CRDO — Q2 FY26 Earnings Review (Atlas)

Date: 2026-02-23 Quarter: Q2 FY26 (October 2025) | Full results: Dec 1, 2025 Q3 FY26 Preliminary: 404 − 408Mpre − announcedFeb9, 2026(fullresultsMarch2, 2026)Marketcap: 23.3B | EV/TTM Rev: ~28x | Revenue growth: +272% YoY

Verdict

Q2 FY26 delivered another step-up in both revenue and profitability — $268M (+272% YoY), record non-GAAP operating margin of 46.3%, and a 5th consecutive guidance beat. The more important signal is the Q3 FY26 pre-announcement: $404-408M vs $335-345M guided — a 63 − 68MbeatthatrepresentsQoQre − accelerationfrom + 201.3B vs an original $800M guide. Five pillars, four hyperscalers at 10%+, zero debt, $813M cash. The only open question is whether Q3 margins held at ~67% gross or compressed as guided (63.8-65.8% GAAP). Answer comes March 2. Conviction 4/5.

Qualification Gate

Six-Factor Score

The Numbers

| | Q3_FY23 | Q4_FY23 | Q1_FY24 | Q2_FY24 | Q3_FY24 | Q4_FY24 | Q1_FY25 | Q2_FY25 | Q3_FY25 | Q4_FY25 | Q1_FY26 | Q2_FY26 | | | Jan-23 | Apr-23 | Jul-23 | Oct-23 | Jan-24 | Apr-24 | Jul-24 | Oct-24 | Jan-25 | Apr-25 | Jul-25 | Oct-25 | |---|---|---|---|---|---|---|---|---|---|---|---|---| | Revenue ($M) | 54.3 | 32.1 | 35.1 | 44.0 | 53.1 | 60.8 | 59.7 | 72.0 | 135.0 | 170.0 | 223.1 | 268.0 | | YoY % | +71% | -14% | -25% | -14% | -2% | +89% | +70% | +64% | +154% | +180% | +274% | +272% | | QoQ % | +6% | -41% | -35% | +25% | +21% | +15% | +12% | +21% | +88% | +26% | +31% | +20% | | GM % [GAAP] | 58.9% | 57.9% | 59.3% | 59.3% | 61.4% | 65.8% | 62.5% | 63.2% | 63.6% | 67.2% | 67.4% | 67.5% | | GM % [NG] | 59.5% | 58.3% | 59.8% | 60.0% | 62.1% | 66.1% | 63.0% | 63.6% | 63.9% | 67.4% | 67.5% | 67.7% | | Op Margin [GAAP] | -5.3% | -51.1% | -41.0% | -20.2% | -11.1% | -13.0% | -24.3% | -11.7% | +19.4% | +19.9% | +27.2% | +29.4% | | Op Margin [NG] | +12.3% | -26.5% | -18.2% | -1.6% | +4.5% | +12.3% | +3.7% | +11.5% | +31.4% | +36.8% | +43.1% | +46.3% | | Net Margin [GAAP] | +5.2% | -17.8% | -33.3% | -15.0% | +0.8% | +19.4% | -15.9% | -5.8% | +21.8% | +38.4% | +28.4% | +30.8% | | EPS [NG] | 0.05|−0.04 | -$0.03 | $0.01 | $0.04 | $0.07 | $0.04 | $0.07 | $0.25 | $0.38 | $0.52 | 0.67||FCF(M) | — | -15.7 | +19.3 | — | — | +1.0 | -13.1 | — | — | +54.2 | +51.3 | +38.5 | | FCF Margin % | — | — | — | — | — | +1.6% | — | — | — | +31.9% | +23.0% | +14.4% | | SBC ($M) | 5.2 | 7.9 | 8.0 | 8.1 | 8.3 | 14.6 | 16.6 | 16.7 | 16.2 | 27.9 | 35.5 | 45.3 |

Q3_FY26 Preliminary (Jan-2026): Revenue 404 − 408M(mid 406M) vs guide 335 − 345M → beat+66M / +19.4%. QoQ re-acceleration: +51%. Full margins reported March 2, 2026.

FCF note: Q2 FY26 FCF = $38.5M (per CFO: $61.7M OCF - $23.2M CapEx). The DB shows $51.3M for Q2 — this appears to be a data error (Q1 figure repeated). CFO transcript is authoritative.

