Credo's 8% rally on a 29K-payroll miss is the AI capex trade's loudest 'no-landing' vote

Credo Technology jumped 8.1% in early Friday trading after the Bureau of Labor Statistics released its September employment report at 8:30 a.m. ET, and by midday the stock was changing hands around $225 versus a $210.17 close the prior session. The 29,000-payroll print missed the 87,000–89,000 consensus by roughly 60,000 jobs; the unemployment rate climbed a tenth to 4.2%; and combined revisions shaved another 60,000 off July and August. On a day a weaker-labor print would normally drag cyclicals and tech together, the AI-infrastructure complex rallied instead. The iShares Semiconductor ETF (SOXX) gained 3.2%, Nvidia added 2.8%, and Marvell Technology rose 2.9% — but Credo led the tape.

TD Cowen's Sean O'Loughlin reiterated a Buy with a $300 price target on the same morning, implying roughly 43% upside from the prior close. Of 19 covering analysts, 16 are at Strong Buy, 1 Moderate Buy, and 2 Hold; not a single Sell rating on the street. The setup is unusual: an 8% intraday move on the same morning that the broader tape was digesting the smallest payroll gain outside of a recession in years.

The plumbing runs through the Fed. On September 16, 2026 the FOMC raised the target range by 25 basis points to 3.75%–4.00% — its first hike in more than three years — and the dot plot pointed to one more move before year-end. That shock sat on top of a 10-year Treasury yield already near a 24-year high, and the AI complex had been compressing on the resulting duration risk. Friday's 29K print broke the math: by 10 a.m. ET the CME FedWatch tool showed the implied probability of another October hike collapsing to 20.5% from 64.2% the week before, a 43.7-point reversal. The market is now pricing that the Fed will hold — or even cut — at the October meeting.

Wall Street's traditional 'no-landing' trade — the economy never slows enough for the Fed to cut — has been turned on its head. The new read is that AI capex will not 'land' even if the labor market does. Goldman Sachs Research now expects global AI-related investment to exceed $1 trillion in 2026, with US hyperscaler capex running at roughly $800 billion and US AI investment hitting 1.8% of GDP this year before climbing to 2.5% in 2027. AI capex has moved from 33% of hyperscaler cash flow from operations in 2023 to an estimated 93% in 2026. The trajectory through 2028 looks more like a prior general-purpose technology buildout than a bubble late-cycle.

The Philadelphia Semiconductor Index is still up roughly 60% in 2026 even after a sharp mid-September scare, when Reuters reported investors were 'punishing the picks-and-shovels layer harder than the hyperscalers' on fears that AI capex could decelerate. Friday's print is the offsetting move: a soft macro is no longer a capex de-rating catalyst because it now implies easier money without meaningfully denting hyperscaler spending plans. Credo's 8% jump is the most concentrated expression of that view — a stock whose revenue is levered to the one capex line the hyperscalers cannot pull back from in a soft quarter: the connectivity fabric inside each AI cluster.

Credo Technology reported Q1 FY2027 revenue of $479 million on August 1, 2026 — up 114.7% year over year — and the latest 10-Q puts trailing-twelve-month revenue at $1.591 billion with a 67.1% gross margin. The company's central product is the Active Electrical Cable (AEC), a copper cable with embedded SerDes retimers that extends reach beyond passive copper at sub-optical cost. Independent estimates put Credo's share of the AEC market at 73–88%, with William Blair citing 85% in 2025. The AEC category is on pace to reach $4 billion by 2028, and at 200G per lane passive direct-attach copper only reaches about one meter while AECs stretch to roughly three meters — the engineering reason Credo's product has been the rack-scale standard inside AI clusters at Microsoft, Amazon, Alphabet, Meta, and xAI.

The flip side is concentration. In Q1 FY2027 the four largest customers accounted for 33%, 28%, 13%, and 10% of revenue — 84% combined from just four buyers. Credo's product wins are real, but the upside is bounded by the capex plans of two or three hyperscalers. With 10-K-cited top-customer exposure of one-third of revenue, any single hyperscaler slowing its AI buildout would hit the model immediately. That risk is what makes Friday's vote so loud: the tape is saying hyperscaler capex will not wobble even if the broader labor market does.

The picks-and-shovels ecosystem around the AI buildout splits into four clean lanes. Upstream demand is set by Nvidia, whose GPUs define how many SerDes lanes each rack needs. The optical lane — where Lumentum and Coherent play — got a $2 billion equity investment each from Nvidia in March 2026, the clearest signal yet that even the GPU leader is underwriting the connectivity fabric that surrounds its silicon. Custom silicon for AI clusters — the lane that ties rack-scale compute to switch and cable silicon — is dominated by Marvell and Broadcom. And at the customer end, Microsoft, Amazon, Alphabet, and Meta underwrite the entire demand curve.

There is also a competitive ceiling built into the AEC story. Per-lane physics cap AECs at roughly three meters at 200G/lane; at 400G/lane the technology is effectively dead. That is why Credo is pre-emptively pushing its ZeroFlap optical transceiver line (production ramp in fiscal Q1 2027 across four customers), its Active LED Cables acquired via the Hyperlume deal in September 2025 (sampling in FY2027, production in FY2028), and its OmniConnect gearbox silicon from the CoMira Solutions acquisition (ramping in FY2028). Management is calling the combined TAM $10 billion by the end of the decade. The risk is timing: the company is still 100% copper inside the rack today and the optical transition begins in earnest in 2027–2029 with co-packaged optics at 3.2T.

The near-term setup is dense with catalysts. The October FOMC meeting is now the single biggest swing factor: a hold or a cut would compound Friday's duration tailwind; a hawkish surprise on inflation data would unwind part of the FedWatch reversal that powered the Credo rally. Credo's fiscal Q2 FY2027 prints in early December, with the street modeling roughly $525–535 million in revenue. The hyperscaler capex lines in Microsoft, Amazon, Alphabet, and Meta Q4 reports — typically mid-October through late November — will reset the 2027 capex envelope and either validate or break the no-landing thesis.

The structural story runs on a different clock. AEC unit demand is expected to roughly double in 2026 to ~10 million units, with 800G AECs becoming the dominant category. Beyond 2027 the constraint becomes physical: 1.6T generation pulls copper to its ~3-meter ceiling at 200G/lane, and co-packaged optics reaches commercial scale at 3.2T/6.4T between 2027 and 2029. Credo's pivot into ZeroFlap pluggables, Active LED Cables, and OmniConnect gearboxes is the explicit hedge, but those lines are still early revenue. The thesis is two-handed: bull case is that the company captures an outsized share of the $4 billion 2028 AEC TAM and extends into adjacent optics, bear case is that the optical transition compresses gross margins from the current 67% toward the 60% range during the product transition.