To sum up: They rehired execs that got them delisted/fined by SEC, those execs are again doing shady rev recognition, there are a number of suspicious related party transactions, possible export violations to Russia, Dell is winning over key customers (Tesla, CoreWeave, Digital Ocean), and customers seeing elevated rate of hardware issues.
… All told, we believe Super Micro is a serial recidivist. It benefitted as an early mover but still faces significant accounting, governance and compliance issues and offers an inferior product and service now being eroded away by more credible competition.
Was down -8% but now back to -2%. … Market seems to be shrugging.
Shady Rev Recognition
In 2018, Super Micro was temporarily delisted from Nasdaq for failing to file financial statements. By August 2020, the company was charged by the SEC for “widespread accounting violations,” mainly related to $200+ million in improperly recognized revenue and understated expenses, resulting in artificially elevated sales, earnings and profit margins.
Less than 3 months after paying a $17.5 million SEC settlement, Super Micro began re-hiring top executives that were directly involved in the accounting scandal, per litigation records and interviews with former employees.
A former salesperson told us: “Almost all of them are back. Almost all of the people that were let go that were the cause of this malfeasance.”
According to a lawsuit filed in April 2024, Super Micro waited only 3 months after the SEC settlement before restarting “improper revenue recognition,” “recognizing incomplete sales,” and “circumvention of internal accounting controls”.
Marquee customers leaving
The huge number of marquee customers moving off SuperMicro is probably the biggest stain, pointing to reliability issues. Dell looks to have won CoreWeave, Tesla, xAI, and Digital Ocean.
CoreWeave was Super Micro’s largest customer over the last year, per Bloomberg Intelligence. But in December 2023, Dell announced a deal with CoreWeave for “thousands” of GPU servers, potentially worth over $1 billion.
Tesla had been sourcing its servers exclusively from Super Micro, per Barclays Research in September 2023. But recent reports in May 2024 and posts by Elon Musk show Dell has now won major deals from Tesla, and Musk’s xAI, eroding Super Micro’s exclusivity.
Super Micro has conceded that it is “under-indexed” with the world’s largest technology companies, called “hyperscalers.” Amazon AWS was a customer but cut ties after delivery issues, per a former employee.
Digital Ocean, a U.S. cloud service provider, switched from Super Micro to Dell after service issues, according to a Digital Ocean employee, describing the relationship as “a train wreck of sorts” fraught with reliability issues.
No position, but never a big fan of the cut-throat commodity server market.
There’s a ton of content in this report as they are using a shotgun approach to make a number of claims. Will be looking into these some more over the coming days to see if anything stands out as potentially valid.
Taking a look into the “Marquee customers leaving” section, every single one of those four claims appears to be primarily sourced directly from Hindenburg Research itself. When I was asking AI tools about these claims, it was pretty much a circular route back to the only sources being Hindenburg themselves.
CoreWeave was Super Micro’s largest customer over the last year, per Bloomberg Intelligence. But in December 2023, Dell announced a deal with CoreWeave for “thousands” of GPU servers, potentially worth over $1 billion.
Other sources have said Meta is the largest customer. Dell announcing a deal with CoreWeave (December 2023) is consistent with the story that Dell is getting overflow orders that Supermicro could not meet for capacity reasons. There are no stories about CoreWeave announcing they are cancelling Supermicro orders, or unhappy with the product. Maybe not coincidentally Bloomberg (mentioned above) was the one who put out fake information in 2018 about Supermicro being infiltrated by the Chinese government.
Tesla had been sourcing its servers exclusively from Super Micro, per Barclays Research in September 2023. But recent reports in May 2024 and posts by Elon Musk show Dell has now won major deals from Tesla, and Musk’s xAI, eroding Super Micro’s exclusivity.
Again this is consistent with the story that Dell is getting overflow orders, and Dell seems to get on a PR campaign anytime they get any win. Supermicro got the other half of the orders. The only source available I am able to find about xAI orders is a story which says both Supermicro and Dell shares jumped as Musk announced they are both supplies for xAI
Super Micro has conceded that it is “under-indexed” with the world’s largest technology companies, called “hyperscalers.” Amazon AWS was a customer but cut ties after delivery issues, per a former employee.
The details about Amazon being unhappy with Supermicro come from claims anonymous employees Hindenburg says they spoke with. They don’t have any timeline on when these supposed conversations happened, but it seems strange to me one of the employees says “Super Micro was in the infancy of that business”, for one of their “early, larger scale rack opportunities”. This seems to imply this was from years ago before AI servers were even invented.
