---
ticker: "CRDO"
company: "Credo Technology Group Holding Ltd"
date: "2026-10-23"
view: "Hold"
fair_value: 153.36
price: 147.98
price_source: "last regular-session trade at 10:13:45 ET; previous official close $162.77 (2026-10-22)"
price_time: "2026-10-23T14:13:45Z"
price_vs_fair_value: "-3.5%"
bear: 29.17
bull: 406.2
exit_price: 290.36
p_price_justified: "20-50%"
restated: "2026-10-26"
method: "case-spread-2026-10-24"
original_fair_value: 153
original_view: "Buy"
page: "https://alpha.dming.si/research/2026-10-23-crdo"
---

# CRDO · Credo Technology Group Holding Ltd · Hold

Research report as of 2026-10-23, partly unblinded. Page: https://alpha.dming.si/research/2026-10-23-crdo

We value CRDO at $153 a share. The report's $147.98 price (Oct 23, 2026) was 4% below fair value: Hold.

_Restated Oct 26, 2026 (method case-spread-2026-10-24); the report's original value was $153, Buy._

| Lens | Range | Marks |
| --- | --- | --- |
| Our report | $29.17 to $406 | fair $153; original $153 |
| Our simulation | $43.09 to $333 | median $127 |
| Street targets (5) | $206 to $310 | average $267, median $272 |
| Price | | $147.98 |

Bands run from our bear to bull (the hollow mark is the original $153), our simulation's 10th to 90th percentile and the Street's low to high. Street targets forecast future prices, not value today. Market data, captured Oct 23, 2026.

| Figure | Value |
| --- | --- |
| Fair value | $153 |
| Price | $147.98 (market data, Oct 23, 7:13 AM PT) |
| Five-year return | 13.4% (a year, at the report's price) |
| Bear / bull | $29.17 / $406 |
| Exit price | $290 (on May 1, 2032) |
| Confidence | 20% |
| Price against our scenarios | The price is justified in 20-50% of our scenarios (4,000 simulated outcomes, each drawing the uncertain drivers from ranges we chose: our judgement, not a measured chance) |

**What the price needs.** Any one of these, on its own, would justify the price:

- FY2028 revenue vs our $4.07B: needs 0.96x, we have 1.00x
- Optical FY2029-32 vs our path: needs 0.92x, we have 1.00x
- Copper growth vs our path, FY2029-32: needs -1.4%, we have 0.0%

_This text argues the original call. The current figures are at the top._

**Original call, Oct 23, 2026 (before the Oct 26, 2026 restatement)**

_As of 2026-10-23 · partly unblinded: one deal-share row in the Q1 FY2027 10-Q implied the 2026-05-28 price, and the accelerator-spend path came from the semis-ai sector note, read before the blind commit (assumptions.md lists six exposures). The blind value was $141; outside review moved it to $153 (section 9)._

