SMCI
Super Micro Computer (SMCI): Real AI Infrastructure Growth, Priced at a Premium to Peers, Clouded by Legal Overhang
- Sector
- Data Center Infrastructure
- Date
- Jun 23, 2026
Speculative Buy (Initiated)
Real, rapid AI infrastructure growth, but the stock trades at a premium to peers once net debt is properly included, with legal overhang still unresolved.
Thesis
Supermicro is caught in a different kind of trap than the typical cheap-versus-peers story I’m used to writing about. Q3 FY2026 revenue grew 123% year over year, and FY2026 guidance of $38.9B to $40.4B points to continued rapid expansion off a $33.7B trailing twelve month base. That growth is real. But I want to be direct about something the market cap alone doesn’t tell you: once you properly include roughly $7.5B of net debt, the actual enterprise value runs closer to $29B to $30B. That puts the EV/EBITDA multiple at roughly 16x to 18x against TTM EBITDA of $1.6B to $1.8B. That is a premium to the peer median, not the in line multiple a market cap only read would suggest.
So this isn’t a valuation discount waiting to close. The case here rests on growth and demand visibility justifying a premium that’s already baked into the price.
Gross margin tells its own story. It bottomed at 6.3% in Q2 FY2026 and recovered to 9.9% in Q3. I read that as an active recovery, not a permanent floor, though it’s still well below where this business has historically run. The 19% sequential revenue decline in Q3 looks scary on its own, but it reflects customer site delays pushing deployments into future quarters, not weakening demand. The order backlog and full year guidance both point the other direction.
In June 2026, Supermicro announced a $39B order backlog alongside a $7B equity raise. That combination tells you two things at once: real demand, and real dilution and execution risk through FY2027.
A 52 week range of $19.48 to $62.36, on a beta of 2.58, shows just how violently this stock has already been re-rated in both directions. This is not a calm name.
Two distinct legal threads run through this stock, and I think it’s important to treat them as genuinely separate issues. The 2024-25 accounting and auditor crisis is resolved. A newer, unresolved March 2026 DOJ indictment of a co-founder is not, and I cover both in detail below.
One more thing worth saying plainly: AI GPU based platforms make up more than 80% of Supermicro’s revenue, which ties growth directly to NVIDIA’s GPU allocation and supply. That’s a real concentration risk. The counterweight is that enterprise channel revenue grew to 28% of the Q3 total, up from 15% the quarter before, a genuine sign that demand is starting to broaden beyond hyperscaler GPU platforms even if the headline numbers stay NVIDIA dependent for now.
Key Metrics
| Metric | Value |
|---|---|
| Price | $33.91 |
| Market Cap | ~$22B |
| TTM Revenue | $33.7B |
| TTM EBITDA | ~$1.6-1.8B (~5% margin) |
| Gross Margin, Q2 to Q3 FY26 | 6.3% to 9.9% (active recovery, not a floor) |
| Q3 FY26 Revenue Growth | +123% YoY (-19% QoQ, customer site delays, not demand) |
| FY2026 Revenue Guidance | $38.9-40.4B |
| Net Debt | ~$7.5B |
| EV (Recalculated) | ~$29-30B |
| EV/EBITDA (Corrected) | ~16-18x (premium to peer median) |
| 52-Week Range | $19.48 - $62.36 |
| Beta | 2.58 |
| New Order Backlog | $39B (alongside $7B equity raise, June 2026) |
Comps
| Company | Notes | EV/EBITDA | P/E |
|---|---|---|---|
| Dell | N/A | 13.8x | 32x |
| HPE | Fwd P/E; declining revenue | 10.6x | 9x |
| NetApp | Fwd P/E | 13x | 18x |
| Pure Storage | Minimal current earnings | 113x | Growth multiple |
| Lenovo | EV/EBIT, not EV/EBITDA; thin analyst coverage | 7.5x | N/A |
| SMCI | Premium to peer median on corrected EV/EBITDA (net debt included) | ~16-18x | ~18-25x |
Once you properly account for roughly $7.5B of net debt, Supermicro’s enterprise value runs closer to $29B to $30B, which implies an EV/EBITDA multiple of roughly 16x to 18x. That’s a premium to the peer median, not the in line read a market cap only comparison gives you. This is a different setup than the typical undervaluation thesis I usually write. The case here rests on growth and demand visibility justifying a premium that’s already in the price, not on the market closing a gap.
If I apply the peer median EV/EBITDA of roughly 12x to 13x, excluding Pure Storage’s outlier multiple, to Supermicro’s current EBITDA, I get an implied enterprise value of roughly $19B to $23B. That’s well below the $29B to $30B EV the stock commands today. My bull case rests on forward growth closing that gap. 123% year over year revenue growth in the most recent quarter, a $39B order backlog, and FY2026 guidance of $38.9B to $40.4B all point toward growing into the multiple rather than the multiple already being supported by where earnings sit right now.
Valuation Range
Here’s my honest range, not a single confident number. I don’t think a tight point estimate is the right call on a name with this much binary risk attached to it.
Downside: if the premium compresses back to the peer median on current EBITDA, implied enterprise value lands around $19B to $23B. Subtract the $7.5B of net debt and that implies a market cap of roughly $11.5B to $16B, well below today’s $22B.
Upside: if FY2026 plays out the way management is guiding and the market sustains something close to today’s multiple as earnings grow into it, there’s real room above current levels.
The width of that range isn’t me hedging. It reflects genuine binary risk sitting in the legal overhang below. A tighter number would overstate confidence I don’t actually have right now.
Legal & Governance Overhang
I want to be upfront that this is the third episode like this for Supermicro, not the second. Most coverage treats the recent crisis as an isolated event, and I don’t think that’s the right read. An earlier SEC settlement in 2020 resolved accounting matters going back to fiscal years 2014 through 2017, well before the 2024-25 crisis or the indictment below. Three episodes over a decade is a pattern, not bad luck.
The 2024-25 accounting and auditor crisis, a delayed 10-K, the auditor’s resignation, and a stretch of Nasdaq non-compliance, is actually resolved now. Supermicro regained Nasdaq compliance in January 2026.
What isn’t resolved: in March 2026, the DOJ indicted co-founder and board member Yih-Shyan Liaw, along with two others, on charges tied to alleged export control violations and chip smuggling to China. There’s also a related securities fraud class action against the company still working through the courts. Neither has a clean resolution yet, and both carry the kind of binary, stock moving risk that has nothing to do with how the underlying business is actually performing.
Catalysts
$39B order backlog execution through FY2027. NVIDIA Vera Rubin NVL4 platform ramp. Potential resolution of the legal overhang.
Risks
The risks here aren’t subtle. Continued margin compression from price war dynamics among server OEMs is real. So is dilution from the $7B equity raise. This stock is also genuinely volatile, a 2.58 beta and a 52 week range from $19.48 to $62.36 say that clearly. There’s concentration risk too: AI GPU based platforms are over 80% of revenue, tying growth directly to NVIDIA’s GPU allocation and supply. And there’s the unresolved legal exposure tied to the DOJ indictment and the related securities litigation, sitting on top of a now three episode pattern of governance issues. I’m not glossing over any of that to make the growth story look cleaner than it is.
Status
This is a live position for me, underwritten on an 18 to 24 month horizon rather than a quick catalyst trade. I still believe the core call here: Supermicro’s AI infrastructure growth is real, and it’s continuing. But my conviction and how I’d size this are both calibrated to the unresolved legal overhang and the recurring governance pattern. Either one could move this stock sharply in either direction, completely independent of how the business itself is doing.