CRDO · US
CRDO.US: FY26 update — AI rack connectivity AEC / SerDes surge
Credo FY2026 revenue $1.335B (+206%), Q4 $437M (+157% YoY), Non-GAAP EPS $1.16, ~68% GM, ~$1.4B cash. Q1 FY27 guide $465–475M. AEC / SerDes is core to AI rack high-speed connectivity, but hyperscaler concentration (Amazon, etc.) is extreme. See AVGO, ANET, NVDA, CLS.
Cite a section with a deep link, e.g. /en/r/crdo-us-research#thesis
FY2026 snapshot
- FY26 revenue (+206%)
- USD 1.3B
- Q4 revenue (+157%)
- USD 437.0M
- Q4 Non-GAAP EPS
- USD 1.16
- Gross margin
- 1
Q1 FY27 guidance
- Q1 revenue low
- USD 465.0M
- Q1 revenue high
- USD 475.0M
- Cash + STI
- USD 1.4B
- FY26 Non-GAAP NI
- USD 662.0M
Thesis
Credo Technology sits at the “last meter” of AI data-center rack connectivity — AEC (Active Electrical Cable) and SerDes IP are indispensable for GPU cluster scale-out. FY2026 validated the explosion with $1.335B revenue (+206%), Q4 $437M (+157% YoY), and 68% GM. Q1 FY27 guide of $465–475M suggests the demand curve has not peaked.
The premium holds only if:
- AI rack deployment density keeps rising — every new GPU rack adds AEC / SerDes demand, tightly linked to NVIDIA GPU shipment slope;
- 112G / 224G technology leadership — high-speed SerDes is the competitive moat; Credo leads in AEC;
- Fabless high-margin model — 68% GM proves pricing power; ~$1.4B cash provides strategic flexibility.
But CRDO is a classic small-cap + extreme customer concentration story — Amazon and a few hyperscalers dominate orders. Any single-customer capex reset can move the quarter. Contrast with Broadcom’s broader mega-customer base.
Business
Model: AEC cables, SerDes IP, and retimers for AI rack high-speed interconnect.
Flagships: AEC (~60%+ sales); 112G/224G SerDes IP; custom retimers.
Competitiveness: Early AEC design wins; cost/deploy vs optics for mid-reach.
Strategy: 224G ramps and customer diversification.
Value chain
Mid-stack interconnect — the “last meter” inside AI racks.
Upstream: foundry/cable/OSAT. Downstream: hyperscalers (e.g. AWS) and ODMs. Risks: customer CapEx concentration; 224G peer competition.
Valuation
FY26 +206% growth + 68% GM drove a massive 2026 outperformance. As of this note (2026-07-30), valuation already prices multi-year hyper-growth. 206% is not sustainable — prepare for multiple compression as growth normalizes.
Valuation frame
| Metric | Status | Read |
|---|---|---|
| 盈利增长 | 爆发式增长 | FY26 +206%；Q4 +157% YoY |
| 毛利率 | 极高 | ~68%；fabless 高 margin 模型 |
| 估值 | 偏高 | AI 互联主题已大幅定价 |
| 客户集中度 | 极高风险 | Amazon 等少数 hyperscaler 主导 |
| 市值体量 | 中小盘 | 波动率显著高于 AVGO / NVDA |
| AI CapEx 周期 | 敏感 | 机柜级部署节奏决定营收斜率 |
Versus AVGO: CRDO does not compete at ASIC / networking silicon; it captures rack connectivity increment. Smaller cap, higher volatility, more extreme concentration. Versus Arista: ANET does system-level switching; CRDO does underlying connectivity components — complementary in the chain.
Financial trend (~24 months)
Eight-quarter revenue and gross margin with YoY and QoQ. Semi/AI hardware seasonality is secondary to supply, ASP, and order timing.
Revenue · last 8 quarters
Interactive chart available in the reader.
Gross margin · last 8 quarters
Interactive chart available in the reader.
Operations
FY2026 operations
FY2026 P&L and segments
| Item | FY2026 | Note |
|---|---|---|
| FY26 营收 | $1.335B（+206%） | FY 截至 2026-05-02 |
| Q4 营收 | $437M（+157% YoY） | +7.4% QoQ |
| Non-GAAP 净利润 | ~$662M | FY26 全年 |
| Q4 Non-GAAP EPS | $1.16 | — |
| Q4 GAAP EPS | $0.88 | — |
| 毛利率 | ~68% | fabless 高 margin |
| 现金 + 短投 | ~$1.4B | 资产负债表健康 |
Product mix
| Product line | Revenue mix | Watchpoint |
|---|---|---|
| AEC 产品 | ~60%+ | AI rack 互联主力；增速最快 |
| SerDes / 光模块 | ~25% | 112G/224G 高速 IP |
| 其他互联 | ~15% | retimer / 定制方案 |
Takeaways:
- AEC ~60%+ of revenue — primary AI rack connectivity growth driver; SerDes IP adds technical depth.
