In short: News mention ("Stocks I'm Watching"): server designers higher as the AI trade continues to find strength.
In short: Openly disbelieved. "Super Micro said an independent investigation found no evidence that current senior management knew about the alleged export diversion scheme involving two former employees and a contractor, basically selling Nvidia GPUs to China. Well, I'm sure that's true. Huh." Consistent with the prior week's treatment of the name as uninvestable on governance rather than on end-market.
Super Micro builds the servers that house Nvidia's AI chips — a good business in a good market. The problem is trust in what the company says.
This week it announced that an independent investigation had found no evidence that current senior management knew about an alleged scheme to divert Nvidia GPUs to China in breach of export controls, blaming two former employees and a contractor. Singh's entire response is four words of sarcasm: "Well, I'm sure that's true. Huh."
The point is one he has made about this name before: being in the right industry does not compensate for not being able to rely on the numbers or the disclosures. When an investigation a company commissions into itself clears its own management, it is not the same as independent proof.
Full passage: premium transcript (PDF).
In short: The worked example of the index-graduation derating — a case study, not a call. It was Haymaker's 2024 illustration of why mid-caps work: "one of the reasons we cited for preferring Mid-Caps in the late winter of 2024 was that this style often catches companies as their growth accelerates and their stock price reflects this uplift. The example we used back then was Super Micro Computer (SMCI)." Then the timing: "our suspicion was that it would soon be pulled up to the S&P 500 out of the Mid-Cap index because of its explosive market cap increase. That happened almost as soon as we hit 'Send'. To say the timing was unfortunate, at least for S&P 500 index investors, is a considerable understatement." And the outcome: "from the time of its 'graduation' to the Large-Cap index, SMCI has experienced a bit of a derating… like by nearly 70%." (A 10:1 split is noted; the split-adjusted peak was "roughly $114.") The generalisation is the point: "of course, it was a small part of the S&P so the cost of this value implosion to its shareholders was negligible. But when multiplied by dozens, if not hundreds, of these events, that can add up to a material performance drag over time." No view is taken on SMCI's shares today.
Super Micro Computer builds the server hardware that AI data centres run on. It appears in this note not as a recommendation but as a cautionary case study — and, awkwardly, one of Haymaker's own.
In early 2024 Haymaker used it to illustrate why medium-sized companies are a good hunting ground: they are big enough to be real businesses but small enough that a growth acceleration moves the share price a lot. SMCI was doing exactly that. Haymaker also predicted what would happen next — its market value would grow so fast that it would be promoted out of the mid-cap index into the S&P 500.
It was, "almost as soon as we hit 'Send'" — and that promotion marked the top. Since graduating into the S&P 500 the stock has fallen nearly 70%. So the mid-cap index captured the run up and handed the collapse to the large-cap index, whose funds were obliged to buy at the peak.
Hay is careful to size the damage honestly: SMCI was a small piece of the S&P, so the direct cost to index holders was negligible. His point is cumulative — this happens over and over, "multiplied by dozens, if not hundreds, of these events," and the sum is a real long-run drag on the large-cap index. No opinion is offered on whether SMCI is worth owning today.
In short: An explicit avoid: 4Q26 revenue $11.1B against $11.55B expected, though EPS beat at $1.70 vs $0.96 — "the stock's too risky to get involved in."
Super Micro builds the servers that house AI chips — a business squarely in the path of the biggest spending boom in technology. It still missed on revenue, at $11.1 billion against $11.55 billion expected, even though earnings per share beat.
Singh's dismissal is one line: "the stock's too risky to get involved in." Given the accounting investigation he flagged on the previous week's call, the point is that being in the right industry does not compensate for not being able to trust the numbers.
Full passage: premium transcript (PDF).
In short: "Super Micro has had problems executing of late. However, this was a good quarter" — better revenue and EPS with strong guidance; revenue +91%, EPS $1.70 "up 315% versus last year," stock +19% on Wednesday. Sells the servers that go into data centers; chief competitor is Dell.
