They guided gross margin down 360bps and called it conservative while raising revenue guidance for the fifth straight quarter.
Thesis: The street is pricing Credo as an AEC-only company facing CPO displacement. Three evidence points say otherwise: ZF Optics was pulled in six months by customer demand with production already shipped, 1.6T transition carries ASP uplift across the entire portfolio while volume grows, and supply is locked across 12/7/5/3nm ahead of a ramp competitors cannot match. The GM guide to 64-66% against a stated 63-65% long-term range is sandbagging consistent with five consecutive beats. Concentration at 88% is real but those three customers are all growing double-digit YoY in the current buildout.
Verdict: LONG — Conviction: HIGH
Catalyst: Fiscal 2027 Q1 when ZF Optics production revenue prints, proving the optics expansion is cash flow. Watch for qualification customers converting beyond the four in pipeline and ASP commentary on 1.6T across AECs, ICs, and optics.
Key Risk: One hyperscaler pulls back CapEx and 39% of revenue evaporates. Top-3 concentration at 88% with fourth customer missing the prior 10% commitment means diversification is running opposite to the mix.
The Tell: CFO walked back a prior commitment unprompted: 'while they are still in line with our expectations from 90 days ago, they may not be a 10% customer for the full year, if that makes sense.' The fourth hyperscaler 10% target from last quarter is dead, buried under the largest customer's outperformance. Diversification narrative is running opposite to actual mix.
Friction Level: HIGH_FRICTION — Bull sees a mispriced multi-product growth engine with supply locked and ASP uplift at 1.6T. Bear sees fragile structure where 88% concentration and a growth cliff from +52% to +5% QoQ override strong execution.
2026Q3 · Mar 2, 2026
4 of 5 readings agreed4 LONG · 1 HOLDRead the full call
88% Concentration, 3-Layer Growth Story, and a GM Guide Nobody Believes
The street is pricing Credo as an AEC-only company facing CPO displacement. Three evidence points say otherwise: ZF Optics was pulled in six months by customer demand with production already shipped, 1.6T transition carries ASP uplift across the entire portfolio while volume grows, and supply is locked across 12/7/5/3nm ahead of a ramp competitors cannot match. The GM guide to 64-66% against a stated 63-65% long-term range is sandbagging consistent with five consecutive beats. Concentration at 88% is real but those three customers are all growing double-digit YoY in the current buildout.
CFO walked back a prior commitment unprompted: 'while they are still in line with our expectations from 90 days ago, they may not be a 10% customer for the full year, if that makes sense.' The fourth hyperscaler 10% target from last quarter is dead, buried under the largest customer's outperformance. Diversification narrative is running opposite to actual mix.
Fifth consecutive beat. Revenue at high end of revised guide, GM above high end, OpEx guided flat despite Comira close. CFO explicitly calls the Q4 GM guide 'a very conservative forecast' and reiterates the 63-65% long-term range while guiding 64-66%.
Management states ASP uplift across the entire portfolio at the 1.6T transition: 'There is going to be an uplift from 800G to 1.6T across the board, across the entire portfolio.' Volume growth plus price lift simultaneously.
Every hostile question met with specifics. CPO: NVIDIA says use copper where you can, forecasters show low single-digit CPO switching share. Concentration: named exact percentages. Supply: locked across 12/7/5/3nm plus packaging commitments.
Management says 'making progress in diversifying' customer base while the fourth hyperscaler commitment walks back: CFO admits 'they may not be a 10% customer for the full year' because largest customer grew so much. Concentration tightened this quarter.
CFO attributes 360bps GM guide decline to 'product mix' without specifying which mix. 68.6% was above the stated 63-65% long-term band. The guide lands squarely inside that band, confirming peak was unsustainable.
Inventory up $57.8M or 38.7% QoQ to $208M ahead of a quarter guided to grow only $18-28M in absolute revenue. Working capital absorbed ~$43M, explaining the gap between $209M net income and $166M operating cash flow.
