The signal has never been stronger. The question is how far it still reaches.
Credo sells the physical layer of AI — retimed copper cables, SerDes, optical DSPs, and now its own silicon photonics. Revenue grew 114.7% last quarter and the stock is down 43% from August. One question decides it: was the AEC franchise a standard, or a head start Credo settled away in March 2026? Six analyst lenses, three scenarios, four horizons.
Log price axis — necessary here, because the cone spans $96 to $840 and the history spans $43 to $309. Gray line = Credo’s actual path from the Dec 31, 2024 close of $67.21 into today’s $175.89 (Sep 18, 2026). Closes from July 2026 onward are sourced daily closes; earlier points are anchored to dated disclosures (annual closes, dated percentage-change statements) rather than invented months — the exact sessions of the $86.49 low and $308.67 high are month-level, and the dots mark those intraday 52-week extremes against a line drawn on closes. Colored paths to the right are synthesized scenario midpoints at bear 30% / base 45% / bull 25%. The clay open circle at 2027 is the Street’s $282 twelve-month consensus (19 analysts, Sep 2, 2026).
Re-weight the scenarios
Those probabilities are a judgment call — so make them yours. Drag to set how likely the bear and bull cases are (base takes the remainder); the blended target below, the dotted line on the chart, and the prob-weighted row of every scenario card update live.
Six analyst lenses, six answers
The same quarter supports a $115 target and a $345 target. Each lens below is a synthesized expert framework — not a real person, not a real firm rating — with its own twelve-month number and the arithmetic behind it. Panel mean: $225. Street consensus: $282.
The Bandwidth Compounder
Credo has tripled revenue twice and guides to grow 85%+ again, yet trades at 24.1× forward — a PEG of 0.48, against Astera Labs at 43.9× and Marvell at 54.1× (Sep 11, 2026). Two engines fire at once: content per rack steps up as 200G lanes and 1.6T ports displace 100G, and optical — guided above $600M in FY27, roughly a quarter of revenue — is a second S-curve from near-zero share. A 47.8% non-GAAP operating margin and 38.2% ROIC say the growth is not being bought.
The Cash Auditor
The franchise economics are real: 67.1% gross margin, capex at 3.9% of revenue, $738M net cash against $26M of debt. Two cautions temper the multiple. Trailing stock compensation of $235.2M consumes more than half of $438.6M free cash flow, so honest owner-earnings sit nearer 160× than the headline 75× P/FCF. And Q1 FY27 converted only $90.2M of cash from $236.3M of non-GAAP profit as inventory swelled to $313.1M. Pay for the franchise, not the adjusted number.
The Expiring Monopoly
Credo dismantled its own legal moat. The March 2026 settlements with Amphenol, Molex and TE Connectivity handed three entrenched connector giants — TE alone at roughly $16B of revenue — an IP-clear path into retimed copper, and qualification cycles put their cables at hyperscalers through 2027. Underneath: 84% of revenue in four customers with no binding commitments, one at 57% of receivables, and zero non-AI revenue. The tell is already visible — a 1.8% revenue beat on Sep 1, 2026 versus roughly 14% earlier, inventory up $62.2M, and a 20% single-day drop.
The Qualification Clock
The barrier is not the patents — it is the six-to-nine-month hyperscaler qualification gauntlet Credo cleared first, having shipped millions of AECs before any rival shipped one. That head start is what sustains 67.1% trailing gross margin and a 67–69% Q2 guide alongside 114.7% growth. The $1.251B DustPhotonics deal (closed May 28, 2026) completes a SerDes→DSP→photonics stack that lowers cost — but Broadcom and Marvell already own that stack and roughly 70% of optical DSP share. Call it a three-to-five-year lead, not a fortress.
The Broken Neckline
The tape says one thing and the multiple says another. CRDO sits 43% below its August high, squarely on the $176 neckline of a double top, below its 50-day average, with measured-move risk toward $125 and no resistance reclaimed until roughly $212. Yet the de-rating has already happened: trailing P/E fell from about 103× in August to 62.1×. Short interest is only 3.55% of shares, so this is momentum de-grossing — Coherent, Lumentum, CoreWeave and Nebius are all off 30%+ — not a crowded short. At beta 3.23, position size matters more than the target.
