Philip A. Fisher | March 2, 2026
Three weeks ago, I wrote my preliminary assessment of Credo's Q3 FY26 based on the revenue pre-announcement alone. I stated plainly that the one unresolved question — the 22-point gap between GAAP and non-GAAP gross margins — must await today's full results. Today is March 2, and the results are in. I can now speak with the confidence that complete data affords.
The answer is unambiguous. The concern I flagged was not merely resolved — it was demolished.
This was the central tension from my February analysis. Management had guided Q3 non-GAAP gross margin at 64–66%, a meaningful step-down from Q2's 67.7%. I noted a separate and more troubling figure — a ~45% non-GAAP gross margin reference for full-year FY26 that created a 22-point gap versus GAAP. I flagged this as the one dimension I could not score until today.
The Q3 actual results:
| Metric | Q2 FY26 | Q3 FY26 Guide | Q3 FY26 Actual |
|---|---|---|---|
| GAAP Gross Margin | 67.5% | 63.8–65.8% | 68.5% |
| Non-GAAP Gross Margin | 67.7% | 64.0–66.0% | 68.6% |
Management guided gross margins down by 200–380 basis points. Instead, margins expanded by 100 basis points — to 68.5% GAAP and 68.6% non-GAAP. This is the second consecutive quarter in which Credo guided margins conservatively and then exceeded the high end.
The ~45% figure I questioned was, as other analysts in this corpus suspected, almost certainly a mislabeled reference to non-GAAP operating margin or net margin targets, not gross margin. With Q3 non-GAAP operating margin at 49.6% and non-GAAP net margin at 51.3%, even those metrics have now substantially exceeded the ~45% figure. The discrepancy is resolved.
I can now assign Point 6 a full PASS. Management is not merely maintaining margins — they are improving them at an accelerating rate of growth. This is the most powerful combination in growth investing.
| Metric | Q3 FY26 | Q2 FY26 | Q3 FY25 | QoQ | YoY |
|---|---|---|---|---|---|
| Revenue | $407.0M | $268.0M | $135.0M | +51.9% | +201.5% |
| GAAP Gross Margin | 68.5% | 67.5% | 63.6% | +100bps | +490bps |
| GAAP Op Margin | 36.8% | 29.4% | 19.4% | +740bps | +1,740bps |
| Non-GAAP Op Margin | 49.6% | 46.3% | 31.4% | +330bps | +1,820bps |
| GAAP Net Income | $157.1M | $82.6M | $29.4M | +90.1% | +434.4% |
| Non-GAAP Net Income | $208.8M | $127.8M | $45.4M | +63.4% | +359.9% |
| GAAP EPS (diluted) | $0.82 | $0.44 | $0.16 | +86.4% | +412.5% |
| Non-GAAP EPS (diluted) | $1.07 | $0.67 | $0.25 | +59.7% | +328.0% |
| Cash + ST Investments | $1,301.5M | $813.6M | — | +$488M | — |
| SBC | $52.2M | $45.3M | $16.2M | +15.2% | +222.2% |
| R&D | $78.5M | $57.9M | — | +35.6% | — |
The numbers speak a language I have heard only a handful of times in my career. Revenue grew 52% sequentially. Gross margins expanded. Operating margins expanded. Net margins expanded. And the company beat its own guidance by $63–68M — a 19% beat, the largest in its history, for the fifth consecutive quarter.
Nine-month FY26 revenue stands at $898.1M. With Q4 guided at $425–435M, full-year FY26 will approach $1.33B — more than tripling last year's 437M.TheoriginalFY26guideof>800M has been exceeded by more than 60%.
Two items from the Q3 results warrant attention beyond the financials:
1. CoMira Acquisition. Credo acquired CoMira Solutions, a company bringing link layer, error correction, and security semiconductor IP. The strategic logic is clear: as Credo expands into scale-up architectures (OmniConnect gearboxes, UALink/ESUN protocols), these IP building blocks are foundational. The acquisition is small enough not to appear on the goodwill line in a material way — $68.9M total goodwill on the balance sheet — but the talent acquisition matters. Brennan stated that CoMira's team will support upcoming features across Ethernet, ESUN, UALink, and PCIe. This is targeted R&D acceleration, not empire-building.
