Credo Technology Group Holding Ltd (CRDO)

Priced as of 2026-09-11 — market data goes stale within days; treat this as a starting point, not a current quote.

A Sum-of-the-Parts (SOTP) valuation — pricing each Strategic Business Unit (SBU) on the multiple its own BCG Matrix quadrant justifies, then adding the pieces together — puts Credo's central fair value at **$145.47 per share**, about **10.3% below** the current $162.20 price. That is barely past the −10% line where this pipeline's Decision Framework switches from HOLD to AVOID, and the conviction is Low for three honest reasons. First, the fair-value band runs from $57.15 to $211.32 — a 3.7-fold spread — and the current price sits **inside** it. Second, the answer flips to HOLD (a −7.2% gap) if you divide by the 187.95 million basic shares that data providers use instead of the 194.4 million diluted shares this methodology specifies; that is a bookkeeping convention deciding the verdict, which is not a sound place for a decision to rest. Third, roughly 74% of the fair value sits in one SBU — Active Electrical Cables (AECs) — whose multiple comes from the midpoint of a comp range whose two ends differ by 3.7 times. What has genuinely changed since August is that the argument is now close. The prior run found the stock worth about a quarter of its price. Work backwards from today's price and the market is implicitly paying about **20.4x** revenue for the AEC business — a number that sits comfortably *inside* the defensible peer range of 7.57x to 27.68x, not outside it. The market and this method are no longer describing different companies; they are arguing over where in one wide range the right multiple sits. Momentum has also inverted: the stock is down about 32% in 30 days and has returned **−1.07%** over the trailing year against **+108.02%** for the iShares Semiconductor ETF (SOXX), and its revenue multiple has compressed from 27.5x to 18.85x. This is no longer a stock running away from its fundamentals — it is a stock that has already given most of that run back, and still screens slightly expensive on a quadrant-disciplined sum of its parts.

Fair-value band

Rather than a single point estimate, this valuation reports a range: every discretionary judgment call set to its defensible low end, its midpoint, and its high end.

Price sits inside the band — the method does not resolve this name. The decision below comes from the central estimate, but the honest read is that it depends on assumptions this analysis could not settle.

