The only question that matters
Super Micro walked into this print as the most distrusted name in the AI-hardware complex, and the distrust was never really about whether it could sell servers. The stock closed at $31.60 on into the after-close release — down roughly 30% over the trailing year while the AI trade compounded around it — carrying a market cap near $18.9B on ~601M shares with 30-day IV rank in the mid-70s (unverified — market data, not filing figures). Read the discount honestly and it had two sources, and both were trust, not demand. One was margin: a year of 9–10% gross margins and the bear frame that an AI-server assembler squeezed between Dell, HPE and the Taiwanese ODMs has no pricing power and never will. The other was governance — the delayed-filing and auditor saga that stapled a permanent haircut onto the multiple. That second discount is the reason this print is hard to underwrite: the gap between what management says and what the numbers prove is exactly what SMCI has spent two years teaching investors to worry about. On the company pre-announced that Q4 margin would land far above its own 8.2–8.4% guide, so the direction was public. What was not settled is the only thing that matters — whether the market should pay for it. I don't think it should pay in full yet, and the rest of this is why.
Read the margin and its explanation side by side
Start with the number everyone will quote. Revenue landed at $11.1B, up from $5.8B a year earlier (~+91%) and $10.2B in Q3 — genuinely strong, and near the low end of the $11.0–12.5B range the company had flagged. Now the number that actually moves the equity: GAAP gross margin of 17.5%, versus 9.9% in Q3 and 9.5% a year ago, with non-GAAP gross margin at 17.6%. That is a near-doubling in a single quarter, and it dropped straight through — net income of $1,178M (from $483M in Q3 and $195M a year ago, roughly 6x year-over-year), GAAP diluted EPS of $1.62 (from $0.72 and $0.31), and non-GAAP EPS of $1.70 against a Street near $0.96 and the company's own guide of just 65–70 cents. Operating cash flow was positive at roughly $747M (unverified — from results coverage, pending the filing). A blowout by any measure.
Now decode the explanation, because this is where a forensic read earns its keep. Management attributed the entire jump to a favorable customer and product mix — richer enterprise weighting plus broader adoption of its Data Center Building Block Solutions and direct-liquid-cooling architecture. Sit with what "mix" means when the swing is ~800 basis points against the company's own guide. It is the softest, most reversible line-item on the P&L: it is not a price increase you can point to, not a cost taken out you can verify, not a contract term that persists. It is the weighted average of what happened to ship this quarter. A margin explained by mix is, by construction, a margin that can un-mix the moment the next hyperscale megadeal ships at box economics. That is not an accusation that the number is wrong — it plainly happened. It is the observation that management chose the one explanation that carries no guarantee it repeats, and then asked the market to treat it as permanent.
The rest of this dive is for paid subscribers.
The headline numbers are above. The 8 sections below carry the mechanism, the peer read, the valuation work, and the dated tests that decide it.
- The guide is the tell — management capitalized the margin
- What actually produced the margin — and whether it survives
- How the rest of the chain should read this
- Valuation: the market is pricing the question, not the print
- The strongest case against me
- What decides it from here
- The trade (analysis, not advice)
- Bottom line
Already a subscriber? Sign in.