The Cost of Demand
September’s collapse was a sector event, not a company event. Combined 2026 hyperscaler capex still runs $660–690B; Alphabet has raised its number three times to $175–185B; TSMC lifted 2026 capex to $60–64B on +53% August revenue — and its stock fell 4% on the news. The complex now trades on what demand costs to support, not whether it exists. The emerging constraint is power and permits: record-low 1.4% data-center vacancy and roughly $156B of stalled projects. Tariffs barely touch a Cayman-domiciled fabless supplier; Section 232 exempts data centers. Own the earnings, not the multiple.
Wall Street 12-month price targets
What the sell-side expects over the next year. Bars are sorted low to high; the dashed line is today’s $175.89. Every single target sits above it — and every one of these was published on Sep 2, 2026, the day after the quarter that knocked 20% off the stock.
Eight of the nineteen firms covering Credo, all dated Sep 2, 2026 (source: StockAnalysis ratings page, retrieved Sep 21, 2026). The distribution is 14 Strong Buy, 4 Buy, 1 Hold, zero Sell; the mean is $282.47 and the median $290, against a full range of $185 to $350. Note what that means: the Street re-published these targets the morning after a print that cost the stock a fifth of its value, and not one desk moved below the market price. Firms and targets as reported; ratings are the firms’ own, not this desk’s.
How far the signal carries
Synthesized scenario midpoints, dated from today. Returns are versus the Sep 18, 2026 close of $175.89. Every price below is an EPS estimate times an exit multiple — the arithmetic is in the drawers beneath. These are illustrative frameworks, not forecasts; real outcomes can land outside the cone.
1 Year
Sep 20272 Years
Sep 20283 Years
Sep 20295 Years
Sep 2031▸ Bull case — show the assumptions & math
▸ Base case — show the assumptions & math
▸ Bear case — show the assumptions & math
▸ Why the weights are 30 / 45 / 25 — and what the EPS ladder rests on
Revenue, capex, free cash flow & cash ($M)
Credo is fabless, so the capex bar is deliberately small — that is the model. Note what happened to the cash bar in the last group: $1.44B became $764M when DustPhotonics closed on May 28, 2026.
Revenue $192.97M → $436.78M → $1,335M → $1,591M trailing, on capex that never exceeded $62M — 3.9% of trailing revenue. That is the whole quality argument: Credo buys no fabs. Free cash flow went from $17.1M in FY2024 to $438.6M trailing, but subtract $235.2M of trailing stock compensation and true owner cash is nearer $203M — the bear’s footnote to the bull’s headline. Total debt is a rounding error at $26.2M, so the slate bar here is cash and short-term investments, not leverage (a net-cash substitution this desk uses for debt-free companies). Fiscal years end in early May. Sources: company releases and StockAnalysis financials, retrieved Sep 21, 2026.
The EPS ladder underneath every target ($)
The price targets are not conjured. Each is one of these bars times an exit multiple. Gray is reported; olive is estimate — consensus through FY2028, this desk’s base-case extension after that.
Non-GAAP diluted EPS, which is how both the company and the Street quote Credo — and the gap to GAAP is large: FY2026 was $3.46 non-GAAP against $2.51 GAAP, almost all of it stock compensation. FY2025 $0.70 and FY2026 $3.46 are reported. FY2027E $6.31 and FY2028E $9.63 are published consensus (19 and 18 analysts). FY2029E $12.40 is derived from the $5.07B FY2029 revenue consensus at FY2026-like conversion; FY2030E–FY2032E are this desk’s base-case extension at roughly 20% annual growth. The base case’s $441 five-year target is simply the last bar, $21.00, times a 21× exit multiple.
Nothing in the business is slowing
Year-over-year growth, most recent reported or guided. Read these against a stock that has lost 43% since mid-August — that disconnect is the bull case, and the bear’s reply is that every bar here is backward-looking.
