Layer 4 · CRDO — Credo Technology
Thesis
Active electrical cable (AEC) pure-play — Credo's SerDes-and-cable platform replaces optical transceivers inside the rack at short reach with a power-efficient copper alternative, and the hyperscalers (Microsoft and Amazon named) have made AECs a standard part of the GPU-rack build. Q3 FY26 printed revenue +201% YoY — that's the kind of growth that earns a Tier A but Credo holds at Tier B for one reason: customer concentration is severe (MSFT + AMZN ~70%+ of revenue) and the June 1 Q4 FY26 print is the test of whether the AEC adoption broadens beyond the first two adopters.
What it does + financial print
Credo sits at Layer 4 — specifically the inside-the-rack interconnect layer that connects every GPU to every NIC to every switch within a single rack. Active electrical cables are the technology bet: copper cables with embedded SerDes / DSP chips that re-time and equalise the signal, achieving 100G/200G/400G per lane at meaningfully lower power than optical transceivers for sub-7-meter reaches. The economics: AECs cost less than optics, consume less power than optics, and are field-replaceable in a way that copper-passive cables (DAC) are not. The hyperscaler adoption pattern is the validation — Microsoft was the first, Amazon followed, and the consensus expectation is that Meta and Google qualify next.
Q3 FY26 (March 4): revenue $407.01M (+201.5% YoY), EPS $1.07 (+13.75% surprise on consensus). Q4 FY26 guide strong — AEC + optics roadmap intact. Median analyst PT $200 (high $260, low $170). Stock $183.32 on May 17 — has consolidated between the Q3 print and the June 1 Q4 print.
Bull case
The June 1 Q4 print confirms (a) MSFT and AMZN volume continuing to scale, AND (b) at least one of Meta / Google / Oracle entering qualification or initial deployment. Revenue scales from a $1.6B annualised run-rate (Q3 implied) to $3B+ in FY27 as the customer count broadens. The AEC TAM expands from the $4B-by-2028 consensus to a higher number as 200G-per-lane and 400G-per-lane SerDes unlock new use cases. Credo's optics roadmap (introduced as a hedge against AEC compression) starts contributing revenue, diversifying the platform away from pure-AEC.
Gap / bear case
Customer concentration is the gap. If either MSFT or AMZN pauses AEC orders for any reason (architecture change, internal cable program, alternative supplier qualification), Credo's revenue can drop 30-50% within a single quarter. Copper-to-optics substitution is a real long-term risk: as port speeds climb to 200G-per-lane and beyond, the reach over which AECs make economic sense shrinks, and the optics-instead-of-copper crossover point moves closer to the rack. Marvell and Broadcom both ship competing SerDes / DSP solutions at scale — Credo's technical lead is narrow.
Trigger to upgrade / downgrade
Upgrade to Tier A on the June 1 Q4 FY26 print confirming (a) revenue growth >+100% YoY again, AND (b) at least one new named hyperscaler customer in qualification. Downgrade on Q4 print showing customer-mix concentration deepening (single customer >50% of revenue) or a guide-cut from a captive hyperscaler insourcing program.
The trade
- Entry zone: $155-175 on pullback (or half-size into June 1 print)
- Stop: $140 close
- Position size: 1% NLV at zone; 1.5% on Tier A upgrade
- Catalyst date: Q4 FY26 print June 1 2026
- Conviction: 7/10