Super Micro builds the servers that fill data centres; Dell is its main competitor. After a stretch of execution stumbles, this was a clean quarter: revenue up 91%, earnings per share of $1.70 up 315%, and strong guidance — the stock also rose 19% on the day.
It belongs to the same category as CoreWeave in his framework: a pure beneficiary of hyperscaler capital spending. Not a moat story, a flow story — which is why he immediately re-attaches the caveat that the whole category depends on the Anthropic/OpenAI risk staying dormant.
15:00Super Micro has had problems executing of late. However, this was a good quarter. Super Micro reported better revenue and EPS and issued strong guidance. Revenue was up 91% versus last year. EPS of $1.70 was up 315% versus last year and the stock was up 19% on Wednesday. The results of Cororeweave and Super Micro should not be a surprise.
In short: "After the close, we have Super Micro, which has had some accounting issues and an investigation" — flagged as under scrutiny into the print rather than as an opportunity.
Super Micro builds the server hardware that AI data centres run on, so it should be a direct beneficiary of the build-out. The reason Singh flags it negatively is unrelated to demand: the company has had accounting problems and is under investigation. That is a governance risk that no amount of order growth compensates for.
Full passage: premium transcript (PDF).
In short: Giroux: named among the losers of multi-silicon orchestration — the neocloud/AI-server value chain erodes as Microsoft, Meta and Amazon in-source the business over time.
In short: "Uninvestable at this point" — Taiwan's Keelung prosecutor expanded a smuggling probe (raids across 12 sites, two SMCI employees detained, nine under investigation) over forged shipping docs to route ~50 Nvidia-equipped AI servers to China/HK/Macau; Taiwan is fast-tracking a law to criminalize chip diversion.
Super Micro builds AI servers. Taiwanese prosecutors widened a criminal smuggling investigation into it — raids across a dozen locations, employees detained — accusing a ring of forging shipping papers to sneak ~50 Nvidia-powered servers into China. With Taiwan now rushing to criminalize exactly this kind of chip diversion, Singh calls the stock flatly "uninvestable."
Full passage: premium transcript (PDF).
In short: An AI-derivative hardware (server) name — announced a $7B raise via equity + equity-linked financing; "$7B off a market cap of roughly $20B is no small thing." Stock down 28% on the news. More capital-intensity evidence.
7:22Now, its fiscal 2027 capital plan is for 40 billion in equity and debt. Like I said before, the tech industry is being transformed from a capital light model to a model of insatiable need for capital and investors are starting to lose their taste for it. After hours Wednesday night, Oracle was down 10%. And in more capital intensity news, Super Micro, which is admittedly not a software company, but a tech hardware company that largely sells servers, announced plans to raise $7 billion through a combination of equity and equity linked
In short: A small vol trade — sold the 10-strike puts at ~300 vol (already profitable). Reads the E&Y resignation/delisting fear as overblown (likely minor early-revenue booking, ~$27M last time vs billions, not WorldCom/Enron); 75% cooling-rack share, ~$1.7M/rack, generates ~$300M cash, repaid the BofA loan. Not buying the common — institutions can't until a 10-K/relisting.
Super Micro builds the servers and liquid-cooling racks that house AI chips. Its stock cratered after its auditor (Ernst & Young) resigned and a delisting from the Nasdaq looked possible. Instead of buying the stock, the house made a small bet by selling put options — getting paid a rich premium in exchange for agreeing to buy the stock at $10 if it falls there.
The view is that the accounting fears are overblown (likely a minor timing issue, not a fraud like Enron or WorldCom), the company has 75% market share in cooling racks and generates about $300 million of cash, and it already repaid a big bank loan. So they pocket the option premium, keeping the position tiny because a delisting is still a real risk. It's a trade, not a long-term holding.
Nothing matches this filter.
Verbatim excerpts from the public transcripts (auto-pulled at each mention's timestamp, lightly cleaned). Timestamps link into the video; "source page" opens that commentator's full analysis at this row. Click a mention's header line to expand it (one open at a time). For personal study — not investment advice.