Every line is up, and the only small bar is the one you want small — share count, up just 2.5% year over year despite a part-stock acquisition. Olive bars are reported results (Q1 FY2027, ended Aug 1, 2026, reported Sep 1, 2026; TTM through Aug 2026). Clay bars are company guidance or frontier lines: the FY2027 “more than 85%” revenue guide, the Q2 guide midpoint of $530M against roughly $268M a year earlier, and management’s description of AEC revenue as having “more than tripled” in fiscal 2026. AEC growth shown as +200% is that description floored, not a reported figure.
Bull vs. Bear
The whole valuation argument compresses into one disagreement: is retimed copper a standard Credo owns, or a window that closes when the connector giants finish qualifying?
▲ THE BULL CASE
- Growth is accelerating off a bigger base. Q1 FY2027 revenue $479.0M, +114.7% YoY and +9.6% sequentially, after FY2026 tripled to $1.335B. Q2 is guided to $525–535M, above the roughly $517M consensus.
- The multiple is the cheapest in the cohort. 24.1× forward and a PEG of 0.48, against Astera Labs at 43.9× and Marvell at 54.1× (Sep 11, 2026) — while growing faster than either.
- A second S-curve is already funded. Optical guided above $600M in FY2027, roughly a quarter of revenue, with silicon photonics now in-house after the $1.251B DustPhotonics close on May 28, 2026 — and Credo holds only single-digit DSP share, so the runway is share gain, not market growth.
- The economics are genuinely elite. 67.1% trailing gross margin, 47.8% FY2026 non-GAAP operating margin, 38.2% ROIC, capex at 3.9% of revenue, $738M net cash and $26M of debt.
- The qualification moat is real even without patents. Six-to-nine-month hyperscaler qualification cycles, millions of AECs shipped before any rival shipped one, and margin that has held 67–69% right through the fastest growth.
- Management is paid on the ladder, not the quarter. The June 2026 CEO grant vests only on both revenue and stock-price hurdles — the first tranche needs $2.5B trailing revenue and a $244.70 average price, the last $7.5B and $489.40.
- The Street did not blink. All nineteen covering firms republished on Sep 2, 2026, the day after the 20% drop; not one target sits below the market price.
▼ THE BEAR CASE
- Credo settled its own moat away. The March 2026 agreements with Amphenol, Molex and TE Connectivity ended the AEC patent suits and cleared three connector giants — TE alone around $16B of revenue and vastly greater manufacturing scale — into Credo’s largest business.
- Four customers, no commitments. 84% of Q1 FY2027 revenue came from four buyers (33/28/13/10), with one at 57% of receivables. Hyperscalers buy on purchase orders and dual-source as policy; a single lost socket is a third of the company.
- Zero diversification. There is no meaningful non-AI revenue. 100% of the business rides hyperscaler capex intent — intent that has already repriced, with Coherent, Lumentum, CoreWeave and Nebius all down 30%+ from their peaks.
- The beat is shrinking. Q1 FY2027 beat revenue by 1.8% versus roughly 14% surprises a year earlier, and the stock fell 20% the next day. Inventory rose $62.2M to $313.1M into a promised second-half ramp.
- Physics is moving against copper. Every lane-speed doubling shortens how far a cable can reach. Co-packaged optics moving into 1.6T switch platforms compresses the retimed-copper window — and Credo spending $1.251B on silicon photonics is its own read on that roadmap.
- Where it is not the incumbent, it is tiny. Single-digit share of optical DSP against Broadcom and Marvell’s combined ~70% — while selling DSPs to Coherent, InnoLight and Lumentum and competing with those same customers through ZeroFlap transceivers.
- The adjusted number flatters. $235.2M of trailing stock compensation is more than half of $438.6M free cash flow, and Q1 converted just $90.2M of cash from $236.3M of non-GAAP profit.
Risk map — likelihood × impact
Where each risk sits over a three-to-five-year horizon, not just how loud it is. Only one risk occupies the hot corner — likely and high-impact — and it is the one the company itself created in March 2026.
- SBC dilution
- De-rating
- Margin mix
- AEC share loss
- Channel conflict
- Goodwill risk
- Customer loss
- Capex digestion
- CPO shock
- Taiwan supply
AEC share loss post-settlement
Amphenol, Molex and TE Connectivity are IP-clear after March 2026 and compete on manufacturing scale Credo cannot match. What breaks: a qualified rival AEC at a top-three hyperscaler in 2027 and price becomes the variable.