AECs are de facto standard for intra-rack and rack-to-rack up to 7 meters, displacing laser optics on reliability and power. Four domestic hyperscalers delivering double-digit YoY growth plus fifth hyperscaler win and neocloud demand accelerating.
+75
CONFIDENT"Brennan gives specific supply commitments across 4 nodes, names production units shipped to TensorWave, and volunteers 'strong expectation' of >4 ZF Optics ramps beyond guidance. CFO calls GM guide 'very conservative forecast' in response to analyst pressure."
Full transcript with specific customer percentages, inventory data, node-level supply detail, and named qualification accounts. CFO gave specific FY27 revenue math ($1.3B to ~$2B). Comira price undisclosed.
Fiscal 2027 Q1 when ZF Optics production revenue prints, proving the optics expansion is cash flow. Watch for qualification customers converting beyond the four in pipeline and ASP commentary on 1.6T across AECs, ICs, and optics.
One hyperscaler pulls back CapEx and 39% of revenue evaporates. Top-3 concentration at 88% with fourth customer missing the prior 10% commitment means diversification is running opposite to the mix.
Revenue $407M up 52% QoQ, 200% YoY. GM 68.6% above guide. Operating margin 49.6% with OpEx mid-30s vs revenue 50%+. Real operating leverage. Three-layer revenue stack: AECs growing, ICs growing, ZF Optics ramping into FY27 after being pulled in six months by customer demand with production already shipped to TensorWave. Four customers in qualification, Brennan expects >4 ramps in FY2027. Supply locked across 12/7/5/3nm with packaging commitments. ASP uplift across entire portfolio at 1.6T transition. FY2027 guide implies ~$2B revenue, >50% YoY. Cash $1.3B after ATM. FCF $140M confirms earnings quality. Street pricing AEC-only company with CPO overhang. Wrong on both counts.
Top-3 concentration at 88% with 39% in one customer. Diversification narrative inverted: CFO admits fourth hyperscaler may miss the 10% full-year target because the largest grew too much. GM guided down 360bps midpoint to 64-66%, landing inside the stated 63-65% long-term band. 68.6% was peak. CFO blames 'product mix' without specifics. Growth cliff from +52% to +5% QoQ with no management acknowledgment of the deceleration. Inventory built 38.7% into a decelerating quarter. Three new products (ZF Optics, ALCs, OmniConnect) carry FY27-28 timelines and contribute nothing today. Comira closed same day with no disclosed price.
Extreme hyperscaler dependency at 88% top-3
39% in one customer, 32% in second, 17% in third. Single hyperscaler CapEx pullback removes a third of revenue. CFO confirms diversification running opposite to actual mix.
Mix Shift Spin on 360bps GM decline
CFO calls 64-66% guide 'very conservative' but it lands inside the stated 63-65% long-term range. 68.6% was above sustainable level. 'Product mix' not specified.
Inventory Divergence
Inventory +38.7% QoQ to $208M ahead of a quarter guided to grow only $18-28M in absolute revenue. Working capital absorbed $43M.
Growth deceleration from +52% to +5% QoQ
Sequential growth cliff not acknowledged by management. FY2027 still implies >50% YoY but the QoQ trajectory flattens materially.
AI Label Without Substance
Three 'multibillion-dollar TAM expansions' (ZF Optics, ALCs, OmniConnect) have shipped no meaningful volume. Timelines are FY2027-2028.
LONG. The asymmetry favors the bull. Street models AEC-only with CPO overhang. Evidence shows ZF Optics pulled in six months by customer pull with production already shipped, ASP uplift structural at 1.6T, supply locked while competitors fight over memory and lasers. GM guide is sandbagging consistent with five straight beats. Concentration risk is real but near-term low while all hyperscalers are growing double-digit in a multi-year buildout. Size it. Add on ZF Optics revenue print.