk in that time frame.

— Bottom line: What do I conclude? I, in my non-professional (and I might add “bad person”… see below) way, am uncomfortable with the rise but don’t plan to make any moves with my SMCI position until I see results AND see their guidance. I think there is an excellent chance the stock will drop around that point because of people wanting to “lock in their profits”… as inane as that sounds. AND I think that I’ll have an SMCI position even into 2025. Maybe beyond… I never promise long term fidelity to a stock.

— Disclosure: I am wildly overallocated again. SMCI is now 35% of my portfolio and NVDA is 35% also, by coincidence… due to the huge recent gains in both of them. These are obviously the bulk of my portfolio… subject to my “Final note” below:

Final note: Please feel free to completely ignore my thoughts here. After all, I’ve been told on this board… by “authority”… that I don’t own a portfolio because I have high allocations. And that makes me a bad guy apparently. Place me on ignore if you like.

Rob
He is no fool who gives what he cannot keep to gain what he cannot lose.

50 Likes
9

“I would argue that the services that SMCI provide are not a commodity that is easily replicated.”

Lots of good replies here to those who say SMCI is pure commodity and has no enduring competitive advantage.

Brilliant visionaries with outstanding executive leadership who provide superb customer service are not commodities. The SMCI CEO is brilliant, but his accent is so severe that he is mostly unintelligible in English. So people underrate him. This stock was one of those small caps unjustly held down for many years through 2018.

As explained by others above, Dell and HPQ are hardly competitors. SMCI is trusted by NVDA, AMD, INTC, and others. I found SMCI in mid 2022 while browsing SA’s quant top picks. When others brought it to this group last year, some called it speculative. My reaction was SMCI was among the least speculative stocks in the market at the time. It’s been a 30 bagger over 5 years, a 10 bagger since I found it 18 months ago, and a 6 bagger since a short report enabled an excellent buying opportunity 12 months ago. It no longer looks so undervalued but certainly is still cheap compared to NVDA or AMD, and remains my preferred way to participate in the AI revolution.

I’ve trimmed along the way. SMCI never participated in the share price collapse of hypergrowth tech stocks in 2021-2022.

I always think there is a correction ahead so I usually limit the size of my holdings. Shift4 Payments (FOUR), SMCI, PSTG, ADYEY, and TSLA remain my top 5 holdings. All brilliantly led high growth cheap stocks under the radar save TSLA which has a valuation tough to justify without better progress on FSD.

19 Likes
10

Super Micro’s competitive advantage is in their engineering and product design which provides AI optimized servers. They are like lego building blocks so a customer can build the rack and server to their exact specifications. This can be applied a variety of verticals, and for generative AI there are certain racks and builds they have which outperform anything on the market.

Super Micro has 5,100 employees, and they’ve mentioned before that over half of their employees are engineers. This means the company is over 2,500 engineers. They are spending a massive amount in the R&D category.

Recently their large client is rumored to be Facebook who is making inroads into AI. My educated guess is the pre-announcement of results is because Facebook contributed significantly to the quarter.

The numbers they are putting up speak for themselves, they are both a growth and value company. This is what attracted me so much to this company was that they are growing top line and growing the bottom line in proportion.

It’s maybe worthwhile for people to understand the breath and depth of their engineering efforts by looking at their product line,

  1. Super Blade - high performant, energy efficient, optimized for AI, data analytics, high performance computing, cloud & enterprise

  2. GPU Servers with PCIe GPUs - advanced accelerators to deliver dramatic performance gains and cost savings, high performance computing, AI/machine learning, rendering

  3. Universal GPU Servers - open and modular, peta-scale storage, unprecedented capacity and performance in a single 1U or 2U chassis

  4. Hyper - performance rack mount servers, built for most demanding workloads with storage/IO flexibility, custom fit

  5. HyperE - optimized for deployments in edge environments, edge data centers and telco cabinets

  6. BigTwin - 2U 2-node or 2U 4-node platform, providing superiority density performance and serviceability. Dual processors per node, ideal for cloud, storage, and media.

  7. GrandTwin - purpose built single processor for performance

  8. FatTwin - multi-node 4U twin architecture

  9. Edge Servers - compact, optimized for edge, can operate up to 55 degrees C

  10. CloudDC - all in one platform for cloud data centers, maximum data throughput

  11. WIO - I/O optimized for specific enterprise scenarios, dual process platforms

  12. Enterprise Storage - large object storage and workloads

  13. SimplyDouble - large scale storage servers with dual tier design

  14. Workstations - portable under desk factors are ideal for AI, 3D design, and media

19 Likes
11

Addendum to my earlier comments on SMCI:

A slightly different look at it…
The pre-announcement suggests a current earnings run rate of $5.50 x 4 = $22/share.

Take the current share price of $470 and divide by $22. That’s a current PE of about 21. For a company growing EPS by 60%+. And, last time I looked, modest debt and an exploding market that shows no signs (YET) of slowing down.

So… yes, it’s gained a lot. But it still seems undevalued to me. And I haven’t heard anyone suggest that data center builds are slowing down.

Rob
He is no fool who gives what he cannot keep to gain what he cannot lose.

22 Likes
12

As far as I saw SMCI has pre-announced Q4 updated expectations but NOT forward guidance. The SP has rocketed by 50% on this news subsequently. On ER date, their guidance better be commensurate with this positively received results update otherwise there could be some serious pressure on the growth story, PE valuation expectations and the share price.

