Deep Research

Credo Technology (CRDO)

4 min read

Snapshot

  • Ticker: CRDO (NASDAQ)
  • Bucket: AI Connectivity & Optics
  • Q1 2026 position: 1,091,233 shares, $102,434,042 — 2.05% of 13F, #15 holding; new position in Q1 2026
  • HQ: San Jose, California (incorporated in the Cayman Islands)
  • One line: High-speed SerDes specialist whose active electrical cables (AECs) have become the default in-rack connectivity for hyperscale AI clusters, now expanding into optical DSPs and PCIe retimers.

Business Overview

Credo’s core asset is power-efficient SerDes IP, productized as: HiWire AECs (copper cables with DSP re-timing at each end — the dominant revenue line), optical DSPs for transceivers, line-card PHYs/chiplets, and newer PCIe retimer products. The pitch against optics inside the rack is brutal in its simplicity: management claims its “Zero Flap” AECs deliver up to 1,000× better reliability than laser-based optical modules at materially lower power and cost — and in Baker’s power-constrained world, every watt not spent on in-rack optics is a watt available for compute.

Fiscal 2026 (ended May 2026) was explosive: revenue reached **437M), with Q4 FY2026 alone at a record **662M. Growth was overwhelmingly AEC-driven — unit shipments to hyperscalers contributed over 95% of the product revenue increase — across both 100G/lane and emerging 200G/lane deployments, at hyperscalers and neocloud operators alike. Guidance for fiscal 2027 calls for >80% revenue growth, including >$600M from the optical (DSP) portfolio — a deliberate second act beyond copper.

The structural caveat: the top three customers are reported at roughly 88% of revenue — among the most concentrated books in semis.

Financial Trajectory

PeriodRevenueYoYNon-GAAP gross marginNotes
FY2024 (Apr)~$193M~+5%~63%pre-AI-ramp base
FY2025 (Apr)~$437M~+126%~65%AEC inflection
FY2026 (May)~$1.3B~+200%68.1%non-GAAP NI $662M; op margin 47.8%
FY2027 (guide)>$2.3B implied>+80%—>$600M optical

Why Atreides Owns It

Credo is the newest expression of the connectivity thesis — initiated as a full ~600M+ guided in FY2027) means it participates either way. The timing of entry — the quarter before Credo printed +157% growth and guided +80% — alongside simultaneous Q1 2026 entries elsewhere in the book suggests Atreides rotated profits from trimmed winners (Astera in 2025, Lumentum, Ciena) into the bucket’s fastest grower. Baker has not, to our knowledge, discussed Credo by name in public; the thesis fit is inferred from the bucket construction (unconfirmed).

Position History

QuarterTypeShares/NotionalValue% of 13F
Q4 2024—not held——
Q1 2025—not held——
Q2 2025—not held——
Q3 2025—not held——
Q4 2025—not held——
Q1 2026Common1,091,233$102,434,0422.05%

A clean, single-quarter initiation at full size (~$94/share implied) — no scaling-in visible. Notable that Atreides skipped the name through its entire 2025 run-up and only entered after the 200G/lane and optical-DSP second acts became visible; whether this becomes a core holding or another whipsaw trade (cf. Astera) is the open question for Q2 2026.

Risks

  • Extreme customer concentration: top three customers ~88% of revenue; one hyperscaler redesigning its rack topology or dual-sourcing AECs would crater growth.
  • Copper’s physical ceiling: at 400G/lane and beyond, or with longer scale-up reaches, optics (or CPO) may reclaim the in-rack domain AECs currently own.
  • Competition: Marvell, Broadcom, Astera (Taurus), Spectra7, and point-solution entrants all target AEC/DSP sockets; 68% gross margins are an invitation.
  • Valuation: the stock more than tripled over 2025–26; at the fund’s entry it priced much of the FY2027 +80% guide already.
  • Optical DSP execution: the $600M FY2027 optical target is a new market against entrenched incumbents (Marvell, Broadcom) — a miss would break the diversification narrative.

Sources

Backlinks

Graph View