Top-customer step-down
One buyer is 33% of Q1 revenue and 57% of receivables, on purchase orders with no binding commitment. What breaks: a single dual-sourcing decision removes roughly a third of forward revenue with a quarter’s notice.
Hyperscaler capex digestion
Credo has no non-AI revenue, so a pause is undiluted. What breaks: 2027 capex guidance plateaus rather than escalating and a 24× forward multiple on a decelerating grower becomes 15×.
Multiple compression
The whole AI-infrastructure complex is de-rating on what demand costs to support — TSMC fell on record results. What breaks: earnings grow and the stock does not, as it has since mid-August.
Gross-margin mix
Cables and transceivers carry lower margin than chips, and the CFO has said expansion “won’t always be linear.” What breaks: non-GAAP gross margin drifts below the 67–69% band and the EPS ladder resets downward.
Co-packaged optics shock
Optics moving onto the switch package collapses the distance range where retimed copper wins. What breaks: the AEC business is structurally capped years earlier than the roadmap assumes — low odds this decade, existential if early.
Taiwan supply disruption
A fabless Cayman-domiciled supplier depends entirely on Taiwanese foundry and packaging capacity already running at full allocation. What breaks: geopolitical or capacity shock and there is no second source.
Channel conflict
Credo sells optical DSPs to Coherent, InnoLight and Lumentum while shipping competing ZeroFlap transceivers. What breaks: those customers move their DSP sockets to Broadcom or Marvell in response.
DustPhotonics goodwill
The May 2026 deal put $895.8M of goodwill and $361.7M of intangibles on a balance sheet with $2.7B of equity, plus $310M of contingent consideration. What breaks: the optical ramp slips and a write-down follows.
Stock-comp dilution
Trailing SBC of $235.2M is over half of free cash flow, and the June 2026 CEO grant alone is up to 1.44M shares. What breaks: nothing suddenly — it just quietly taxes per-share value at roughly 2.5% a year.
The jargon, decoded
Hover the dotted terms above, or scan the desk’s working definitions here.
- SerDes
- Serializer/deserializer — the mixed-signal circuit that packs parallel data onto one very fast serial lane and unpacks it at the other end. Credo’s core intellectual property; everything else is built on it.
- Retimer
- A chip that receives a smeared, attenuated signal, recovers the clock, and regenerates a clean copy. It is what lets a copper cable run further than physics would otherwise allow.
- AEC
- Active electrical cable — copper with retimer chips in the connector heads. Cheaper and lower-power than optics inside a rack; useless beyond a few metres. Credo’s largest business.
- Reach
- How far a link can carry data before errors overwhelm it. Every doubling of lane speed cuts reach, which is why the copper-versus-optics boundary keeps moving inward.
- Optical DSP
- The digital signal processor inside an optical transceiver that cleans up the light-borne signal. Broadcom and Marvell hold roughly 70% of this market; Credo has single digits.
- Silicon photonics PIC
- A photonic integrated circuit — lasers and modulators built on silicon. What Credo bought with DustPhotonics, so it can make the optical engine rather than buy it.
- Co-packaged optics
- Putting the optical engine directly on the switch chip package. If it arrives early and cheaply, the distance range where retimed copper wins shrinks fast.
- Non-GAAP EPS
- Earnings per share excluding stock compensation and acquisition amortization. Credo’s FY2026 non-GAAP EPS was $3.46 against $2.51 GAAP — almost all of the gap is stock.
- Forward P/E
- Price divided by the next twelve months of expected earnings per share. 24.1× here, versus 43.9× for Astera Labs and 54.1× for Marvell on Sep 11, 2026.
- PEG
- Forward P/E divided by the expected growth rate. Below 1 is conventionally cheap; Credo screens at 0.48 — which is either a bargain or a market saying the growth is temporary.
- Exit multiple
- The P/E assumed at the end of a forecast. Multiply it by projected EPS and you have a target price. Every number in the scenario cards is built this way.
- Prob-weighted
- Each scenario’s price times its probability, summed into one expected value across bear, base and bull. The clay row in every card — and the only one the sliders move.