Credo 2025-26: Revenue $437M (+126%), FY26 Revenue >$800M (+85%)

FY25 (Apr year-end) revenue $437M (+126% YoY); op income $37M (vs -$37M FY24); NI $52M (vs -$28M FY24); EPS $0.29 (vs -$0.18). FCF $29M (+70%). Q4 revenue $170M (+180% YoY, +26% QoQ). Three hyperscalers each contributing >10% of Q4 revenue (vs 86% concentration in single customer Q3) — meaningful customer diversification milestone. AEC product line: ZeroFlap AECs replacing optics at largest customer + multi-hyperscaler ramp. Optical DSP business: 50-gig + 100-gig per lane solutions; new DSPs at OFC; 3nm tape-out for leading-edge. Retimer business: 50/100 gig per lane Ethernet + PCIe retimer entry + AI server ODM platform commitment. PILOT software platform. Q1 FY26 guide: revenue $185M-$195M (+12% sequential midpoint); non-GAAP gross margin 64-66%; non-GAAP opex $54-56M. FY26 guide: revenue >$800M (+85% YoY); non-GAAP net margin approaching 40%.

Key takeaways

• Revenue $437M (+126% YoY); Q4 $170M (+180% YoY) — AI infrastructure connectivity hypergrowth. Credo's FY25 (April year-end) saw revenue more than double to $437M, with Q4 hitting $170M (+180% YoY, +26% sequential). This is one of the cleanest hypergrowth stories in the AI infrastructure value chain. The growth is driven by AEC (Active Electrical Cable) adoption at hyperscalers + optical DSP scaling + retimer momentum + entry into PCIe retimers — multiple converging product lines all benefiting from the hyperscaler AI capex cycle.

• FY26 revenue >$800M (+85% YoY); non-GAAP net margin approaching 40% — sustained hypergrowth + margin expansion. Management explicitly guided FY26 revenue to exceed $800M (vs $437M FY25) — implying +85% YoY growth. Combined with non-GAAP net margin guide of approaching 40%, FY26 non-GAAP net income approaches $320M+ implied. This is a remarkable inflection: a company that lost money on $193M revenue in FY24 now guiding to $800M+ revenue with ~40% net margins in FY26. Operating leverage flowing through at scale.

• Customer diversification milestone: 3 hyperscalers each >10% of Q4 revenue (vs 86% single-customer concentration Q3). Q3 FY25 had Credo's largest customer at 86% of revenue — a major concentration risk. Q4 FY25 marked a meaningful diversification: three hyperscalers each contributing >10% of revenue. Management projected 3-4 customers contributing >10% of revenue in coming quarters. This is the cleanest evidence that Credo's product positioning + design wins are scaling across multiple hyperscalers (likely Microsoft, Amazon AWS, Google, Meta, Oracle Cloud) — multi-customer adoption confirms the AEC + optical DSP + retimer technology leadership.

• AEC product line: ZeroFlap AECs replacing optics — structural cost + reliability advantage. Credo's flagship AEC (Active Electrical Cable) products, particularly ZeroFlap variants, are increasingly replacing traditional optical solutions in short-reach hyperscaler interconnect. Advantages: lower power consumption, lower cost per port, higher reliability (link flap reduction = key data center operational metric). The AEC market opportunity is multi-billion dollar TAM as hyperscalers scale GPU clusters from 32K → 100K+ → 1M+ accelerators per cluster. Each cluster generation requires more interconnect bandwidth + reliability.

• Multi-product roadmap: 3nm tape-out + PCIe retimer + 50/100-gig per lane scaling — multi-year compounding. FY25 included 3nm tape-out for leading-edge optical DSPs, PCIe retimer market entry (with AI server ODM platform commitment), 50/100-gig per lane Ethernet retimer scaling, and PILOT software platform expansion. Combined, this represents a multi-product technology roadmap that supports the compounding revenue trajectory beyond FY26. PCIe retimers alone open meaningful new TAM in AI server interconnect (CPU-GPU-memory).

Business

Credo Technology Group Holding Ltd is a US-listed (Cayman-domiciled) semiconductor company providing high-speed connectivity solutions for hyperscaler data centers, with multi-product portfolio:

• AEC (Active Electrical Cable) (~50%+ of revenue, fastest growing): ZeroFlap AECs + traditional AECs replacing optics in short-reach (rack-to-rack, intra-rack) hyperscaler interconnect. Multi-hyperscaler customer base scaling.
• Optical DSP (~25-30%): 50-gig + 100-gig per lane optical DSP chips. LRO (Linear Receive Optics) concept adoption. 3nm tape-out for next-gen.
• Retimer (~20-25%): Line card retimers (400-gig, 800-gig Ethernet); PCIe retimers (Gen 6, 64-gig PAM4); 50/100-gig per lane.
• SerDes Licensing + Chiplets (~5%): IP licensing + chiplet building blocks.
• PILOT Software Platform: Configuration, monitoring, telemetry software for connectivity products.