Digital Ocean, a U.S. cloud service provider, switched from Super Micro to Dell after service issues, according to a Digital Ocean employee, describing the relationship as “a train wreck of sorts” fraught with reliability issues.
The only source for any discussion about Digital Ocean’s relationship with Supermicro from AI tools was the Hindenburg report and ironically Saul’s board as this thread appears to be only the second mention on the internet about Digital Ocean and Supermicro’s relationship souring.
These claims about marquee customers leaving seems tenuous at best and they seem to imply that any overflow order Supermicro could not handle for capacity reasons is equated with a customer leaving Supermicro.
Interestingly I also recently saw some articles claiming Dell AI servers have better performance than Supermicro. When I looked into it some more the actual reporting data showed that Dell computers perform better in “offline” mode or not connected to the internet. Looking into that further the site which was making claim used the exact same formatting that Dell did on their official reports so it seemed sponsored by Dell or taken from Dell in one way or another.
The accounting fraud and insider deals to family members sounds like it could be an issue so it’s worth looking into those claims. However, based off this marquee customers section I’m guessing this approach by Hindenburg is based off flooding the narrative with tons of claims that will take days to debunk. Hindenburg has already probably closed their short/put positions as of this morning.
If Hindenburg had Supermicro dead to writes on an accounting issue, I doubt they’d be mentioning any product, customer or reliability issues.
I think the timing of this short report just one day before NVDA earnings is suspect and has a hint of market manipulation.
Supermicro has fared well since the previous short report in January 23 (it was $85 then).
As you say, apart from the big drop pre market today, the market hasn’t worried too much about this Hindenburg report.
I continue to hold. Still, I would like Supermicro to address these points on the report before too long.
“Short interest” implies self-interest, sometimes promoted with press releases and reports. Thanks for checking their facts. Makes me want to buy SMCI, out of self-interest.
Hindenburg went short 2.2 million naked put options with 11 day expiration just as the report came out. I’m sure they’ve closed out with a huge return in one day.
I believe in listening to the bears but this is a short hit piece pure and simple with the goal of creating a one or two day killing for Hindenburg.
Frankly, everything said on that report sounds a lot like Tesla. Hard to be cooking the books or losing marketshare when you had to move $800 million orders to the next quarter because did not have the capacity to deliver this quarter.
Also, difficult to be losing marketshare when you went from 1% to something like 15% )market leader) and now around 80% of the liquid cooled market that just begin to take off from 1% or less of all server shipments to 30% of all shipments starting in June/July of 2024.
Also, if books are being cooked then why is SMCI reporting gross margin declines when they could have cooked the books.
“former employees” talking about losing exclusivity with customers, when everyone knows they can’t come close to meeting demand for any customer is not exactly “mind blowing news”! Of course Dell is getting some orders!
I’m also not a fan of the low-margin server business (see Muji above), but I have bet that margins would NOT be an issue for some time and I am a bit perplexed that margins were an issue in the last ER. I am holding out the idea that it’s due to ramping completely new tech (such as liquid cooling) and working thru the supply chain issues that comes with that, to establish themselves as the clear runaway leader in the highest class of compute. We will see. There’s a reason that Jensen is so cozy with these folks.
This insider dealing is all spelled out in the 10K. They actually track on their 10Q exactly how much money they give to their family members business.
They explained the declining margins during the last earnings call. They gave some deep discounts during the quarter. Yes, they were trying to buy share and apparently they were successful. The remainder of the decline came from expedite charges related to the unanticipated steep ramp up for DLC servers. They were not expecting those to be heavily ordered until Blackwell demanded DLC due to heat dissipation problems associated with the higher power requirements. But, the energy savings with DLC are so large (apx 40%) that orders for DLC Hopper racks pretty much exploded. So discounts and expedite charges drove down margins.
They also asserted that going forward they will exercise more discipline with respect to discounting and felt that the supply chain problems will progressively improve such that they will return to their target 14% - 17% margin range in 2025.
First for me. But I broke one rule with SMCI (I’m out premarket) with small profit but let large profit disappear, is what I’m now calling a maxim: cheap is cheap.
That thought had crossed my mind a few times the last few weeks. But I followed the fundamentals. Can’t follow fundamentals if books are cooked.
Barclays analysts discussed the issue with Super Micro management this morning.