4 Likes
13

All true!

Having followed this company for a while, I think it’s a safe bet that the trajectory is not going to severely change for a while. That might prompt a “So why is the pre-announcement a big thing if it’s not a big change?”

Glad you asked. Answer: SMCI performance has been VERY strong for the last year and the stock price gains to date have been impressive. It seems that the market just hasn’t realized just HOW strong it’s been until recently.

“Will this continue?” Another good question! Hyper growth is short lived if your time horizon is long enough. But if you’re just looking a year ahead… or two years… it seems that growth can continue for a while. At least a year, I figure. And… and this “and” is important… the company makes money and the PE is very modest compared to the growth. Works for my sort of investing, which is “Stay for the sunny days, but don’t stay on the beach when the clouds roll in.” In other words, if a company fails to meet my expectations… or if a better one comes along… I move. Buying a company does not involve loyalty.

Rob
He is no fool who gives what he cannot keep to gain what he cannot lose.

27 Likes
14

Ted

This is a 2nd example of perhaps 20 members inputting on whether the seldom held company in Sauls Group should be considered. I still continue to be impressed.

Jer

1 Like
15

The update did say:

With a strong market and end customer demand for our rack-scale, AI and Total IT Solutions,

Whether the guidance to be given is commensurate with the stock price gains is certainly something to question, but not that the growth story is intact.

12 Likes
16

Welp, guidance is out and we’re looking at $14.3 to $14.7b, which would be 104% YoY at the midpoint - absolutely insane

Margins are indeed going down - from about 19% to 15.5%, still I’ll take 104% revenue growth if it means losing 3.5% of margin

Currently trading at $535 after hours, so almost 90% gain YTD

congrats to all the longs on this board for holding through the 2nd half of 2023

21 Likes
17

I bought a 5% position recently. Really happy with the report and blowout guidance. They are a conservative management team, so at least 15 million in revenue for FY24 is not in doubt at all. I’m hearing that META orders might part of the giant growth story here.

FY2024 guidance raise from $10-11 billion to $14.3-14.7 billion - is crazy good.

$15.5+ billion for FY2024 which is 118% YoY growth from FY2023 is a likely scenario.

Not really a Saul stock, not a wide moat and has low margins, but I’ll keep it small. The AI chip capex is so large I think they have a shot to continue their hyper-growth for a while, especially if they can lower costs with their new factories.

Edit:

If SMCI is doing this and ASML orders just tripled, as they announced last week, imagine what NVDA might report? I’m very bullish. Hopefully we can all pan for some gold as the semi space is projected to go from a $600 billion dollar a year industry to $1 trillion by 2030.

Things looking good so far.

Long NVDA, 10%

27 Likes
18

Just to temper that thought a bit - remember that as of November, Nvidia isn’t selling its top of the line chips to China - which means losing some of these top previous purchasers:

https://www.instagram.com/theinformation/p/C09YTf2MGAp/

IMG_7074
IMG_70741200×675 46.5 KB

So, losing Tencent, Baidu, Alibaba and TikTok at least. Which means more chips for the rest, which means companies like SCMI getting more chips to build into US friendly customer’s servers, but not necessarily more AI chips produced and sold by Nvidia overall.

I don’t believe Nvidia is going to have a problem selling every AI chip they make, but the export ban does cloud the extent of their growth beyond the next couple of quarters a bit.

FWIW, I’m still long both companies, and actually bought more NVDA last week.

33 Likes
19

I think this video provides interesting perspective for SMCI investors.

It basically says NVDA is so desperate for server suppliers for its chips (to sell more chips), and it knew the server suppliers would take too long to design super micro like solutions, that it went ahead and created the designs for them (which SMCI gladly took as well).

This could very well be the source of the pricing war and reduction of margins for SMCI. He does point out that SMCI still offers solutions the others can’t but that maybe that will go away over time.

For me, it’s a data point to watch and follow but I after researching the competitors I’ve decided to take a wait and see attitude on whether any of them will go beyond a nip at SMCI’s heels.

All the competitors listed are Taiwan companies, many of them offer other solutions (wheras SMCI is dedicated) and the one I found that appeared to be dedicated - QCT - lists on their website as having “a couple of hundred employees.”

Something worth watching, to be sure, but maybe just benefiting SMCI the most at the end, because of its scale.

18 Likes
20

Anyone who says that doesn’t understand the chip market. The title of that video is either pure click-bait (at best), or a demonstration of not understanding the market.

Every chip manufacturer that I’m aware of has References Designs available. Here’s a link to Texas Instruments:

Here’s Qualcomm’s:

Reference designs are not commercial designs - they’re made to show how the chip works and can be integrated, but different companies will want to vary the reference design for their particular needs/desires/capabilities.

This is not indicative of being “so desperate,” it’s part of getting your product sold.

SCMI doesn’t have any real secret sauce that I’m aware of, but it does make what many (I’ve spoken with a couple directly but can’t give names, sorry) consider the best Nvidia servers. That Meta reportedly buys them instead of building their own says a lot.

As for moat/margins, one has to expect that competition is heating up.

29 Likes
21

Smorgasbord1 is correct. Chip vendors release “reference designs” when they release specs for their new chip. Their engineers have time to design and build test circuit boards in house without concern for leaking proprietary data. They they share the design with their customers. Customers use the design as a starting place for their own circuit board design. Faster time to market, faster money comes in to everyone. I have been in those industries for much of my career. It is common practice.

13 Likes