Strategic moves FY25:

• Revenue +126% YoY to $437M
• Q4 revenue +180% YoY to $170M
• 3 hyperscalers each >10% of Q4 revenue (customer diversification)
• AEC ZeroFlap replacing optics at largest customer + multi-hyperscaler ramp
• Optical 50G + 100G per lane + 3nm tape-out
• Retimer 50/100G per lane + PCIe retimer entry
• AI server ODM platform commitment for PCIe retimer
• PILOT software platform expansion
• New DSPs at OFC (Optical Fiber Conference)
• Q4 hyperscaler AI cluster scaling

FY25 financial performance

The earnings progression: FY22-24 was the multi-year investment phase with revenue growing modestly from $106M to $193M while sustaining operating losses. FY25 marked the structural inflection: revenue +126% to $437M, op income flipping positive ($37M), net income $52M (vs -$28M FY24), EPS $0.29 (vs -$0.18). The inflection reflects (a) AEC adoption ramp at hyperscalers, (b) operating leverage on the cost structure that was built ahead of revenue, (c) multi-product breadth.

FCF $29M FY25 (+70%); capex $36M (+125%) reflecting tooling + tape-out investment. Total debt minimal ($16M). Cash position strong from prior fundraising.

Capital allocation

• Capex: $-36M FY25 (+125% YoY) — tape-out + tooling investment.
• Dividends: $0 (no dividend; growth phase).
• Buybacks: $0 (no buyback program; growth phase).
• Total debt: $16M (minimal).
• FCF: $29M FY25.
• Operating expenses growing at less than half the rate of revenue — operating leverage thesis.

FY26 outlook (per Q4 FY25 call, 2025-06-02)

Management noted continued surging demand for innovative connectivity solutions, customer-driven innovation, multi-hyperscaler diversification, AEC ZeroFlap replacing optics, and PCIe retimer market entry.

Key risks

Hyperscaler AI capex cycle. Credo's revenue is heavily dependent on hyperscaler AI infrastructure spending. Any deceleration in hyperscaler capex (Microsoft, AWS, Google, Meta, Oracle, Coreweave, Tesla, etc.) creates immediate revenue impact. Multi-quarter visibility limited.

Customer concentration (Q3 FY25 86% single customer). While Q4 FY25 showed meaningful diversification (3 hyperscalers >10% each), historical concentration is a clear risk. Single-customer order changes can swing quarterly revenue dramatically.

Competitive landscape — Marvell, Broadcom, Cisco, Astera Labs, others. Credo competes with Marvell (DSPs), Broadcom (SerDes), Cisco (Acacia DSPs), Astera Labs (smart cable connectivity, IPO in 2024), Innolight, Eoptolink, others. Each major hyperscaler typically maintains 2-3 vendors per product category. Pricing pressure + design-win competition ongoing.

AEC vs optical migration pace. AEC adoption depends on hyperscaler architecture decisions (DAC vs AEC vs optical for various reaches). Any pause in AEC adoption (e.g., shifting to longer-reach optical) impacts Credo's primary revenue driver.

Technology transition execution (3nm tape-out + PCIe Gen 6). Multi-year semiconductor product development requires flawless execution at 3nm + 2nm nodes. Tape-out delays, yield issues, or first-pass silicon problems create multi-quarter revenue slips.

Cayman / China relationship. Credo is incorporated in Cayman Islands; founders / R&D have China-related history. Continued export control + CFIUS / Investment Screening dynamics in semiconductors could affect customer relationships.

Foundry capacity (TSMC). Credo depends on TSMC for advanced node manufacturing. TSMC capacity allocation + pricing affects supply availability.

Optical DSP competition. Marvell + Broadcom + Inphi-Marvell legacy + Cisco-Acacia all compete in optical DSP. 50/100-gig per lane scaling competitive intensity.

PCIe retimer entry execution. PCIe retimer market dominated by Astera Labs + Texas Instruments + Microchip. Credo's entry needs to ramp design wins to translate to revenue.

LRO (Linear Receive Optics) adoption pace. LRO concept adoption across hyperscalers depends on architecture timing.