- **Fair value today:** $153 per share, discounting at a 12.62% cost of equity. Weighting the pessimistic, central and optimistic cases 25/50/25 (our weights, not calibrated) gives $168.
- **Price:** $147.98, market data, last regular-session trade 2026-10-23 10:13:45 ET → 3.3% below fair value
- **Five-year exit price:** $290 a share on 2032-05-01: the average of two estimates, one from letting Credo's returns fade toward its cost of capital ($184) and one from what similar chip companies have traded at ($396, at 27.5x operating profit before acquisition amortization). Fading over 10 or 20 years instead gives $285 or $295.
- **Five-year return:** 13.0% a year from $147.98 to the central exit (Credo pays no dividend and buys back no stock), against a 12.6% cost of equity
- **Bear / bull today:** $35 / $329: our pessimistic and optimistic cases, each one consistent business outcome.
- **Confidence:** About 20%: our judgment of the share of simulated values within 20% of $153 (subjective, not calibrated)
- **Moat:** Our judgment: Credo created the active electrical cable market and leads it, on its own SerDes (high-speed signalling) designs. It owns no factory; its edge is speed to market and power use, and competitors have entered.
- **Market expects:** Less than our central: fiscal 2028 revenue 4% below our $4.07B, or the peers' multiple at about 25.9x instead of 27.5x, with everything else as in our central case
- **Price needs:** Any one of: fiscal 2028 revenue at 0.96x ours; optical revenue FY2029-32 at 0.92x our path; copper growing 1.4 points a year slower; or an operating-margin cap of 48.7% instead of 50%
- **Odds:** P(price justified): 20-50%, the share of our simulated scenarios in which the value is at or above the price. Eight in ten simulated values fall between $43 and $333; the median is $127 and the mean $164. Subjective, not calibrated (odds.json)
- **Key driver:** Fiscal 2028 revenue and the optical ramp: each $100M of fiscal 2028 revenue is about $3.6 a share; optical at 0.5x / 1.5x our path is -$42 / +$34; copper growth 10 points slower / 5 faster is -$33 / +$21. Credo's own beta instead of the peers' costs $30.
- **Street:** Average 12-month target $267 (median $273, range $206-310): 30 targets published over the last year averaging $251, the last quarter's 5 at $271 and the last month's 1 at $310 (market data, fetched 2026-10-23; market data gives no date for each target). An expectations reference, not validation

**Buy, at the edge of Hold: at $147.98 Credo trades 3% below our $153 fair value.** Our
blind work, done before we looked at the price, said $141, a Hold at this price. Outside
review found three errors that together add $12.70 a share, and we accept them (section 9).
The largest is a data gap: Marvell, one of our six comparison companies, tagged none of its
fiscal 2018-2021 amortization of acquired intangibles in its machine-readable SEC data, so
our like-for-like peer multiple left those years out. From Marvell's 10-K cash flow
statements the peers' multiple is 27.5x, not 24.2x: +$10.93. All three changes rest on
filings, not on the price, and all three moved the value up; the call changed with them.

The margin of safety is thin, and the stock is volatile. A value 3% above the price is a
Buy under our rule (buy at or below the central fair value), but in our own simulated
scenarios the value reaches the price in only 43% of draws, Credo's own beta would cost
$30, and the chart shows no support until the $86-101 base of last spring (section 4).

Our value takes management's fiscal 2027 guide (revenue up more than 85%, to about $2.5B)
and its "billion-dollar quarter" early in fiscal 2028, which makes fiscal 2028 revenue
$4.07B, then slows growth to 8% by fiscal 2032 as competitors arrive in copper cables and
the optical lines mature. We hold the operating margin before stock pay at 50%, the level
management guides for fiscal 2027; Credo gives no long-term target.

Terms. The **central case** is our best estimate; the **pessimistic** and **optimistic**
cases are coherent alternative outcomes we weight 25% each. **Copper** is Credo's active
electrical cables (AECs: copper cables with Credo chips at each end that link servers and
switches within a rack, up to 7 m), its retimer chips and licensed designs. **Optical** is
its optical signal processors (DSPs), the silicon photonics chips it bought with
DustPhotonics in May 2026, and its own ZeroFlap optical transceivers. The **fade** values the
business after fiscal 2032 by letting its returns fall toward the cost of capital; the
**outside view** values it at what comparable companies traded at. Credo's fiscal year ends
near 30 April: fiscal 2028 runs May 2027 to April 2028.

## 1. What Credo sells and what limits it

Credo's revenue rose from $193M in fiscal 2024 to $437M in fiscal 2025 and $1,335M in fiscal
2026 (market data); AECs drove more than 90% of the Q1 fiscal 2027 increase (10-Q MD&A). It is
fabless: TSMC makes its chips (10-K). Three things limit it:

