CRDO — Financial Model Notes

Every figure below is traceable to a primary filing or a named API. Reproduction scripts are in data/. As of 2026-07-29. Fiscal year ends the Saturday nearest 30 April.

---

1. Sources

Recency assertion: the most recent filing is the FY2026 10-K, filed 2026-06-15 — 44 days old. The most recent reported period ended 2026-05-02, 88 days before this memo. Q1 FY2027 has not been reported. No data here is stale in the GOOGL (485-day) or Alcon (decade-old) sense.

---

2. Revenue — quarterly series, derived

XBRL tags cumulative fiscal-year-to-date periods only. Q4 is derived as FY − Q3 cumulative; Q2 and Q3 as differences of cumulative periods. Reproduction: data/multiples.py.

Fiscal-year totals: FY2020 $53.8m · FY2021 $58.7m · FY2022 $106.5m · FY2023 $184.2m · FY2024 $193.0m (+4.8%) · FY2025 $436.8m (+126.4%) · FY2026 $1,335.1m (+205.7%).

---

3. Income statement, FY2026 (as filed)

Scale cross-check: $472.279m ÷ 188.232m = $2.509 vs filed $2.51. Passes. Tax: provision is 0.2% of revenue — Cayman Islands domicile. Any model must not apply a US statutory rate.

Quarterly FY2026: gross profit Q1 150.368, Q2 181.046, Q3 278.868, Q4 298.067 (Q4 derived). Q4 gross margin 68.2%. Q4 operating income ≈ $155.9m (revenue 437.003 − COGS 138.936 − opex 142.2 as stated in the release).

---

4. Balance sheet and the correction

No debt. LongTermDebtNoncurrent is absent from companyfacts and total liabilities are operating in nature.

The correction the model must apply

10-K Note 16 discloses that DustPhotonics closed in May 2026 for $770m cash, after the balance-sheet date. Net cash used in valuation: $673.3m, excluding any Q1 FY2027 cash generation. Share count adds ~0.8m for the deal, with up to 2.8m contingent shares not included.

---

5. Cash flow, FY2026

Accrual quality: OCF $464.3m vs net income $472.3m → accrual ratio ≈ −0.02. Earnings are cash-backed despite a $174.0m inventory build.

Share-based compensation is 13.7% of revenue and 41.0% of GAAP operating income. Non-GAAP net income of $662m for FY2026 (company-stated) versus GAAP $472.3m is very largely this. Any model should carry SBC as a real cost.

---

6. Working-capital diagnostics

DSO on a 91-day quarter; days inventory on derived quarterly COGS. Quarterly gross profit is only derivable from FY2026 onward from cumulative tags, hence the blanks.

Deferred revenue (ContractWithCustomerLiabilityCurrent) is $0 at 2026-05-02. Remaining performance obligation is $31.9m (10-K Note 4) — 2.4% of FY2026 revenue.

---

7. Customer concentration — the source tables verbatim

Direct customers, % of total revenue: FY2026 A 49%, B 32%; FY2025 A 67%; FY2024 A 39%, C 15%. Accounts receivable: FY2026 A 53%, B 20%, C 19%; FY2025 A 86%. End customers (supplemental look-through): FY2026 D 33%, B 32%, E 19%; FY2025 D 63%; FY2024 Z 26%, Y 20%, B 15%; FY2023 Z 55%, E 13%, C 12%. Top 10 customers: ~90% of FY2026 revenue (risk factor).

* in the filings means "less than 10%", so entries shown as <10% are bounded, not zero.

---

8. Geography — revenue by ship-to destination

Destination of shipment, which the filing notes "may differ from the end customer's principal offices."

---

9. Product/IP revenue split — the disclosure that stopped

Bounded by the FY2025 figure at ~3% of $437m ≈ $13m — so the omission cannot conceal a material amount.

---

10. Market data

Within-year peak-to-trough drawdowns: 2022 −39.3%, 2023 −62.0%, 2024 −33.4%, 2025 −61.1%, 2026 YTD −45.6%.

---

11. EV/TTM-sales history — construction

Computed daily, 2022-06-06 → 2026-07-28, n=1,039. For each trading day: TTM revenue from the last four fiscal quarters whose results were public (35-day filing lag applied), reported shares outstanding, reported net cash on the same lag. Basis is constant across the whole series, so percentiles are internally comparable — but it uses outstanding shares and reported net cash, so the current 25.6x is on the screen's basis, not the corrected 27.1x. Both are stated wherever used.

Percentiles are in CRDO_Valuation.md §5.2. Reproduction: data/multiples.py and data/regime.py.

---

12. Reverse DCF — exact invocations

python3 reverse_dcf.py --spot 191.63 --shares 186.478 --net-cash 1443.286 \
    --revenue 1335.116 --years 5 --wacc 0.10 --terminal-margin 0.164 --exit-multiple 27.0
    ->  56.3%   (reproduces the screen exactly)

python3 reverse_dcf.py --spot 191.63 --shares 192.0 --net-cash 673.3 \
    --revenue 1335.116 --years 5 --wacc 0.10 --terminal-margin 0.25 --exit-multiple 27.0
    ->  45.2%   (base case)

Full sensitivity grids over exit multiple and terminal margin are in CRDO_Valuation.md §3.5. Reproduction: data/reverse_dcf_grid.py.

---

13. Known limitations of this model

1. No consensus estimates — Alpha Vantage quota exhausted. All forward paths are built from company guidance and the observed sequential series, not from Street numbers.
2. No speaker-tagged transcripts — mention frequency runs on 8-K Ex-99.1 releases, so the prepared-remarks / Q&A split required by the method could not be produced.
3. DustPhotonics' standalone financials are undisclosed — FY2027's organic/inorganic split is not computable.
4. Q4 quarterly figures are derived by difference from cumulative XBRL periods, not directly tagged.
5. Peer multiples are point-in-time; ALAB's net cash is understated (absent tag) so its EV/sales is an upper bound.
6. Post-year-end cash is estimated, not filed; ~$100m of Q1 FY2027 free cash flow is excluded from the valuation base as a conservatism.
7. No Excel model is produced. The analysis that decides this name is the reverse DCF and the own-multiple-history percentile, both of which are scripts in data/ and both of which are reproducible; a three-statement build would add presentation, not information, and the time-box was spent on the accounting-quality and concentration work instead. Recorded as a deliberate omission, not an oversight.