CRDO

Current Thesis

Custom-silicon AI-interconnect royalty is re-accelerating into the 1.6T/800G hyperscaler cycle, and the 2026-04-13 $750M DustPhotonics acquisition just gave the bull case a new leg: CRDO is no longer just a copper-AEC vendor — it's now bidding for the optical/CPO layer where the next 3 years of hyperscaler capex actually sits. Three sell-side moves inside 7 days (Jefferies $175 init, Needham $220 reiteration, Goldman raise to $170) tell us the street is being forced to reprice the TAM, and retail hasn't fully digested it yet. We're buying the narrative-acceleration window before the ~2026-06-03 FY Q4 print either confirms the new run-rate or breaks it. Classic Picks & Shovels (a2) setup — ride the tape, trim into mania, cut on any hyperscaler customer-concentration scare.

Bull Case

• DustPhotonics M&A (2026-04-13): $750M cash + ~0.92M common stock for a silicon-photonics pureplay. This is a narrative multiplier — CRDO was the "copper guy"; now it's in the CPO/optical engine conversation where NVDA, AVGO, MRVL play. Deal closes mid-FY2027, but narrative rerating happens NOW.
• Analyst cluster velocity: 3 positive actions in 7 days (2026-04-13 Jefferies Buy init $175, 2026-04-15 Needham Buy reiterated $220, 2026-04-16 GS raised PT to $170). Street PT dispersion $170–$220 on a name that was sub-$100 weeks ago = consensus catching up, not ahead.
• Hyperscaler interconnect tailwind: AWS, MSFT, META, GOOGL all in the middle of 800G → 1.6T ethernet transitions. Every GPU rack needs 10–20x more interconnect bandwidth. CRDO's AEC (Active Electrical Cable) is the cheapest solution inside-rack; DustPhotonics gives them the optics for rack-to-rack.
• Earnings revision cycle: Last 2 prints (Dec-2025, Mar-2026) beat consensus and raised guide. Pattern of "beat+raise" is still intact — next print ~2026-06-03 likely sees another guide-up on 1.6T ramp.
• Momentum confirmation: Named in "top 10 large-cap gainers week of April 13-17" (2026-04-19) — ripping tape with follow-through, not a one-day pop.
• Narrative state: ACCELERATING. We are ahead of CNBC mainstream coverage and ahead of full retail chase (no WSB front-page yet).

Bear Case

• Customer concentration: Historically ~40%+ revenue from AWS. Any whisper of Amazon insourcing connectivity (they already have Nitro/Graviton) is a -15% overnight.
• Multiple already full: At ~$130-ish handle and fwd P/E well above sector, the margin for operational miss is thin. Sell-side PTs at $170-$220 imply only 30-70% upside — not 5x asymmetry.
• M&A execution risk: $750M + stock for DustPhotonics is a big bite. Integration risk, dilution overhang from 0.92M shares, plus photonics is a different engineering culture. Street will punish any closing delay or synergy miss.
• Photonics competitive moat: Optical/CPO is where AVGO, MRVL, COHR, LITE all want to play. Being a late entrant via M&A is not the same as organic silicon IP. Could end up paying $750M for a "me-too" in a commoditizing layer.
• Retail froth signal: Multiple "stocks moving premarket" / "why these 5 stocks are on investors' radars" mentions inside a week = we are not early anymore, we are mid-acceleration. If tape turns, exits get crowded fast.
• Macro/semi cycle risk: If hyperscaler capex guide gets cut at any MAG7 print, every AI-interconnect name gets re-rated -20%+ in a session.