- **Few buyers.** By contracting party, the company that places the order, often a contract
  manufacturer: Customer A was 43% and Customer B 28% of Q1 fiscal 2027 revenue, A 49% and B
  32% of fiscal 2026, and A 67% of fiscal 2025 (10-Q Q1 FY2027 note 3; 10-K FY2026 note 3).
  By end customer, the company that deploys the cables (its own table): four were 33%, 28%,
  13% and 10% of Q1 fiscal 2027; in fiscal 2025 one was 63%. The filings name no revenue
  customer; they name Amazon as a warrant holder and Microsoft and Oracle as development
  partners. Management has named xAI and Microsoft as AEC customers (calls). One program won
  or lost moves revenue by a quarter or more.
- **The cloud build itself.** Credo's revenue is 0.43% of our sector note's accelerator
  spending in fiscal 2027; our central lifts it to 0.76% by fiscal 2032 as optics adds to
  copper (output.txt). The sector note holds the big buyers' spending flat in 2027-2029.
- **Competition in copper.** Competitors have entered AECs (CEO, Goldman Sachs Communacopia
  conference, 2026-09-10). Credo's non-GAAP gross margin is 68.0-68.6% in the last four
  quarters; it was 57.7% in fiscal 2023 and 61.9% in fiscal 2024, before the ramp (market data). We
  fade copper's gross margin by 0.75 points a year.

## 2. Recent results against the bar

Q1 fiscal 2027 (to 2026-08-01): revenue $479.0M against a $470M guide midpoint (+1.9%);
non-GAAP gross margin 68.0%; non-GAAP operating cost $95.2M against $88M guided. Credo has
beaten the midpoint of its own revenue guide in all eight quarters we read, by +1.6% to
+19.7% (median +10.4%), but by only +1.6% and +1.9% in the last two, after the ramp matured
(releases; preview.txt). We read that as a measured pattern, not as a statement of intent.

The Q2 guide is $525-535M of revenue, a 67-69% gross margin and $100-105M of operating cost.
For the year management guides revenue up more than 85%, optical above $600M and non-GAAP
net margin near 50% (Q1 FY2027 call, 2026-09-01). Its finance chief spoke of about +20%
sequential growth into Q3 and +30% into Q4, a "billion-dollar quarter" early in fiscal 2028,
and possibly revenue growing twice as fast as operating cost entering fiscal 2028, which he
qualified (Goldman Sachs conference, 2026-09-10; Quartr transcript, not saved in data/).

Stock pay was $88.0M in Q1 fiscal 2027, 18.4% of revenue, after the DustPhotonics
replacement awards and a special performance award to the CEO (10-Q). We charge stock pay
as a cost, at 12% of revenue in fiscal 2028 falling to 9%, less the cost of awards already
granted by 2026-05-02: those shares are already in our share count, so their remaining
$404.9M of expense (10-K, over about 2.3 years) is not charged again. Awards granted from May
to August, including the DustPhotonics replacement awards, are in the share count too, but
the 10-Q gives no remaining expense for them, so their cost is still charged: each $100M of
it would add about $0.26 a share.

## 3. Our forecast and value

The model (model.py, output.txt) forecasts copper and optical separately to fiscal 2032 and
values them with the shared method: free cash flow kept in the business (Credo plans no
buyback or dividend; CFO, Q4 FY2026 call, 2026-06-01) plus an exit price in May 2032.

| Fiscal year | 2028 | 2029 | 2030 | 2031 | 2032 |
| --- | --- | --- | --- | --- | --- |
| Copper ($B) | 2.73 | 3.14 | 3.39 | 3.59 | 3.77 |
| Optical ($B) | 1.34 | 1.95 | 2.34 | 2.69 | 3.01 |
| Revenue ($B) | 4.07 | 5.08 | 5.72 | 6.28 | 6.78 |
| Gross margin | 68.0% | 67.2% | 66.5% | 65.8% | 65.1% |
| Operating margin before stock pay | 50.0% | 50.0% | 49.4% | 48.3% | 47.0% |
| Operating profit after stock pay ($B) | 1.72 | 2.03 | 2.25 | 2.47 | 2.58 |
| Free cash flow ($B) | 1.18 | 1.34 | 1.62 | 1.82 | 1.91 |

**Margin.** Credo gives no long-term margin target, so we assume one: the operating margin
before stock pay is capped at 50%, management's fiscal 2027 net-margin guide and near the
48.2-49.6% of the last two quarters. It binds in fiscal 2028-29. Without it, operating cost
growing at half of revenue growth would carry the margin higher: worth +$21.

