Penguin Solutions - not a typical Saul stock, but worth a look?

First thread I’m starting here, so go easy on me :slightly_smiling_face:

I’ll say up front this isn’t a classic Saul-style company. Gross margin is in the high 20s, there’s debt, it’s a mix of three businesses, and a lot of the growth comes from memory pricing. But the growth just turned sharply and I didn’t see a thread on it, so I’m curious what others think.

Penguin reported fiscal Q4 on Oct 6:

  • Revenue $567M, up 68% YoY (Q3 was +48%, first half was roughly flat)
  • FY26 revenue $1.73B, up 26%; non-GAAP EPS $2.87, up 51%
  • FY27 guide about $2.43B, roughly +40%

The catch is where the growth comes from. Integrated Memory was $341M, up 158%, and is now about 60% of sales. Advanced Computing, the AI infrastructure business they want to be known for, was $154M and only up 11%, and down 14% for the full year because of the Penguin Edge wind down and lower hyperscale sales. The part they’re excited about, non-hyperscale AI infrastructure (enterprise, sovereign, neocloud), nearly doubled and is now two thirds of that segment, so there’s something real growing underneath.

The bull case as I see it: memory demand stays tight for a while, and by the time it cools, the AI infrastructure business is big enough to carry growth. They raised the FY27 Advanced Computing outlook from mid-teens to about 40% in one quarter, which helps that argument.

The bear case: it’s mostly a memory-cycle stock with low margins, and those businesses usually get valued that way once pricing turns. Two details from the release lean that way. The CFO said Q4 gross margin improved because memory got more profitable, partly offset by lower Advanced Computing margins, so the AI side isn’t carrying profits yet. And non-GAAP diluted shares went from 52.8M in Q3 to 62.0M in Q4, which I assume is tied to the convertible notes they issued in July. Worth keeping in mind before valuing it on EPS.

The thing I keep coming back to is how much of the memory growth is price versus volume. If it’s mostly price, the numbers could fade fast once the cycle turns, and then it comes down to whether the AI infrastructure business is big enough by then. Curious if anyone has a read on that.

No position. Mostly trying to decide if it belongs on my watchlist.

31 Likes

Hey Wilson, I am trying to look into Penguin and just getting distracted. That is not a ding on your report, but says a lot about what I am finding for research. The ~20% margins is relatively low for those of us who invested through the SAAS days, and well, it’s ok. This is mostly a hardware company which also limits it’s ability to get that gross margin higher.

Their investor presentation shows a step called “Manage” on slide 11…but I don’t see a product for that. Later on slide 22 they break out “services” as 251M and that’s about 16% of revenue for the FY. Since they predict their own margin dropping by 2% for fy27 that implies that services are not going to grow? It would be nice to see if they are leaning into anything that is reoccuring, but I cannot find it.

With the bit of research I could do, it seems like a good trend play, but not really differentiated enough to catch my eye.

(This is not as good a response as others on the board can do, but it how I go about deciding if I want a stock on my radar and possibly open a trial position. This one feels like it would need more monitoring that I am willing to do.)

10 Likes

Wilson,

My main concern is that they are significantly behind every other memory player in seeing revenue increase. Which to me feels like they are just a boat going up with the tide.

Drew

4 Likes

@drew1618t Are you looking at Penguin strictly as a memory investment, or have you considered its AI infrastructure operations business? Because I agree the memory segment alone probably isn’t the most compelling reason to own it.

1 Like

Fair points. On the “Manage” side, they do have ClusterWareAI, which they use to monitor and manage GPU clusters. Management also mentioned that their managed services contracts typically run 3 to 5 years.

What caught my attention were the recent neocloud wins. They’re not just supplying hardware, they’re also helping run these clusters over multiple years. Non-hyperscale AI infrastructure revenue grew 99% in Q4, so there’s clearly some traction there.

I wouldn’t necessarily take the lower FY27 margin guidance to mean services won’t grow. Memory and hardware are growing much faster, which could pull down overall margins even if services improve.

Services revenue did decline 11% last quarter, mainly because older hyperscale projects wound down, partially offset by growth in newer AI managed services. They also don’t break out recurring software revenue or services margins, so it’s hard to know how valuable that part of the business really is.

That’s what I’m interested in watching. If these new projects start bringing in meaningful recurring revenue, there may be more than the memory cycle. Otherwise, I agree it’s mostly a hardware play.

5 Likes

Wilson,

I see their AI (Advanced Compute) growing 11% Year-over-year. I find the other sectors seems to be less of a compelling reason.

Penguin Solutions (PENG) – Quarterly net sales by segment ($M) and YoY growth, fiscal 2026

Segment Q1 FY26 YoY Q2 FY26 YoY Q3 FY26 YoY Q4 FY26 YoY
Integrated Memory 136.5 +41.2% 171.6 +63.1% 275.1 +111.4% 340.8 +157.9%
Advanced Computing 151.6 -14.5% 115.6 -42.2% 137.6 +3.8% 154.0 +11.4%
Optimized LED 55.2 -17.6% 55.8 -7.4% 66.1 +7.2% 71.9 +6.6%
Total 343.1 +0.6% 343.0 -6.2% 478.7 +47.6% 566.7 +67.7%

Drew

8 Likes

@drew1618t that chart is excellent – thanks for putting it together.