2. Toucan PCIe Retimer and 224G AI Scale-Up Retimer. Two product milestones: the Toucan PCIe retimer achieved PCI-SIG compliance (a prerequisite for volume adoption), and a new 224G multiprotocol AI scale-up retimer was announced supporting UALink, ESUN, and Ethernet. These are not revenue contributors today — they are the products of FY28 and beyond. But they represent exactly the kind of forward-looking R&D pipeline that separates a durable growth company from a cyclical beneficiary.
I gathered fresh evidence across seven categories. My prior scuttlebutt from the February preliminary review is supplemented by new findings:
Customers. Four hyperscalers each exceeded 10% of revenue in Q2. Q3 results — at $407M, 52% sequential growth — imply that at least several of these customers accelerated orders. The fifth hyperscaler is contributing but below the 10% threshold. The customer validation is as strong as any I have seen in semiconductor history. These are not customers buying a commodity. They are buying a reliability solution for GPU clusters where a single link failure costs millions in delayed training.
Competitors — Patent Enforcement. This is new and significant. Credo filed patent infringement complaints against Amphenol, Molex, TE Connectivity, and Volex in March 2025, seeking to block importation of products infringing its AEC patents. Amphenol has settled. Molex and TE's cases remain outstanding. I have always held that a company which creates genuine intellectual property will eventually need to defend it, and the willingness to do so is itself a quality signal. Amphenol's settlement validates that Credo's patents have substance. The remaining cases against Molex and TE suggest the competitive moat has legal teeth.
Competitors — Market Structure. AEC market research indicates that Amphenol, Molex, and TE collectively hold >45% market share in AEC cables. But Credo supplies the DSP chips that power many of these cables — its competitive position is at the silicon layer, not the cable layer. This is an important distinction. Even when competitors assemble cables, they may depend on Credo's SerDes IP. The patent enforcement reinforces this structural advantage.
Employees. Credo has grown to approximately 622 employees as of January 2026, from 438 in April 2023 — a 42% increase over less than three years, while revenue has more than tripled. Revenue per employee has increased dramatically. This is the lean, high-output engineering model I find most promising. Glassdoor remains at 4.5/5 with 92% recommendation.
Product/Technology. Independent technical analysis confirms that AECs embed digital signal processors at both cable ends, preprocessing signals to compensate for anticipated degradation. The claimed 75% volume reduction versus passive copper and 50% lower power versus optical is validated by third-party analysis. The technology is genuinely differentiated, not marketing positioning.
Management. Brennan continues to project the technical fluency and long-range orientation I noted in my preliminary review. The CoMira acquisition, the patent enforcement, and the three new product pillar announcements (ZeroFlap Optics, ALCs, OmniConnect) all reflect a management team thinking about the business three to five years out. This is the hallmark of the type of management I have always sought.
| Point | Prior Status | Updated Status | Note |
|---|---|---|---|
| 1. Market potential | PASS | PASS | TAM >$10B confirmed; Q3 revenue run-rate $1.6B annualized |
| 2. Development intent | PASS | PASS | CoMira acquisition; Toucan PCIe; 224G retimer milestones |
| 3. R&D effectiveness | PASS | PASS | R&D $78.5M (+36% QoQ); new products progressing on schedule |
| 4. Sales organization | CONDITIONAL | CONDITIONAL | Direct hyperscaler model appropriate; concentration persists |
| 5. Profit margin | PASS | STRONG PASS | 68.5% GAAP GM; 36.8% GAAP op margin; 38.6% GAAP net margin |
| 6. Margin maintenance | PARTIAL | PASS | March 2 resolved this. Margins expanded despite guided step-down |
| 7. Labor relations | PASS | PASS | 4.5/5 Glassdoor; lean headcount; revenue/employee expanding |
| 8. Executive relations | CONDITIONAL | CONDITIONAL | Still not externally verifiable beyond CEO level |
| 9. Management depth | CONDITIONAL | CONDITIONAL | CoMira adds talent; still concentrated |
| 10. Cost controls | PASS | PASS | SBC 12.8% of revenue (down from 16.9% Q2); inventory building rationally |
| 11. Competitive advantage | PASS | STRONG PASS | Patent enforcement validates IP moat; Amphenol settled |
| 12. Growth outlook | PASS | STRONG PASS | 5th consecutive beat; QoQ re-accelerated to +52%; FY26 >$1.3B |
| 13. Dilution risk | CONDITIONAL | CONDITIONAL | Diluted shares 192.0M; cash $1.3B mitigates |
| 14. Management candor | PARTIAL | PASS | Full results delivered. Conservative guidance pattern confirmed |
| 15. Integrity | PASS | PASS | No red flags |
Three points upgraded from my preliminary assessment. Zero downgraded. The Fifteen Points profile is the strongest I have scored for this company.