**Tax.** 15% from fiscal 2028, the global minimum the 10-K says may apply; about 1% today.

**Exit.** Letting returns fade toward the 12.6% cost of capital gives $184 a share; the
peers' 27.5x EV over operating profit before acquisition amortization (Marvell 30.7x,
Broadcom 19.3x, MACOM 40.1x, Coherent 18.3x, Lumentum 24.2x, Semtech 31.6x, ten-year medians
of fiscal year-ends; anchor_ebita.txt) gives $396. The method averages them: $290. The
peers' ratio is put on our basis: our operating profit is after stock pay and before
amortization of acquired intangibles, so the peers' amortization is added back (SEC XBRL,
and Marvell's 10-K cash flow statements for fiscal 2018-2021, which its XBRL lacks). At the
method's default 34.9x, which divides by profits after amortization, the central would be
$178. Semtech's amortization looks undertagged after 2023, which raises its ratio; it sits
above the median, so it does not set the multiple. The two estimates differ by 115% because
the fade, at a 12.6% cost of capital and a 25% long-run return on new capital, pays 11.9x
year-six after-tax operating profit, about 10x before tax, against the peers' 27.5x: the
peers' history prices a growth runway the fade does not grant. The method keeps both halves.

**Beta.** We use the peers' median beta, Blume-adjusted toward 1: 1.746, a 12.62% cost of
equity. Credo's own measured beta is 3.25 (market data), 2.51 adjusted, far above every peer but
Astera Labs; at that beta the cost of equity is 15.8% and the central is $124.

**Shares.** 200.9M today, counted gross: 187.8M issued, 8.3M restricted units, 3.1M
performance units and 1.7M options (10-Q note 10). The DustPhotonics earnout adds up to 2.8M
shares and $31.6M cash on financial milestones (10-K note 16). Credo books the cash part at
$10.0M, 32% of the maximum ($310.0M contingent consideration less $300.0M recorded in
equity); our central takes 32% of both parts (0.89M shares, $10.0M), the pessimistic none and
the optimistic all. That share is the cash leg's fair value, not a stated probability for the
shares: none costs nothing and adds $0.70; all costs $1.50.

**Claims today.** Cash $466.9M and certificates of deposit $297.4M, less $9.5M of licence
fees and $200M of operating cash we assume the business keeps; no debt (10-Q, 2026-08-01).

The pessimistic case ($35) is the sector note's pessimistic shape (flat 2028, then -20%) with
one AEC program lost: fiscal 2028 revenue $2.47B, flat on fiscal 2027, then $2.05B in fiscal
2029, gross margin falling to 58.6% and a mature return of 15%. Operating cost keeps rising
3% a year through the fall, as it did in fiscal 2024, when revenue grew 5% and operating cost
23% (market data). The optimistic case ($329) has fiscal 2028 revenue of $5.1B, optical overtaking
copper by fiscal 2031 and a 55% margin.

## 4. Market expectations and how the price reacted

Weekly closes (market data, regular session): $95.24 in the week of 2026-03-23, the year's
low; $271.83 in the week of 2026-06-15, with an intraweek high of $308.67 the week after;
$232.75 in the week of 2026-08-24, then $170.57 the week of the Q1 results (2026-09-01),
down 27% while SOXX rose 2%; that quarter beat the revenue guide by 1.9%, against a median
of 10% before, but we did not check the week's news for another cause. $218.64 in the week
of 2026-09-28; $169.70 in the week of 2026-10-12, down 22% while SOXX fell 12% in the
sector's longest losing streak in more than two decades (Seeking Alpha headline,
2026-10-17). On 2026-10-22 Credo closed at $162.77; at 10:13 ET on 2026-10-23 it traded at
$147.98, down 9.1%, while SOXX was up 0.5% at $478.26 (market data). The vendor's headline
feed showed no company news for the day's fall by then; we did not search further. From the
June high Credo is down 52%; SOXX is down 27% from its own high ($655.95 intraweek, week of
2026-06-22). The stock recovered during the day: market data quoted $153.70 when research-check.py ran
later on 2026-10-23, at our fair value; the view uses the market-pass price.

