Date: 2026-03-03 | Analyst: Bear (PaulWBryant) | Quarter: Q3 FY26 (Jan 2026) Position: None (evaluating)
I wrote five open questions on February 22nd when the prelim dropped. Let me be honest about how each resolved.
Prior Belief #1: Q3 GAAP gross margin — did 67.5% hold at $406M revenue? → Updated: 68.5% GAAP. Not only held — expanded. Beat the high end of guidance by 270bps. I could not have been more wrong to worry about this. Fleming guided 63.8-65.8% and delivered 68.5%. Third consecutive quarter of systematic margin guidance undershoot. The "45% non-GAAP" confusion from the prelim was indeed a misattribution to operating margin. Resolved cleanly.
Prior Belief #2: Clarify 45% non-GAAP guidance — operating margin, not gross margin? → Updated: Confirmed. Non-GAAP operating margin: 49.6%. Non-GAAP gross margin: 68.6%. GAAP/NG spread on GM is 10bps — trivially small. The large spread is at the operating line (36.8% GAAP vs 49.6% NG), driven by $52.2M SBC (12.8% of revenue, down from 16.9% Q2). No mystery.
Prior Belief #3: Formal Q4 guidance vs "mid-single-digit sequential" language → Updated: $425-435M. Midpoint $430M = +6.1% QoQ. GM guided 64-66%. Based on the three-quarter pattern (250-270bps beats on GM), I'd expect actual Q4 revenue of $460-490M and GM of 67%+. But that's speculation, not conviction.
Prior Belief #4: Did 5th hyperscaler cross 10%? → Updated: No. 10%+ customers dropped from 4 to 3. Fourth hyperscaler's results were "in line with expectations from 90 days ago" but the largest customer's upside changed the denominator. Top-3 combined: 88% (up from 82% Q2). This gives me pause.
Prior Belief #5: Cash buildup explanation → Updated: ATM equity program. Cash: $813.6M → 1, 301.5M(+487.9M). FCF was 139.7M, leaving 348M from ATM. Raising equity while stock is strong, building a fortress. $1.3B cash, zero debt. Sensible capital allocation.
Summary: All five resolved favorably. Especially the gross margin question, which resolved spectacularly.
| Quarter | Revenue | YoY% | QoQ% | GM% [NG] | OpM% [NG] | NetM% [NG] | FCF ($m) | FCF% |
|---|---|---|---|---|---|---|---|---|
| Q3 FY24 (Jan-24) | $53.1M | -2.2% | +20.7% | ~57% | — | — | — | — |
| Q4 FY24 (Apr-24) | $60.8M | +89.4% | +14.5% | ~58% | — | — | — | — |
| Q1 FY25 (Jul-24) | $59.7M | +70.1% | -1.8% | ~60% | — | — | — | — |
| Q2 FY25 (Oct-24) | $72.0M | +63.6% | +20.6% | ~62% | — | — | — | — |
| Q3 FY25 (Jan-25) | $135.0M | +154% | +87.5% | 63.9% | 31.4% | 33.6% | — | — |
| Q4 FY25 (Apr-25) | $170.0M | +180% | +25.9% | 67.4% | 36.8% | 38.4% | $54.2M | 31.9% |
| Q1 FY26 (Jul-25) | $223.1M | +274% | +31.2% | 67.5% | 43.1% | 44.1% | $51.3M | 23.0% |
| Q2 FY26 (Oct-25) | $268.0M | +272% | +20.1% | 67.7% | 46.3% | 47.7% | $38.5M | 14.4% |
| Q3 FY26 (Jan-26) | $407.0M | +201% | +51.9% | 68.6% | 49.6% | 51.3% | $139.7M | 34.3% |
Every margin line expanded. Every profitability metric at record levels. FCF went from $38.5M to $139.7M — a $101M sequential jump. That's an inflection, not incremental improvement.
| Quarter | Guide Mid ($m) | Actual ($m) | Beat ($m) | Beat % |
|---|---|---|---|---|
| Q3 FY25 | $120 | $135 | +$15 | +12.5% |
| Q4 FY25 | $150 | $170 | +$20 | +13.3% |
| Q1 FY26 | $190 | $223 | +$33 | +17.4% |
| Q2 FY26 | $235 | $268 | +$33 | +14.0% |
| Q3 FY26 | $340 | $407 | +$67 | +19.7% |
Five consecutive beats. Each bigger in dollar terms. FY26 original guide was 800M; they′lldeliver 1.33B. This is rare management credibility. The one blemish — Q4 FY25 miss ($170M vs $185-195M guide) — feels like ancient history now, but I note it because it happened.
Revenue re-acceleration. +20% QoQ → +52% QoQ at this scale. Dollar adds: $45M → $139M sequentially. Fleming on the upside: "our top three customers all grew sequentially from Q2 to Q3." Not one-offs. Not pull-forwards. Just demand.
Operating leverage is textbook. 68.6% NG GM, 49.6% NG op margin, 51.3% NG net margin. For a semiconductor company at $400M/quarter. I've watched a lot of companies and this margin profile is software-like.
FCF inflection is real. OCF 166.2M(record, +104.6M QoQ). CapEx $26.5M (production mask sets). FCF 139.7M.Annualized 560M. No working capital tricks — this is genuine cash generation.
Management credibility. Brennan's call performance was substantive. Direct answers on optical vs copper. Specific customer counts for ZF Optics ("4 in qual, expect >4 in FY27"). No salesman behavior. Fleming's guidance framework: "just north of $1.3B... 50% growth gets you to nearly $2B." Clear. Specific. Trackable.