“Management believes that the fundamental business remains strong and that they just need more time to assess internal controls and governance,” said Barclays analysts George Wang and Tim Long, in an investor note. “Management says SMCI is fully in compliance with export controls.”
“Management believes that the fundamental business remains strong and that they `just need more time to assess internal controls and governance,” said Barclays analysts George Wang and Tim Long, in an investor note. “Management says SMCI is fully in compliance with export controls.”
I’m not sure, if that is really convincing and what I want to hear as a shareholder… therefore I am out.
For the most part this board has advocated making decisions based on objective, verifiable information. We all make our decisions with the intention of making good decisions. As I review many of the posts regarding SMCI it appears to me that there’s a lot of knee-jerk emotional reactions. I say that more as an observation than a criticism. We all make our decisions based on how we react to whatever information is at our disposal. Usually, there’s some analysis involved. I don’t see much of that in these posts.
Does anyone think SMCI will become insolvent? I don’t. Does anyone think they will get a bunch of canceled orders due to the short report and/or the late filing? I don’t. Might they have to restate revenues or margins or something else material. OK, that could happen. But honestly, I’m more of a mind that while this certainly does not look good, I’m hard pressed to identify how the business will actually be damaged. Export control violations can be a big deal, but they asserted that they are full compliance with export controls, so I’m inclined to believe that the asserted violation was just another invalid assertion by Hindenberg.
Please feel free to correct me if you think I’m wrong.
From anecdotal experience restated financials become irrelevant to the share price as business has moved two to three quarters ahead and we are already into 2025.
Share price usually reflects the new business and ignores the restated financials. Price usually higher.
The exception is cases of real fraud or corruption and not just negligence with a business growing too fast for management systems to keep up with.
If it is to be believed, $800 million in sales are being moved to next Q. Imagine trying to keep your financials in order with that sort of chaos with business worldwide if you don’t have the proper systems in place. I doubt SMCI focused too much on updated financial accounting when the doors blew off and they couldn’t keep up with demand.
Related transaction fraud is concerning if it exists. One would have to wonder why if business is growing that fast that they’d need to pad sales with related transactions.
When sales are recognized could be an issue, but that is a matter of definition and probably not material to any future business.
Finally, can we trust comment on margins improving?
In general, however @brittlerock’s point is what usually happens absent real fraud and corruption.
I ended up selling all my shares today. While I am doubtful there is intentional fraud, it is hard to understand how they would not be able to prepare the 10-K form in time.
I am thinking it probably has to do with an overly complex way of recognizing revenue where they may do so on certain SKUs when they ship but not others. On the other hand they’ve been in business for 30+ years and should know how to account for the systems they are selling. The company just reported earnings and the 10-K has a ton of boilerplate disclosures where they should be able to plug in numbers from the last time around. I’m just guessing though and that’s part of the problem that my confidence in Supermicro is shot now.
On the short report my guess is Hindenburg Group got a hold of some inside information that Supermicro would announce these delayed financials, and threw a kitchen sink of accusations against them so when they actually did announce it would have this type of impact. Either way the stock was going to get hammered for announcing a financial reporting delay.
This is also after having two supply blocked quarters in a row, a large margin decrease, and a now a delayed financial report. Usually you see these types of delayed reports when it’s a newer company, a small company, there is a new CFO, a complex acquisition, or a change in financial software.
I can see why people would be very tempted to stay in this stock. I don’t think it’s a bad move to hold here, but there is a lot of momentum against the company and the stock right now for valid reasons.
I agree that it is not good that they could not file on time. I’m not sure if the reason they couldn’t is because their auditors are demanding they answer the points in the Hindenburg report before they sign it off, or maybe they had some internal issues that Hindenburg heard about and they knew they would file late so they issued the report? In other words, was the late filing caused by the short report, or was the short report issued because they knew of the potential late filing?
I hovered multiple times over the sell button yesterday, but i still have not sold my oversized holding ( it was 30% before this drop).
I wanted to wait till after Nvidia earnings. Now it appears to be rising again. I’m quite sure Jensen has been in touch with Charles requiring him to get things sorted out too.
At the moment, I cant bring myself to sell until I know for sure from supermicro itself what is actually happening and what is true and what isnt. I can also hear Saul’s reminder in my ears to not fall in love with a stock, and to get out when the story changes. Im not sure it has fully changed yet?
It does help that Ive owned SMCI since May 2023 with an average cost basis of $197.
Ive followed your comments, brittlerock, on SA and have found them helpful to continue to hold my nerves on this and not panic sell.