Running our model backwards (odds.json), the price is justified by any one of: fiscal
2028 revenue 4% below ours; optical revenue 8% below our path; copper growing 1.4 points a
year slower; or a margin cap of 48.7%. The price asks slightly less than our central, which
itself takes management's fiscal 2027 guide and its fiscal 2028 ambition at face value.

The simulated median, $127, is 17% below our $153 central, and the mean, $164, is above
it. Both come from the shape of the scenarios, not from a different central: the
pessimistic world has a median of $55 and the optimistic one $234 (odds.json), and each
draw also scales the exit price by a lognormal factor with a median of 1 and a mean of about
1.14, which spreads values more upward than downward. Without that factor the simulated
mean and median sit close together, below the central (the advocate's run on the blind
model: mean $136, median $132 against $141). The odds count where the price falls in that
spread: 43% of draws are at or above $147.98.

Alpha, our momentum and fundamentals ranking model, ranks Credo 132nd of 702 on 2026-10-23
(tier Strong): growth A, quality B, momentum C, expectations D, valuation D (model cache
partly stale). It ranks on growth and quality; we value the business. The two agree that
the business is strong and that the stock is not cheap on its multiples.

Timing, a quick chart read from the weekly bars: a downtrend from the June high, with lower
highs in August ($286) and October ($232), and today's trade below the September lows of
$149-159. The next support on the chart is the March-April base at $86-101. Grade D: the
chart gives no reason to hurry. Timing is beside the view, never an input to it.

## 5. Where we differ from the Street

The Street's revenue is ours: its averages for fiscal 2027-31 are $2.5B, $3.9B, $5.0B, $5.7B
and $6.2B (market data, 2026-10-23; ours $2.52B, $4.07B, $5.08B, $5.72B, $6.28B). So are its profits,
roughly. Loading each Street year into our model in turn (street_check.py, street_check.txt;
the model auditor's method):

| Step | Central value | Change |
| --- | --- | --- |
| Our central | $153.36 | |
| The Street's revenue, fiscal 2027-31 | $151.77 | -$1.59 |
| The Street's EPS, fiscal 2027 | $151.74 | -$0.03 |
| The Street's EPS, fiscal 2028 | $152.53 | +$0.79 |
| The Street's EPS, fiscal 2029 | $153.33 | +$0.80 |
| The Street's EPS, fiscal 2030 | $153.78 | +$0.46 |
| The Street's EPS, fiscal 2031 | $153.50 | -$0.29 |
| Fiscal 2032 at the Street's fiscal 2031 operating margin | $151.13 | -$2.37 |
| The rest, to the average target discounted one year at 12.62% ($237.37) | | +$86.24 |

The Street's EPS ($9.67 for fiscal 2028, $12.38 for fiscal 2029) is a little above ours
($8.78 and $11.18: operating profit before stock pay, our 15% tax, interest on kept cash), and its fiscal 2031 EPS falls on higher
revenue; with our stock pay and our 15% tax, its earnings path is worth about what ours is.
The difference is the price on those earnings: the average target is 27.7x the Street's
fiscal 2028 EPS. Revenue and EPS alone do not show the Street's tax, stock-pay treatment or
valuation method, so we do not assign the $86 to one driver. What would move us toward the
Street is a margin held above 55% through fiscal 2028 or growth that holds up after fiscal
2029; both are in our optimistic case ($329).

## 6. Why the other side could be right

The case for Hold or worse: concentration and the multiple. A single contracting customer
was 67% of fiscal 2025 revenue, and AEC competitors have entered. An AI spending pause in
2028 with one program lost is our pessimistic case: revenue flat, then down 17%, and $35 a
share. Credo's own measured beta, 3.25, says the stock has moved like that before; at that
beta the central is $124. And our value is above the price only because of the outside view:
at the fade alone the exit would be $184 and the value well below the price. A peer multiple
of 25.9x instead of 27.5x leaves the price exactly at value.