- Non-GAAP NI ~$662M validates profitability — not a “growth without earnings” story.
- Cash ~$1.4B vs ~$1.3B annual revenue — exceptionally healthy balance sheet; ammo for R&D and potential M&A.
Products & AEC
Core product matrix:
- AEC (Active Electrical Cable) — cables with embedded retimer / SerDes, replacing traditional DAC and optics for 800G / 1.6T interconnect inside and between AI racks. Simple deployment, lower cost than optics, adequate reach — a preferred hyperscaler option at scale.
- SerDes IP (112G / 224G) — underlying high-speed connectivity technology; IP can be licensed to Broadcom and other networking silicon vendors.
- Retimer / custom solutions — bespoke connectivity for specific customer requirements.
AI rack deployment chain: NVDA GPU → CLS ODM rack → CRDO AEC connectivity → AVGO network switching. Credo is a non-optional link.
Customers & concentration
Customer structure is the biggest risk factor:
- Amazon is among the largest customers — AWS AI infra capex directly drives CRDO order slope;
- Other hyperscalers (Microsoft, Google, Meta) are in the design-win pipeline, but concentration remains extreme;
- Versus AVGO’s Google / Meta / OpenAI diversification, CRDO’s base is narrower.
Watch mega-customer capex commentary, design-win announcements, AEC shipment QoQ slope, and 224G new-customer adoption.
Competition
Strategy / product / risk map
Three axes:
| Axis | Credo | AVGO | Arista |
|---|---|---|---|
| Role | AEC / SerDes components | Network ASIC + XPU | System-level switching |
| Growth | +206% (FY26) | +48% (Q2) | ~+20% band |
| Concentration | Extreme | High | Moderate |
| GM | ~68% | ~69% EBITDA | ~60% |
Credo’s moat is AEC technology leadership + early mega-customer design wins, not ecosystem lock-in. 224G-era competition (Marvell, Broadcom internal SerDes, etc.) is the FY27–28 key variable.
Peer comparison
| Company | Share / role | GM | Strength | Weakness |
|---|---|---|---|---|
| Credo (self) | AEC / SerDes | ~68% | AEC design wins | Hyperscaler concentration |
| Broadcom | SerDes / switching | High | Scale + IP breadth | Less AEC pure-play |
| Marvell | Connectivity / optics | ~60% | Custom silicon | AEC share race |
| Arista | System switching | ~64% | System attach | Not cable/SerDes pure-play |
Management
CEO Bill Brennan bench stable; no material 24m C-suite change. Stability: steady — risk is customer concentration more than personnel.
Key management (24m)
| Role | Name | Since | 24m change |
|---|---|---|---|
| CEO | Bill Brennan | 2014 | No change |
| CFO | Dan O'Neil | 2022 | No material change |
| CTO / Founders bench | Founding tech team | — | No material C-suite turnover disclosed |
Outlook
Company guide · Q1 FY2027
| Item | Guide | Note |
|---|---|---|
| Q1 FY27 营收 | $465–475M | 延续高增路径 |
| QoQ 趋势 | +6–9% | 相对 Q4 $437M |
| 毛利率 | ~68% | 维持高 margin 结构 |
Framework:
- Q1 FY27 $465–475M vs Q4 $437M implies ~+6–9% QoQ — still high but slope is moderating (normal base effect).
- Sustaining 68% GM supports valuation; any GM compression warrants competitive-intensity concern.
- 224G SerDes ramp is the FY27 catalyst — determines whether Credo keeps leadership into the next speed bump.
Scenarios
| Scenario | Conditions | Implication |
|---|---|---|
| Bull | Q1 FY27 >$475M; 224G new design wins; Amazon capex accelerates | AI connectivity theme extends |
| Base | Q1 roughly hits $465–475M; QoQ +6–9%; GM ~68% | Hold; watch deceleration signals |
| Bear | Q1 miss; Amazon capex slows; 224G competition intensifies | 20–30% drawdown possible |
Positioning frame (research only): small-cap AI connectivity satellite; size discipline; avoid chasing at +206% peak growth.