Inventory + supply chain dynamics. Q3 FY25's 86% single-customer concentration was driven by one hyperscaler ramp. Working capital + inventory swings sensitive to single-customer order dynamics.

Regulatory / export controls. US export controls on semiconductors + Cayman/foreign listings create regulatory monitoring requirements.

FX (USD revenue). Most revenue USD-denominated; cost structure mixed. Limited direct FX exposure.

Talent retention. Semiconductor design talent + AI infrastructure expertise highly competitive.

Bottom line

Credo Technology FY25 (April year-end) is the structural inflection year: revenue +126% to $437M; op income $37M (vs -$37M FY24); NI $52M (vs -$28M); EPS $0.29 (vs -$0.18). Q4 revenue $170M (+180% YoY, +26% QoQ). 3 hyperscalers each >10% of Q4 revenue (customer diversification milestone vs 86% single-customer Q3). AEC ZeroFlap replacing optics at largest customer + multi-hyperscaler ramp. Optical 50G + 100G per lane + 3nm tape-out. Retimer 50/100G per lane + PCIe retimer entry + AI server ODM platform commitment. PILOT software platform.

FY26 guide: revenue >$800M (+85% YoY); non-GAAP net margin approaching 40%. Q1 FY26 revenue $185M-$195M (+12% sequential midpoint); non-GAAP gross margin 64-66%; non-GAAP opex $54M-$56M.

The risks are real — hyperscaler AI capex cycle dependence, customer concentration (recent Q3 86% single-customer), competitive landscape (Marvell, Broadcom, Cisco, Astera Labs, Innolight), AEC vs optical migration pace, technology transition execution (3nm tape-out + PCIe Gen 6), Cayman / China relationship + export controls, foundry capacity (TSMC), optical DSP competition, PCIe retimer entry execution, LRO adoption pace, inventory + supply chain dynamics, regulatory / export controls, FX, talent retention.

But the structural thesis (US-listed semiconductor + high-speed connectivity for hyperscaler AI infrastructure + AEC ZeroFlap replacing optics + multi-hyperscaler diversification milestone + optical DSP 50/100G per lane + 3nm tape-out next-gen + retimer 50/100G + PCIe retimer market entry + AI server ODM platform commitment + PILOT software + FY26 revenue >$800M (+85%) + non-GAAP net margin approaching 40% + multi-product roadmap + minimal debt + operating leverage thesis) is intact and FY25 confirms.

Quality AI infrastructure connectivity hypergrowth compounder mid-cycle, with multi-product breadth + multi-hyperscaler customer diversification + AEC market leadership + optical DSP roadmap + PCIe retimer optionality + structural cost advantages over optics + multi-year design-win pipeline. The FY25 +126% revenue + Q4 +180% YoY + customer diversification milestone + 3 hyperscalers each >10% + AEC ZeroFlap adoption + 3nm tape-out + PCIe retimer entry + FY26 >$800M revenue guide + ~40% net margin guide creates one of the cleaner AI infrastructure compounding setups for investors seeking direct exposure to hyperscaler AI capex + connectivity TAM + design-win-driven semiconductor model. The FY26 framework + multi-product roadmap + operating leverage trajectory + customer diversification + technology leadership provides multiple paths to outperformance over a multi-year horizon. Hyperscaler capex cycle + customer concentration + competitive landscape + technology transition risks remain ongoing, but the multi-product breadth + customer diversification + structural cost advantages + multi-year design-win pipeline + operating leverage support continued compounding through cycles.

Citations

• Credo Technology Group Holding Ltd FY25 (Apr year-end) Form 10-K (filed June 2025, SEC EDGAR).
• CRDO Q4 FY25 earnings call, 2025-06-02 — Q4 revenue $170M (+26% sequential, +180% YoY); FY revenue $437M (+126% YoY); 3 hyperscalers each >10% of Q4 revenue; AEC growth + customer diversification with hyperscalers; ZeroFlap AECs replacing optics; advantages of AECs over traditional solutions; optical 50/100G per lane DSPs + new at OFC + 3nm tape-out; Retimer 50/100G per lane Ethernet + PCIe retimers + AI server ODM platform commitment; PILOT software platform; Q1 FY26 revenue $185M-$195M (+12% sequential midpoint); non-GAAP gross margin 64-66%; non-GAAP opex $54M-$56M; FY26 revenue >$800M (+85% YoY); non-GAAP net margin approaching 40%.
• CRDO Q3 / Q2 / Q1 FY25 earnings calls — supporting hyperscaler ramp + product portfolio expansion + operating leverage progression.
• Internal financial_statements view (consolidated annual + cash flow + capital structure).