The case for more, made by our opposite-side reviewer: the 50% margin cap rests on a fiscal
2027 net-margin guide, not a long-term limit; at 55% the value rises $21. Credo announced
1.6T optical transceivers using DustPhotonics silicon photonics (2026-09-15), a 50-metre
active light cable shipping in 2027 (2026-10-12) and an OmniConnect platform with AMD
(2026-10-13); optical at 1.4x our path adds $27. Both together, on the blind model with the
earnout paid in full, gave $187. These are launches without revenue yet, and we did not move
our path on them.

## 7. Next results: Q2 fiscal 2027, expected 2026-11-30 or 2026-12-02 (not confirmed by the company)

Earnings calendar lists 2026-11-30 after the close, marked unverified, and market data
lists 2026-12-02 (both fetched 2026-10-23); we found no announcement from Credo. Last year's
Q2 was reported on 2025-12-01, and our blind record said about 2026-12-01. The record grades
the forecasts from 2026-12-02, the later date. Ranges aim to hold about 80% of outcomes, set
from Credo's record against its own guidance midpoint (preview.py, preview.txt; unchanged by
the review).

| Forecast | Ours | 80% range | Why |
| --- | --- | --- | --- |
| Revenue ($B) | 0.539 | 0.525-0.585 | guide $530M plus the last two beats (1.8%); high end at the eight-quarter median beat |
| Non-GAAP gross margin (%) | 68.3 | 67.0-69.0 | last three quarters; guide 67-69% |
| Non-GAAP operating expenses ($M) | 110.5 | 100-116 | guide midpoint plus the last three quarters' overrun (7.8%) |
| Diluted EPS ($, non-GAAP) | 1.31 | 1.18-1.55 | the above, 1.2% tax, about 200M shares |
| Diluted EPS ($, GAAP) | 0.76 | 0.50-1.12 | stock pay about $97M, amortization $11.6M, deal costs |
| Third quarter FY2027 revenue outlook, midpoint of the range ($B) | 0.647 | 0.604-0.743 | our Q2 x 1.20 (CFO: about +20% into Q3) |

The quarter matters through fiscal 2028: each $100M of fiscal 2028 revenue above ours is
worth about $3.6 a share.

## 8. Dates to watch, and what would change the view

- 2026-11-30 or 2026-12-02 (unconfirmed): Q2 results and the Q3 guide; optical progress
  against "more than $600M" for the year.
- The four big cloud buyers' 2027 capital spending guides (late January to early February
  2027): a cut of 20% or more is our pessimistic path.
- Q3 fiscal 2027 results (about early March 2027): the first read on a billion-dollar
  quarter in early fiscal 2028. Fiscal 2028 revenue below about $3.9B (0.96x ours) takes the
  value below today's price.
- Customer notes in each 10-Q: a contracting customer above 25% of revenue that cuts or
  ends a program is our pessimistic path.
- A non-GAAP operating margin above 55% for fiscal 2028: our cap is too low.

## 9. What changed from the blind value, and the review

The blind commit (ac8d96c) valued Credo at $141. A report check (research-eval, Claude Opus
5.5) and two outside reviewers (Codex, run separately: a model auditor and an
opposite-side advocate arguing Buy) reviewed draft cba616890. Each effect below is the
central value with that one change undone (output.txt, model.py); together they move the
central from $140.66 to $153.36, +$12.70 or 9%.