Prior Beliefs / Updated Beliefs

Prior beliefs (entering Q2 FY26 results):

• Revenue: ~$240M (midpoint of 230 − 240Mguidance)—+28M guide midpoint
• Non-GAAP GM: 64-66% (guided; a meaningful step-down from Q1's 67.5%)
• Non-GAAP op margin: continued expansion, maybe 43-44%
• Customer concentration: 3 customers at 10%+, with fourth approaching threshold
• FCF: ~$45-55M
• Beat magnitude: expected +5-10% above guide, consistent with prior quarters

Delta assessment:

The GM "step-down" that management guided never materialized. Q2 NG GM came in at 67.7% vs a 64-66% guide — Credo sandbagged margins as aggressively as it sandbags revenue. The FCF miss is the only real number that disappointed, and it's explained by CapEx investment in MACsec production equipment — not a demand problem. The ATM equity offering ($334M cash raise) at elevated prices was opportunistic and well-timed.

The Q3 pre-announcement ($404-408M vs $335-345M) is the session's most important data point. A $66M beat is not noise at this scale — it is the largest beat in company history and reflects real demand acceleration, not base effects.

Leading Indicators

CRDO does not report traditional SaaS-style leading indicators (no ARR, no RPO of scale). RPO was reported at $33.9M in Q2 — negligible relative to $268M quarterly revenue, indicating a short-cycle, purchase-order-driven model with near-zero backlog visibility.

Proxy leading indicators:

Divergence analysis: No leading indicator is flashing early weakness. The inventory build (+28.7% QoQ) preceding a +51% QoQ revenue quarter in Q3 is consistent with demand-driven production staging. The short-cycle business model means the best predictor is customer purchase order cadence — and management states 12-month visibility with "strengthening" forecasts.

Bearish divergence to watch: AR at 91% of quarterly revenue suggests some hyperscaler payment timing concentration. If Q3 revenue is $406M but AR balloons further, that warrants investigation. Not a current flag, but something to watch on March 2.

Scuttlebutt Findings

(Pre-computed by scuttlebutt stage — supplemented with my assessment.)

• AEC market share: 73-88%. Multiple independent sources confirm Credo as de facto AEC standard. Explicitly named in xAI's Colossus cluster ("Credo AECs"). Hyperscalers deepening, not diversifying away. (Source: Chipstrat, CNBC, everyticker.com)
• CEO credibility: High. Brennan built Credo from 15 employees to 400+ over 12 years, delivering >90% CAGR revenue since 2014. Active CNBC presence. GSA and Axiado board seats signal industry standing. No controversies. (Source: GSA, Glassdoor, Credo IR)
• Employee sentiment: Solid. Glassdoor 4.4/5, Blind 4.3/5. "Chaotic / start-up energy" descriptor on Blind — expected at a company growing 270% YoY. No dysfunction or exodus signals. (Source: Glassdoor, Blind)
• Hiring accelerating: Job postings tripled Sep→Feb (9→26 open roles). No layoffs found anywhere in 2025 tracker databases. Revenue-per-employee at 400+ employees and $800M+ ARR is exceptional. (Source: AltIndex, TechCrunch tracker)
• Competitor threats real but not imminent: Broadcom dominates optical DSP; Marvell shipping 1.6T DSPs; Astera Labs competing in PCIe/CXL. None have matched Credo's AEC position. (Source: Nasdaq.com, ainvest.com)
• IP licensing to Siemon + ITC filing: Positive signal — Credo asserting its moat rather than ceding ground. "3 conversations settled, couple more in flight." This enriches IP licensing as a revenue stream while not threatening core market position. (Source: Credo IR, Ackerman Q&A)
• ALC / Hyperlume combination: Brennan confirms "more of an execution play" — technology validation largely done, microLED manufacturing now the challenge. If it delivers at FY28 revenue ramp, the TAM doubles. (Source: Q2 FY26 call — O'Loughlin Q&A)
• Seeking Alpha bear case — valuation: The main bear argument is not business quality but price (65x forward EPS, optical headwind risk). "I Called Credo's Massive Rally. Here's Why I'm Downgrading It" — valuation-based, not thesis-based. (Source: SA)

Nothing in scuttlebutt contradicts the financial picture. The risk to watch is optical transceiver cost curves eroding AEC economics — not a current event but a 2027+ question.

Valuation Context

EV/FY26 at 17x is the right way to frame this. A company growing >200% YoY with 67%+ gross margins and 46%+ NG operating margins trading at 17x forward revenue is not obviously overvalued in the context of AI infrastructure hyperspend. The more relevant comparison is EV/FY27 revenue: if Q4 FY26 implies ~425MandCredoguidesmid − single − digitsequentialgrowthintoFY27, FY27revenuecouldreach 2.0-2.2B (5-10% sequential from $425M, compounded). At $23.3B market cap, EV/FY27 Rev ≈ 10-11x — that is genuinely cheap if FY27 growth is even 50%+.

