Phase Space AI

Trade Construction

Credo Technology Group [CRDO]

CRDO — Trade Construction

Analysis only. This memo issues no position verdict — the book decides whether any of this is taken. Spot $192.28 (2026-07-28 close). As of 2026-07-29.

---

1. What the trade would be expressing

Not "AI infrastructure is big." The specific, falsifiable proposition is:

Credo's optical business reaches the >$500m FY2027 target management stated on 2026-04-13, diluting the concentration risk that currently caps the multiple, while AEC holds at a mid-single-digit sequential rate — and the market re-rates a $2.1bn-revenue, 33%-operating-margin business off a multiple that has already compressed to the 14th percentile of its own trailing twelve months.

It is refuted by any of: a fourth quarter of sub-8% sequential growth with no optical contribution disclosed; days-of-inventory above ~180; or a change in the customer-concentration table showing one of D/B/E stepping down.

---

2. Equity liquidity — PASS, trivially

Any position this book would take is a fraction of a percent of one day's volume. Exit is not a constraint.

---

3. The options chain — pulled before anything was proposed

The Liquidity Criteria requires the actual chain first. It was pulled, and it disqualified the expiry that matched the horizon. Source: Alpaca paper-api/v2/options/contracts for discovery and data.alpaca.markets/v1beta1/options/snapshots for quotes and Greeks, 2026-07-29.

3.1 January 2027 (expiry 2027-01-15) — investable

Open interest in the hundreds-to-thousands and quoted size of 100–435 contracts. This chain can be filled.

3.2 June 2027 (expiry 2027-06-17) — NOT investable

Maximum open interest across every strike examined on this expiry is 78 contracts. This is the HCA failure mode exactly — an 18-contract chain that nothing tested for. The 12-month-matched expiry is uninvestable and no structure is proposed on it. Any option expression of this thesis must use January 2027 and accept a 5.6-month horizon against a 12-month target, which is a real mismatch and is stated as one rather than papered over.

3.3 Implied vs realised volatility — options are expensive here

Long-premium structures are paying a substantial spread over realised. That is the single most important input into vehicle choice below and it argues against buying naked calls.

---

4. Vehicle assessment

Vehicle conclusion: common stock. If defined risk is required, the Jan-2027 200/300 call spread is the only structure that clears the liquidity test — with its 5.6-month expiry stated as a known mismatch against the 12-month horizon.

---

5. Sizing inputs

Provided for the book; the memo does not set a size.

Inverse-volatility sizing will size this name down hard, and that is correct. A name with a 62% realised drawdown in each of two of the last four calendar years, three customers at 84% of revenue and $31.9m of backlog should not carry a full slot.

Correlation warning — this is the material one for the book. CRDO's revenue driver is hyperscaler AI capex, which is the identical single factor driving NBIS, MU, SNDK, AMZN, MSFT, META and GOOGL already in this corpus. A CRDO position adds concentration to that factor, it does not diversify it. If the book is already at its limit on AI-infrastructure exposure, that constraint binds before any of the analysis above matters.

---

6. Entry and monitoring

Entry. Momentum Criteria is MEASURED and governs timing, not admission. The reading is mixed and deteriorating: 12-1 momentum +135.1% (top decile cross-sectionally) but −19.2% over the last month, with the reversal coincident with the Q4 sequential-growth break. The market is already repricing the finding this memo makes. Two coherent readings:

• Wait for the Q1 FY2027 print (~early September 2026, not company-confirmed). It resolves the single largest uncertainty — whether mid-single-digit sequential is a floor or a way-station — at the cost of giving up the move.
• Scale in ahead of it on the view that a 36% drawdown already discounts a deceleration that management pre-announced five months ago.

The memo does not choose. It notes that the catalyst is close, dense and binary, which argues for either a small pre-print position or none.

Invalidation triggers — any one of these breaks the thesis as written:

1. Q1 FY2027 revenue below the $465m guidance floor.
2. A fourth consecutive quarter of sub-8% sequential growth with no disclosed optical revenue contribution.
3. Days of inventory above ~180 (currently 164, up 16 sequentially).
4. Any FY2027 10-Q showing a top-three customer stepping down materially in the concentration table.
5. The ">$500m optical revenue in fiscal 2027" target being withdrawn, deferred or quietly not repeated.
6. GAAP operating margin below 28% for two consecutive quarters (Q1 FY2027 is guided to ~31.9%).

Confirmation triggers:

1. Optical revenue disclosed and tracking to the $500m FY2027 target.
2. Sequential growth re-accelerating above 10% on an optical contribution.
3. A fourth material end customer appearing in the concentration table (concentration falling, not just rotating).
4. Days of inventory falling back below 150.

---

7. Ledger entry — pre-registered before the outcome is known

Per references/recommendation-ledger.md. No specific contract is being recommended, so no trade line is appended to trade_recommendations.jsonl. What is pre-registered is the analytic call, for scoring:

The pre-registered falsifiable claim, for the record: FY2027 revenue lands between $1.72bn and $2.25bn, and Credo discloses optical revenue as a separate line or a stated figure at least once during FY2027. Both are checkable within four quarters.