| Finding (who) | Answer | Central effect |
| --- | --- | --- |
| Peer multiple left out Marvell's fiscal 2018-2021, which its XBRL does not tag (auditor) | Fixed from Marvell's 10-K cash flow statements (FY2019, FY2021; data/): 24.2x to 27.5x | +$10.93 |
| Stock pay for awards already in the share count charged again (report check) | Fixed: the 10-K's $404.9M of remaining expense on granted awards, over 2.28 years, is not charged again | +$1.02 |
| Fiscal 2027 working capital on a run-rate basis, fiscal 2028 on annual revenue (auditor, report check) | Fixed: fiscal 2028 builds from the fiscal 2027 exit run rate | +$0.76 |
| Gross-margin draws pass through the 50% margin cap (auditor) | Fixed: the cap holds when gross margin is higher | $0 (odds only) |
| The Street check fitted one EPS year and called the rest the multiple (both reviewers) | Fixed: every Street year loaded in turn; the remainder is not assigned to one driver | none |
| Margin cap at 55% (advocate) | Rejected: the only support is the CFO's qualified remark on 2026-09-10, read before the blind commit; shown, +$21.17 | none |
| Optical at 1.4x on the September-October launches (advocate) | Rejected: launches without revenue; shown, +$27.24 | none |
| Copper competition charged in both growth and gross margin (report check) | Rejected: our copper growth below the 31%-a-year unit market Credo cites is share; the margin fade is price, toward the 57.7-61.9% of fiscal 2023-24. No fade is +$6.79 | none |
| Pessimistic operating cost rises while revenue falls (report check) | Rejected: in fiscal 2024 revenue grew 5% and operating cost 23% (market data); falling cost was +$3.64 on the blind pessimistic case | pessimistic only |
| Opening cash counted today, not carried to the exit (auditor) | Disclosed: a convention; carried in the exit bridge it is -$0.93 today and +$3.51 on the exit | none |
| Earnout share fraction taken from the cash leg's fair value (auditor, advocate) | Disclosed: none +$0.70, all -$1.50 | none |
| Capacity deposits never refunded (report check) | Disclosed: +$0.27 if refunded by fiscal 2032 | none |
| Gross-margin range widened after the blind commit for the odds tool (report check, advocate) | Disclosed: the median must sit in the middle of the range; odds only | none |
| Fiscal 2028 quotes from unsaved transcripts (report check) | Disclosed: Q4 FY2026 call (2026-06-01) and Goldman Sachs conference (2026-09-10), Quartr; the advocate found the 2:1 remark in a public transcript | none |
| Q1-week fall attributed to the small beat (report check) | Fixed: stated as a fact beside SOXX, cause unchecked | none |
| Model output called today's claims part of the exit bridge (report check) | Fixed: label only | none |

All three accepted changes rest on filings, not on the price, and all three moved the value
up, past the price: the view changed from Hold to Buy. That is flag 2's case, so the model
auditor reran on the revised model (below). The move, 9%, is under the 20% at which we flag
a review's effect for the reader.

The model auditor's second round (Codex, on b2ceb91de) reproduced $153.36 and each fix's
effect with no interaction left over, and found no correction that reverses the Buy. It
found three smaller points: the stock-pay relief covers awards granted by 2026-05-02 while
the share count includes those granted to 2026-08-01 (disclosed in section 2, about +$0.26
per $100M of their remaining expense, not measurable from the 10-Q); the report misstated
the odds in one sentence (fixed); and the PSUs' 2.15-year period was rounded to the RSUs'
2.28 years ($0.00). odds.json was rerun on the final drivers.json.

## 10. Sources

Filings masked before reading and pinned in data/: 10-K FY2026; 10-Qs Q2 and Q3 FY2026 and
Q1 FY2027; releases Q1 FY2025-Q1 FY2027; 8-Ks 2025-10-06 to 2026-10-16. Marvell 10-Ks for
fiscal 2019 and 2021 (amortization only, fetched after review). Transcripts (Quartr via
Market data): Q1 FY2027 call (2026-09-01, saved), Q4 FY2026 call (2026-06-01) and the Goldman
Sachs Communacopia conference (2026-09-10). Market data statements, segments and peer betas
(2026-10-23). Peers' amortization: SEC XBRL company facts (2026-10-23). Sector note
docs/research/sectors/semis-ai/note.md (2026-10-15). Market pass: market quote, weekly
bars and earnings calendar (2026-10-23); headlines feed (2026-10-23); Alpha rank
(2026-10-23); Street targets and estimates via scripts/research-check.py (2026-10-23).