Product pipeline expansion. AECs (core), PCIe Gen 6 AECs (sampling, production H1 FY27), Blue Heron 200G retimer (Upscale AI win), ZF Optics (production to TensorWave, 4 in qual), ALCs (FY28), OmniConnect (FY28). Multiple growth vectors, not just one product.
I wouldn't be doing my job if I just pounded the table.
| Quarter | Top 1 | Top 2 | Top 3 | Combined | 10%+ Customers |
|---|---|---|---|---|---|
| Q4 FY25 | 61% | 12% | 11% | 84% | 3 |
| Q1 FY26 | 35% | 33% | 20% | 88% | 3 |
| Q2 FY26 | 42% | 24% | 16% | 82% | 4 |
| Q3 FY26 | 39% | 32% | 17% | 88% | 3 |
Top-3 at 88%. Up from 82% in Q2. The 10%+ count dropped from 4 to 3. Yes, individual relationships are rotating healthily — no single customer above 40%, and Customer #2 surged 8 points. But if Customer #1 at 39% has a bad quarter, that's ~$160M at risk. The concentration is real.
I said it in the MU thread: "I think ALAB and CRDO, MU, and most hardware companies" face the question of exit timing. CRDO is executing brilliantly right now. So was AEHR before it fell off a cliff. So was Zoom. The question isn't whether current results are good — they're exceptional. The question is: when does hyperscale AI infrastructure build plateau? I don't know the answer, and I'm skeptical anyone does. This keeps my position sizing conservative.
Q4 guide midpoint $430M = +5.6% QoQ after +52%. Even with a likely beat to $460-490M (+13-20% QoQ), that's a significant step-down. Fleming framed FY27 as "mid-single-digit sequential quarterly growth." That means +5-7% QoQ through FY27. The market may not reward that after the Q3 fireworks.
The 201% YoY growth rate is not sustainable. Even Fleming says FY27 is >50% YoY. The relevant comparison is ~50-60% growth at current multiples, not 200%. At $20B market cap and $2B FY27 revenue with ~35% NG net margin, you get $700M net income, or ~29x. Reasonable if $2B is a floor. Not reasonable if it's a peak.
I think the panic is largely wrong — the Lumentum/Coherent deals are about optical components (lasers, silicon photonics), not copper AECs. Brennan addressed it directly: "where you can use copper, you will use copper." The deals are nonexclusive. CRDO's ZF Optics competes on reliability/telemetry, not raw photonics. Different value axes.
But the market's concern isn't entirely stupid. NVIDIA investing $4B in optical connectivity suggests the direction of travel is toward more optical, not more copper. CRDO's ZF Optics is an answer to this, but it's pre-revenue at scale. The core business is copper. If the industry moves faster than expected to optical for shorter reaches, copper's TAM contracts.
I'd feel better if NVIDIA had also announced a CRDO partnership. They didn't. Shows what I know about reading tea leaves — maybe it means nothing, maybe it means something. I'll watch.
| Metric | Value |
|---|---|
| Market Cap | ~$20.3B |
| EV | ~$19.0B |
| Run-rate Revenue (Q3 x4) | $1.63B |
| Run-rate P/S | 11.7x |
| Run-rate P/E (GAAP) | 32x |
| Run-rate P/E (NG) | 24x |
| EV/FCF (run-rate) | 34x |
| Revenue Growth / P/S | 17.2x |
| FY27 P/S (on ~$2B) | ~10x |
| FY27 EV/FCF (on ~$600M est.) | ~32x |
By my revenue-PEG heuristic, even at the sustainable 50-60% growth rate, 11.7x P/S is reasonable. Not a screaming buy at $20B market cap, but reasonable. Compared to ALAB at ~36x P/S with negative FCF, CRDO is meaningfully cheaper on every metric.
Can this be a 40Bcompanyin2 − 3years?At 2B FY27 revenue with 35% NG net margins, you'd need a 20x P/S or 57x P/E. Possible if growth sustains into FY28. Not guaranteed.
ESTABLISHING → INTACT
All five prior beliefs resolved favorably. The numbers match the theory. Management is credible. The business is generating real cash flow.
But I'm not ready for "Strengthening" yet:
I could be wrong. This might be one of those companies where the numbers are so good that my caution costs me returns. It's happened before. But I don't invest based on hope.
If I were building a position, I'd start at 4-5% here. Not 10%+. The numbers warrant ownership. The risks warrant sizing discipline.
I'd add on:
I would NOT go above 10% until:
Brennan on AEC dominance: "Our AEC product line once again delivered strong growth, driven by existing customers and new wins, including our fifth hyperscaler."
Brennan on copper vs optical: "They [NVIDIA] have been really outspoken that where you can use copper, you will use copper. The reasons are very basic — reliability, number one; power efficiency, number two; and ultimately, total cost of ownership, number three."
Brennan on ZF Optics: "My strong expectation is that we will ramp more than four [customers in FY27]. We have four in qual now."
Fleming on FY27: "Just north of $1,300,000,000 [FY26]. Fifty percent growth gets you to nearly $2,000,000,000 for next year."
Fleming on Q3 upside: "Our top three customers all grew sequentially from Q2 to Q3."
Fleming on GM conservatism: "We are conservative in the way we forecast... long-term expectation in the 63% to 65% range."
Brennan on optical narrative: "There has been a bit of a signal-to-noise ratio issue in the market, and the noise right now is dominating the signal."
Brennan on 1.6T: "NVIDIA is going to lead the charge with Vera Rubin, but many other customers will follow on a slower timeline."
Bear