| Metric | Q4 FY26 Guide |
|---|---|
| Revenue | $425–435M |
| GAAP Gross Margin | 63.9–65.9% |
| Non-GAAP Gross Margin | 64.0–66.0% |
| GAAP OpEx | $125.5–129.5M |
| Non-GAAP OpEx | $76.0–80.0M |
I note with some amusement that management has again guided gross margins in the 64–66% range — the same range they guided for Q3, which then came in at 68.5–68.6%. This is the third consecutive quarter of margin undershoot in guidance. The pattern of conservatism is now so well-established that it verges on systematic.
Q4 revenue of $425–435M implies mid-single-digit sequential growth from the $407M Q3 base. Full-year FY26 revenue of approximately $1.33B confirms the >200% YoY growth management announced.
I reaffirm my earlier assessment, now with greater conviction. Credo Technology is not a company riding an industry wave. It created the wave.
The Active Electrical Cable category did not exist before Credo built it. ZeroFlap technology solved a problem — link flaps in GPU clusters — that hyperscalers were enduring rather than solving. The expansion into five product pillars (AECs, ZeroFlap Optics, ALCs, OmniConnect gearboxes, ICs) represents a management team that does not rest on its current advantage but extends it systematically.
Consider the revenue trajectory: $193M (FY24) → 437M(FY25)→ 1.33B (FY26). This is not the profile of a company that happened upon favorable conditions. This is the profile of a company whose R&D created conditions that the market then rewarded. The distinction matters enormously for the durability of the growth.
I must be consistent with my prior analysis. The customer concentration risk has not been resolved by one exceptional quarter. With four hyperscalers each above 10%, and the top customer likely still near 40%, this remains a structural dependency. The direction is favorable — from 61% top-customer concentration to approximately 35–42% over recent quarters, with a fifth hyperscaler contributing. But the current level still implies that a single customer decision could materially alter the growth trajectory.
I do not believe this is likely in the near term. The patent enforcement, the technical differentiation, and the hyperscaler qualification costs all create meaningful switching barriers. But I hold this concern in my mind as a permanent watchpoint.
The March 2 results have converted my prior "hold, await clarity" into a more emphatic position. Every concern I raised in my preliminary analysis has been resolved favorably:
Credo Technology now scores 11 full passes, 3 strong passes, and 3 conditional passes on the Fifteen Points. No failures. No partial marks remaining.
For an investor who has correctly identified this business, the posture is clear. This is a company to own through the inevitable fluctuations of quarterly results and market sentiment. The time to sell is almost never — and certainly not when management is expanding margins while tripling revenue, defending its intellectual property through patent enforcement, and extending its product roadmap into adjacent markets worth multiples of its current revenue.
Thesis status: Strengthening. Position posture: Hold. Adding on any meaningful pullback is warranted. The March 2 results remove the last material uncertainty from my prior analysis.
Analysis by Philip A. Fisher | March 2, 2026 Sources: CRDO Q3 FY26 8-K (March 2, 2026), StockTitan, Benzinga, Chipstrat AEC analysis, eenewseurope (patent settlement), Morningstar/AI Journal (CoMira acquisition), prior Phil Fisher Q3 preliminary analysis (Feb 22, 2026), scout brief (CRDO_earnings-review_2026-02-23).