Setup & Price Structure

• Price context: No live price feed in this dossier — operator must pull before acting. Qualitatively: stock gapped up on 2026-04-13 DustPhotonics news and held the gap through the week, which is tape-confirmation.
• Structure read: Post-gap consolidation inside the analyst cluster = bullish continuation pattern. We want to buy pullbacks to rising 20-EMA / 21-EMA, not chase vertical prints.
• Volume: Deal-day volume elevated; follow-through days into 2026-04-17 kept it in the top-10 large-cap gainers list — institutional accumulation, not just retail.
• Invalidation level: Weekly close below 20-EMA = trend broken, exit. Single-day -10% on customer-specific news = immediate stop.
• Entry playbook: HIGH conviction on pullback entries. Do NOT chase +5% green candles; wait for sideways/red day into moving-average support. If the name rips another +10% without consolidation, wait or probe small.
• Size: Up to 4% of book on clean pullback entry given HIGH conviction + archetype-2 (not a retail-squeeze cap).

Catalyst Calendar (next 30 days)

• ~2026-05-15 to 2026-05-30 — DustPhotonics deal proxy/8-K filings, any HSR review updates. Any regulatory friction = negative.
• ~2026-05-28 (est.) — Pre-announcement window for FY Q4 (May quarter). CRDO historically does NOT pre-announce, but guide-raise whisper season starts here.
• ~2026-06-03 to 2026-06-05 (est.) — FY Q4 2026 earnings print. Binary event. This is why we do NOT hold size through — trim 50%+ minimum 3 trading days before regardless of tape.
• Ongoing: Any hyperscaler capex commentary at MSFT (2026-04-24 est.), GOOGL (2026-04-24 est.), META (2026-04-29 est.), AMZN (~2026-05-01 est.) prints — read-through is direct and tradeable intraday.
• Industry: OFC 2026 follow-on commentary from competitors (MRVL, AVGO, COHR) on optical roadmap. Any competitive product announcement from AVGO on CPO is negative read-through.

What Would Change Our Mind

• Bull → Max conviction (SUPREME): CRDO pre-announces or whispers a guide-raise before 2026-06-03, OR a 4th analyst move to $200+ PT from a tier-1 (MS/JPM). Then size up on next MA-retest.
• Bull → Neutral (DEFER): 2-week sideways chop between current level and gap-fill, no new news. Not broken, just paused — re-enter on breakout.
• Bull → Dead (SKIP / exit if long):
• Weekly close below 20-EMA = structural trend break, exit immediately no excuses.
• Amazon/AWS announces in-house interconnect silicon or a non-CRDO partnership = thesis-ender.
• DustPhotonics deal unwinds, gets regulatorially blocked, or closing gets delayed >6 months.
• FY Q4 print misses OR guide-down — cut regardless of after-hours bounce.
• Sector-wide: if AI-capex narrative breaks (any MAG7 capex guide-cut), trim across the book not just CRDO.
• Archetype flip: If CRDO becomes a WSB/StockTwits front-page retail meme with 3d velocity >200%, re-classify as archetype-6 and tighten position cap to 1%/name with RSI>75 trim trigger.

Correlation Notes

• Tightest correlation: AVGO (custom silicon / ASIC peer), MRVL (DSP + optics direct comp), ALAB (Astera Labs — smart cable/AEC adjacent), COHR, LITE (optical component). When MRVL prints, CRDO moves intraday.
• Hyperscaler read-through: MSFT, AMZN, META, GOOGL capex prints drive CRDO sentiment 1:1. A single hyperscaler capex guide-cut is a -8% to -15% event.
• Theme basket: Sits in ai-chip-infra-memory + ai-interconnect + silicon-photonics + hyperscaler-capex. If we already hold AVGO or MRVL at size, CRDO adds correlated exposure — cap combined basket at ~12% of book, not additive 4%+4%+4%.
• Inverse tell: If TSM utilization commentary softens or optical component peers (COHR/LITE) guide down on datacenter mix, CRDO is a leading-indicator short from the same desk — watch for cross-confirmation.
• Macro: High-beta, long-duration semi. Rate-cut tape helps, yield spikes hurt. 10Y >4.75% = meaningful headwind for the whole basket.