Credo Technology: Optical Business Emerges as Second Growth Engine While New Inference Products Point to Multibillion-Dollar TAM Expansion
Q1 Fiscal 2027 earnings call, September 1, 2026 — Credo posts record revenue and reaffirms aggressive optical ramp targets
Credo Technology delivered its seventh consecutive quarter of triple-digit year-over-year revenue growth, with Q1 fiscal 2027 revenue reaching $479 million, up 10% sequentially and 115% year-over-year, beating the high end of guidance. Non-GAAP gross margin held at 68%, and non-GAAP net income more than doubled year-over-year to $236.3 million, a 49.3% net margin. Management guided Q2 revenue to a range of $525 million to $535 million and reiterated that fiscal 2027 will bring more than 85% full-year revenue growth, driven by an inflection in the second half tied to optical products.
OmniConnect: a new multibillion-dollar bet on inference memory bottlenecks
The most significant new disclosure on the call centered on OmniConnect, a SerDes-and-gearbox architecture aimed squarely at the memory fanout problem constraining inference performance. CEO Bill Brennan described the offering as solving a "universal issue" where memory bandwidth and capacity limits, rather than compute, cap inference throughput. The first product, a gearbox called Weaver, bridges an embedded high-speed SerDes reaching up to 10 inches to an LPDDR5 interface, with an LPDDR6 successor already planned so that customers can upgrade memory generations "without another tape-out." Credo named Positron as its first customer, highlighting a roadmap toward 2-terabyte memory configurations that Brennan said could make frontier models fit entirely in memory, calling this "redefining what is competitive" in inference hardware. Management pegged OmniConnect as a multibillion-dollar annual opportunity, with revenue beginning in fiscal 2028 — a new growth vector layered on top of the optical and AEC franchises investors already track.
Active LED Cables add a third pluggable transceiver category
Credo also detailed progress on Active LED Cables, or ALCs, which use micro-emitter technology to extend copper-like reliability and power efficiency out to 30 meters of reach — well beyond the roughly 7-meter ceiling of AECs. Brennan positioned ALCs as "the third leg of the stool" after AECs and ZeroFlap Optics, both product categories Credo itself created. Initial revenue is targeted for fiscal 2028, with a demonstration planned at OCP in October. Longer term, management is angling to apply the same micro-emitter technology to scale-up architectures, which Brennan called "as big of an opportunity" as the pluggable ALC market itself, given the reliability and serviceability problems that have so far constrained scale-up interconnect adoption.
Optical guidance reaffirmed at $600 million-plus, with three distinct legs
Credo reiterated that fiscal 2027 optical revenue will exceed $600 million, with ZeroFlap Optics, silicon photonics PICs, and optical DSPs each contributing more than $100 million. CFO Dan Fleming's framing implies the remaining balance still comes largely from DSPs' existing base, but the structural point for investors is that Credo now has three separately monetizable optical product lines rather than a single transceiver SKU. Management disclosed that the DustPhotonics acquisition has already produced two major silicon photonics PIC design wins for next-generation ramps expected in fiscal 2028, though Brennan clarified these initial wins do not include DSP integration — leaving upside from future co-marketed DSP-PIC bundles. He credited the acquired Dust team's traction to leading-edge technical execution, noting it was "amazing that a small team like that had such great traction."
Joining the Open CPX consortium signals a scale-up optics push
Credo confirmed it has joined the Open CPX MSA consortium, a near-package optics standard aimed at the 10x density increase required for scale-up networks as architectures move beyond pluggable form factors. Brennan described Credo's strategy as "agnostic," supplying silicon photonics PICs initially but planning to offer complete optical engines as the company adds functionality. Confirmed NPO design wins are expected to begin ramping in fiscal 2028, positioning Credo's addressable market to extend well beyond the front-end scale-out network into the emerging scale-up segment — a market Brennan sized qualitatively as "tens of billions of dollars" of combined opportunity once ALCs and NPO are included alongside existing AEC and optical DSP franchises.
PILOT telemetry platform positioned as a data moat
In response to an analyst question about whether accumulated telemetry data constitutes a competitive advantage, Brennan gave one of the call's more detailed technical answers. Credo's PILOT software captures real-time eye height, signal-to-noise ratio, and post-histogram data across all six directional links between XPUs and switches — a granularity he said is "completely different" from telemetry approaches discussed elsewhere in the industry. He described the system as adding "a check engine light" between the industry's existing binary green-light/red-light failure indicators, enabling customers to proactively pull a failing GPU from a cluster before a hard failure occurs. Brennan also disclosed that customer requests pushed Credo to develop sensing for latent ESD damage on transceivers and even dust contamination on fiber connections that can cause multipath interference — capabilities he framed as compounding advantages as the installed base and data set grow. He quantified the financial stakes bluntly, noting that industry-reported contracts imply "hundreds of millions or even $1 billion" in value tied to shortening cluster bring-up from six-to-eight weeks to five-to-six days, with sustained uptime above 99% as the parallel goal.
Customer concentration remains high but is broadening by product
Fleming disclosed that Credo's top customer represented 33% of Q1 revenue, the second 28%, followed by 13% and roughly 10% for a fourth customer that rotated in from a different name the prior quarter. Management continues to expect three to four customers above the 10% threshold going forward. Fleming noted that concentration is becoming less about single-product exposure, stating there is "strengthening and broadening across the board" as hyperscalers adopt AECs, optical DSPs, and ZeroFlap products simultaneously rather than a single interconnect type.
Balance sheet reflects the DustPhotonics deal
Cash and equivalents fell to $764.3 million from a $679 million sequential outlay tied to the DustPhotonics acquisition, while inventory rose $62.2 million sequentially to $313.1 million as the company builds supply ahead of the anticipated second-half optical ramp. Operating expenses of $95.2 million came in above the high end of guidance on elevated R&D spending, up 16% sequentially, and management guided fiscal 2027 opex growth of approximately 55% year-over-year — comfortably below the targeted 85%-plus revenue growth rate, implying continued operating leverage with non-GAAP net margin expected to stay near 50% for the year.
Market sizing underpins the long-term growth narrative
Brennan cited third-party forecasts showing the optical transceiver pluggable market growing from 60 million units in calendar 2026 to 175 million units by 2030, a figure he used to frame Credo's diversification strategy across copper AECs, optical DSPs, silicon photonics PICs, ZeroFlap Optics, and now ALCs and OmniConnect as necessary rather than opportunistic. He was candid that AEC growth, while still expected to continue, will decelerate relative to prior years given the law of large numbers — AEC revenue more than doubled from fiscal 2024 to 2025 and more than tripled from 2025 to 2026 — with optical products growing faster precisely because they are scaling from a smaller base.