SBC dilution is the valuation headwind. $45.3M SBC in Q2 = 16.9% of revenue. GAAP EPS of $0.44 vs NG EPS of $0.67 — a 52% spread. Share count grew from 150M (Q2 FY24) to 188M (Q2 FY26), +25% dilution over 9 quarters (~11% annualized). This is the right number to watch — not headline-alarming but real economic cost to shareholders.

Platform & Secular Position

Secular trend: AI training and inference infrastructure. Hyperscalers (Amazon, Microsoft, xAI, Meta, Oracle) are building GPU clusters measured in hundreds of thousands to millions of GPUs. Every GPU requires connectivity. Credo provides the connectivity layer — AECs, retimers, optical DSPs — plus emerging new categories.

Platform vs point solution: Credo started as a point solution (AECs) and is deliberately evolving into a platform. The five-pillar strategy (AEC, IC, ZF Optics, ALCs, OmniConnect) with the PILOT software/telemetry layer is a legitimate platform foundation — especially as PILOT creates switching costs by embedding diagnostics into cluster management software.

TAM penetration: AEC TAM was estimated at ~$3B; Credo claimed $2B+ in TAM 18 months ago. Now TAM re-estimated at $10B+ with new pillars. At $1.3B FY26 revenue, Credo has captured roughly 13-20% of the current addressable market. The question is whether the TAM grows as fast as Credo penetrates it (the AI capex cycle expanding the denominator faster than the company closes it).

One important structural insight from the call: Brennan describes AECs as "displacing optical rack-to-rack connections up to 7 meters." The TAM boundary is expanding as cluster densification pushes more connections into the copper-viable range. This is a market creation dynamic, not just market share capture.

Key Risks

1. GM compression in Q3. Management guided Q3 non-GAAP GM at 64-66%, a 170-370bp step-down from Q2's 67.7%, citing product mix ramp costs. Revenue beat was massive (+19%), but if margins compressed to 64% on 406M, non − GAAPoperatingincomewouldbe 183M (45% margin) — still excellent but confirming a GM headwind as optical DSP scales. Full results March 2.

2. Customer concentration. Top 4 customers = ~93% of Q2 revenue (42/24/16/11). If a single hyperscaler's build cycle pauses — as we saw with the top-1 customer dropping from 61% (Q4 FY25) to 35% (Q1 FY26) — quarterly revenue can swing sharply. The business has 12-month visibility, not multi-year contracts.

3. Optical cost curves. AEC economics depend on optical modules remaining power-hungry and unreliable relative to copper. If next-generation optical transceivers improve reliability and/or CPO achieves hyperscaler adoption, the AEC TAM contracts. Brennan dismisses CPO as "not anytime soon" — credible near-term, uncertain beyond 2027.

4. Platform execution (three new pillars). ZF Optics (revenue FY27), ALCs (revenue FY28), OmniConnect (revenue FY28) are all 12-36 months from meaningful revenue. If any of these miss their timelines, the 10BTAMstoryshrinksbackto 3-4B. The FY27-FY28 bridge is unproven.

5. SBC / dilution creep. $45.3M SBC in Q2 = 16.9% of revenue and growing in absolute terms. Share count up 25% in 9 quarters. If revenue growth decelerates while SBC stays elevated, the GAAP-to-NG spread will become harder to ignore.

Key Catalysts

1. March 2, 2026 full Q3 results. The key question: did gross margins hold at 67%+ on $406M revenue, or did they compress as guided? If margins hold, the full-year profitability story gets dramatically stronger and the bear case on optical mix evaporates.

2. Q4 FY26 guide. "Mid-single-digit sequential" from 406Mimplies 425M — but management has sandbagged every quarter for five consecutive beats. Q4 guide will set the tone for FY27 expectations.

3. 5th hyperscaler scaling. Customer #5 is contributing "initial revenue." If it crosses 10% by Q4 FY26 or Q1 FY27, customer concentration risk improves materially and TAM expansion is validated.

4. PCIe design wins in FY26. Management "on track." Production revenue in FY27. This unlocks the scale-up network TAM and creates a second major product line within the IC pillar.

5. ZF Optics second hyperscaler sampling (H2 FY26). Brennan committed to sampling a second US hyperscaler in the current fiscal year. If this becomes public, it validates the optics expansion thesis before FY27 revenue arrives.

Position Disclosure

Atlas does not hold positions. Conviction 4/5. The business is executing at an elite level — five consecutive beats, margin expansion, platform expansion, and now the largest guidance beat in company history. The open question (Q3 margins) gets resolved in 7 days. At 17x FY26 revenue growing 200%+, the valuation is defensible if the business sustains even 40-50% YoY growth into FY27. The risk is not business quality — it is cycle dependence on hyperscaler AI capex and the optical headwind in the medium term.