Portfolio at a glance

Key assumptions & swing factors

• AEC Star valued at 17.63x EV/Revenue, the Rule A midpoint of the defensible peer range 7.57x (Amphenol) to 27.68x (Lumentum), stockanalysis.com, 11 September 2026. This single input drives 74% of Total Fair Enterprise Value; each 1x of multiple is worth about $5.95 per share.
• Rule A cross-check PASSED: the endpoints differ by 3.66x, triggering the second-metric test. 17.63x EV/Revenue at Credo's disclosed GAAP EBITDA margin of 27.6% (Q1 FY2027 GAAP operating income $120.7M plus $11.6M amortisation, on $479.0M revenue) implies 63.9x EV/EBITDA, inside the comp set's own 23.20x-104.20x EV/EBITDA range and within 0.3% of its 63.70x midpoint. No endpoint was dropped. The range remains wide on both metrics, so per Rule A step 3 the result is reported as unresolved and conviction is capped.
• Rule B applied once per SBU and stated: for AEC the Job Audit 'Star on borrowed time' / Contingent verdict selects the comp anchor (it is why the midpoint is used rather than deviating upward toward Lumentum, whose 62.54% forecast growth matches AEC better than Amphenol's 26.98%) and NO haircut is applied to the multiple; for the optical SBU the 'Blocked' verdict selects the harvest comp (MaxLinear at 12.01x) and does NOT also cut the probability weight.
• Question Mark weighted at the default 50/50 invest-and-win / harvest. Quantified signals point both ways and offset: Marvell plus Broadcom hold over 90% of the high-end PAM4 DSP market, against reiterated FY2027 guidance of more than 80% YoY optical growth in a 17% CAGR market. Market-implied probability is unsolvable: the price requires $10.11B for this SBU while the 100% invest-and-win value is only $8.90B, a $1.21B shortfall (versus roughly $14B in the superseded August valuation).
• Rule C base refreshed: valuation runs on LTM revenue of $1,591.0M for the twelve months ended 1 August 2026 (FY2026 $1,335.1M minus Q1 FY2026 $223.1M plus Q1 FY2027 $479.0M), not FY2026's $1,335.1M. Credo reported Q1 FY2027 on 1 September 2026 and reiterated FY2027 guidance of more than 85% revenue growth. A matched-basis forward check (comp forward EV/Revenue midpoint 9.51x on at least $2,470M of FY2027 revenue) gives $124.64 per share, a -23% gap, pointing the same way.
• Rule B2: both SBUs use the EV/Revenue fallback because Credo discloses no segment-level operating income or EBITDA. Peer multiples and the company figure are both GAAP and both LTM. The prior valuation's flag that Credo's 'GAAP operating margin was reported around 49.6%' is withdrawn: the disclosed Q1 FY2027 GAAP operating margin is 25.2% and 49.6% was a non-GAAP figure. On the matched GAAP basis Credo's margin is not high relative to Amphenol's 32.62% EBITDA margin, so no upward adjustment is warranted.
• Revenue split is estimated, not disclosed - Credo reports one consolidated product revenue line. Approximately 72.7% AEC ($1,157M), 25.3% optical ($402M), 2.0% SerDes IP Licensing ($32M) of LTM revenue. This is the least-evidenced input in the file and is carried into the band rather than resolved.
• Scope ambiguity new since the prior run: Q1 FY2027 guidance splits the optical portfolio into optical DSPs, silicon photonics PICs and ZeroFlap Optics, each above $100M. Read strictly, the 'Optical DSPs, Retimers & SerDes Chiplets' SBU may be nearer one third of the portfolio. The conservative case uses the narrow reading (about $134M), the central and generous cases use Stage 8's broad reading; this accounts for roughly $3.2B of the band's width.
• Inherited Stage 4 borderline call: if optical DSP market growth resolves nearer the 9.8% CAGR estimate than the 17% figure used, this SBU is a Dog rather than a Question Mark. That reading is built into the conservative end of the band and points the same way as the AVOID, not against it.
• The share-count convention is decision-relevant at this gap and is disclosed rather than resolved: 194.4M GAAP diluted weighted-average shares (Q1 FY2027) give $145.47 and AVOID; the 187.95M basic count reported by data providers gives $150.46 and a -7.2% gap, which is a HOLD. This is the clearest single reason conviction is Low.
• Price sits inside the fair-value band ($57.15 to $211.32), so per the Decision Framework the method does not resolve this name and conviction is capped at Moderate; it is set below that cap at Low because the -10.3% gap clears the AVOID threshold by only 0.3 percentage points. Momentum is Still Room (34.1% of the 52-week range, -1.07% over twelve months against SOXX at +108.02%, LTM EV/Revenue compressed from 27.5x to 18.85x), so the prior run's Overrun Beyond Fundamentals downgrade no longer applies.
• Placeholders (SerDes IP Licensing at Synopsys' 8.86x EV/Revenue) are 1.0% of Total Fair Enterprise Value, well below the 20% threshold at which the whole SOTP would become a range rather than an estimate.
• This decision reflects market prices as of 11 September 2026 and should be re-run, not assumed current, before acting on it. The price moved 37% between the superseded 10 August 2026 valuation and this one, more than the entire valuation gap reported here.

Portfolio-level tensions

• Customer concentration is undisclosed by SBU: two customers at roughly 32-33% of total revenue each and the top 10 at about 90% give a small number of hyperscalers leverage over both product lines simultaneously
• Both scored jobs trace to the same upstream trigger (a hyperscaler's AI capital-spending decision), making the two SBUs more correlated than a standard Star/Question Mark pair — a Rule 6 shared-trigger flag, not a diversified two-quadrant portfolio
• AEC's Contingent (boom-dependent) durability means the SBU funding the Optical DSP push has its own uncertain horizon, which could tighten Optical DSP funding at the exact moment sustained investment is most needed
• Genuine cross-SBU opportunity: AEC's deep, already-qualified hyperscaler relationships are a distribution channel Optical DSP could use to enter design-slot conversations, but this cross-sell motion is not evidenced as being systematically pursued today
• Shared, undisclosed R&D budget and fabless supply chain (TSMC, Amkor, ASE) mean the $750M DustPhotonics acquisition and AEC's 1.6T platform/patent enforcement compete for the same internal capital pool