Since I’m posting I’ll also use this opportunity to thank you for your offer to post the rules of the board on Mondays. Yes, please do!

…

One of the things we can no longer do on the Motley Fool boards is send a direct response by email, nor is there any other way to send a private message. That’s why I’m posting this on the board. It’s not ideal to clutter up the board this way…but I don’t know what else to do. With WPR and Ryshab and others who are on Twitter, I can message them there, and yes, Reddit allows that too.

@drew1618t are you on Twitter or Reddit?

Bear

on twitter

and on reddit

2 Likes

Drew, you’re right, Advanced Computing only grew 11% in Q4. But I think there’s a little more to it.

On the earnings call, management said non-hyperscale AI infrastructure revenue grew 99% in Q4 and 72% for the full year. The rest of the segment has been held back by the wind down of Penguin Edge and lower hyperscale hardware sales.

They also added 17 new AI infrastructure customers during the year, with several new neocloud deals announced in Q4.

I agree the overall numbers aren’t particularly impressive yet, but I’m more interested in where that business is heading. The bigger question for me is whether those new customers eventually turn into meaningful recurring services revenue, rather than just hardware sales.

That’s still unproven, and I think the next few quarters should give us a better idea.

5 Likes

Nice write up on Penguin Solutions! It is an interesting question if this stock is a Saul stock or not. Going by a strict definition from the Knowledge Base, the business is lower margin and sells a physical product which would go against aspects of a business that Saul was looking for. However, I believe Saul would have adapted his criteria for the new group of hardware based companies that have emerged over the last few years.

I had a position back in August after their prior fiscal Q3 2026 in July which was strong. However, I sold them back as their CEO has been there less than a year, and there was a interim CFO this summer. Additionally, it was more of a one quarter step from the prior revenue amounts they were getting.

I’m back in Penguin wth a small starter as I thought this latest fiscal Q4 2026 report in October was even stronger. It took me awhile to understand the product lineup, because as it has been pointed out, most of their growth comes from Integrated Memory. However, the press release really made a focus on sales to Neoclouds which I found confusing at first.

The sales to NeoClouds with their ClusterWareAI product goes in the Advanced Computing segment. As @drew1618t pointed out, this Advanced Computing segment only grew 11% yoy. This is because the company had a legacy product called Penguin Edge that had customers not renew, and is subsequently being winded down.

The timeline with the new CEO is no coincidence from my point of view. The CEO Kash Shaikh started roughly nine months ago, and the business has had two booming quarters since then.

These are just some of the NeoCloud deals they announced in their press release, all of which will go into the Advanced Computing segment,

  • “Won six new AI Infrastructure data center customers in the fourth quarter, including four neocloud providers, reflecting a surge in demand from neocloud customers for our full-stack AI Factory Platform.”
  • “Continued to execute our land-and-expand strategy. Across fiscal 2026, we added seventeen new AI Infrastructure customers, and twelve customers expanded their business with Penguin in the same period.”
  • “Won a neocloud customer backed by a leading South Korean technology company that selected Penguin to design, build, deploy, and manage an NVIDIA GB300 NVL72-based platform.”
  • “Won a publicly traded neocloud customer with more than $3 billion in signed, multi-year contracts to provide AI infrastructure deployment and 24x7 operations services, supported by ClusterWareAI™, our AI factory operating system software.”
  • “Won a neocloud customer with $10 billion in contracted compute from a leading AI lab. The customer selected Penguin to deploy and operate a 36,000-GPU AI factory in Norway. This multi-year engagement demonstrates the scale and capabilities of our AI Factory Platform.”
  • “Won another neocloud customer, Lektra, which selected our AI Factory Platform to deploy and optimize distributed AI micro data centers powered by existing carbon-free energy. We provide validated reference designs, NVIDIA-based AI compute, and expert services with ClusterWareAI and support.”
  • “Advanced ClusterWareAI with new self-managing agentic AI capabilities, building on the AI Factory Operations Agent introduced last quarter. ClusterWareAI now automatically detects and remediates GPU performance issues across inference environments, helping customers maintain higher uptime and reduce operational overhead as they scale AI workloads.”

From what I gather with this ClusterWareAI solution, it sounds like a full solution for operations of a NeoCloud. All of this new business is going into the slower growing division of Advanced Computing, so I would expect this segment to start showing strong growth.

One last detail that is worth mentioning with this business is that while their HQ is in California, the business is optimized around low cost locations. Engineering is primarily in India, with some in Taiwan. Manufacturing is in Malaysia, and a smaller base in China. This is how the company is getting a GAAP net income margin of 16.5%, which is pretty surprising for a 27.5% gross margin metric.

Overall, I do think the two strong quarters in a row, and the new Neocloud business line would make this a stock Saul would be interested in if he was still an active investor.

26 Likes