Optical interconnect giant Credo Technology (CRDO.US) released its latest financial results after the US market close on Tuesday. The report revealed that for the first quarter of fiscal 2027, which ended on August 1, 2026, the company's revenue surged 114.7% year-over-year to $479 million, edging past the analyst consensus of $472 million. Adjusted earnings per share climbed 130.8% year-over-year to $1.20, also surpassing the consensus estimate of $1.17. Looking ahead, Credo guided for second-quarter fiscal 2027 revenue between $525 million and $535 million, with the midpoint of $530 million exceeding the analyst forecast of $516.5 million.

However, the company's GAAP gross margin for the first quarter contracted by 290 basis points year-over-year and 370 basis points sequentially to 64.5%, with the midpoint of second-quarter guidance easing further to 63.9%. Despite the robust core metrics and revenue outlook, the comparatively softer margin figures weighed on the stock in after-hours trading. Management's commentary during the earnings call indicated that AI infrastructure remains in a period of rapid expansion. Credo is transitioning from an AEC-centric high-speed interconnect supplier to a "system-level connectivity platform" that encompasses copper interconnects, optical DSPs, silicon photonics, complete optical transceivers, and next-generation NPO/scale-up connectivity solutions. The optical business is poised to become the most significant growth engine moving forward. Here are the key takeaways from Credo's first-quarter fiscal 2027 earnings conference call.

Management Commentary

William Brennan, President, Chief Executive Officer, and Chairman, opened the call by highlighting another robust quarter. Revenue reached a record $479 million, up 10% sequentially and more than doubling year-over-year. Non-GAAP gross margin stood at 68%, with non-GAAP net income surpassing $236 million, a 140% increase year-over-year. He emphasized that Credo's growth trajectory is attainable by very few semiconductor companies, while profitability continues to expand. This growth is anchored by seven consecutive quarters of triple-digit year-over-year revenue increases. In fiscal 2027, the company anticipates continued substantial growth, with the optical business accelerating at the fastest pace.

Brennan noted that investment in AI infrastructure is growing rapidly, with expanding cluster sizes, increasing data rates, and more complex connectivity demands. Beyond mere bandwidth, reliability, power efficiency, signal integrity, telemetry, and maintainability have become paramount. He also projected that AI infrastructure will grow increasingly heterogeneous, with no single architecture, protocol, or physical medium dominating all connections. Future AI systems will integrate optical and copper interconnects based on varying transmission distances, protocols, and topologies, with customers selecting optimal technologies for their architectures. This is where Credo excels. While network reliability remains a critical metric, the ability to innovate, execute, qualify, and deploy tailored solutions for diverse customer needs is a core differentiator. The company's focus is on accelerating cluster deployment, maximizing processor utilization, and sustaining reliable operations at scale. The current product portfolio addresses connectivity needs spanning millimeter to kilometer distances, encompassing both optical and copper solutions.

Regarding Active Electrical Cables (AEC), Brennan described it as still the company's largest and growing business. Credo maintains deep relationships with five hyperscalers, with expanding engagement with NeoCloud customers. Among existing customers, AEC penetration continues to rise as deployments scale. The transition to 200Gbps per lane and 1.6T ports presents an additional growth driver. He reiterated that AEC has been Credo's flagship system-level product, with co-optimized chip, firmware, manufacturing test, and system qualification. This system-level approach has been a core differentiator since the company pioneered the category. As AI clusters scale, the value proposition remains clear: high reliability and low power consumption for short-reach connections are increasingly critical in large-scale deployments. AEC's healthy growth trajectory continues, driven by deeper penetration with existing and new customers and rising bandwidth requirements in next-generation clusters.

Brennan then discussed the optical business, which encompasses optical DSPs, silicon photonics PICs, and ZeroFlap optical transceivers. The optical DSP segment achieved record revenue in the first quarter, including deployments of 50Gbps and 100Gbps per-lane solutions. Even as the transition to 1.6T solutions begins, he believes 800G ports will enjoy a long lifecycle. Customer engagement for 1.6T DSPs is strong for both fully retimed and LRO solutions. The first 1.6T DSP revenue remains on track for later this fiscal year. In the quarter, Credo also recognized its initial silicon photonics PIC revenue following the completion of the DustPhotonics acquisition. The first design wins are for 800G and 1.6T optical transceivers, expected to ramp throughout the year. Silicon photonics PICs add another critical technology to the optical platform, enabling co-optimization of DSP and PIC with firmware, telemetry, and PILOT software. This level of integration creates opportunities to enhance reliability, power efficiency, signal integrity, and diagnostics, positioning Credo favorably as scale-up architectures evolve toward near-package optics (NPO). The company is witnessing increased customer activity around NPO for scale-up networks, with confirmed design wins expected to begin ramping in fiscal 2028. As part of the Open CPX MSA consortium, Credo will bring many advantages of the current pluggable ecosystem, including telemetry, interoperability, and maintainability, with opportunities spanning both optical components and complete system-level solutions.

Brennan highlighted the progress of the ZeroFlap Optics business, which integrates optimized optical hardware, PILOT software, and switch-level SDKs to continuously monitor link health and identify or mitigate potential instabilities. The objective is to shorten cluster bring-up time and enhance long-term network availability, both delivering significant financial and end-customer value. Production shipments have commenced, with expectations for broader customer ramps across 800G and 1.6T products during fiscal 2027, serving both hyperscalers and NeoCloud customers. With DSPs, PICs, and ZeroFlap Optics, Credo now covers a larger portion of the optical link. This fundamentally expands the company's addressable opportunity: selling components where preferred, and crucially, integrating these elements with hardware, firmware, and software to deliver complete optical transceivers with unprecedented system-level reliability. The combined momentum across DSP, PIC, and ZeroFlap Optics reinforces the company's commitment to surpassing $600 million in optical revenue for fiscal 2027.

Turning to Retimers, Brennan noted record first-quarter revenue, driven by scale-up deployments using the Screaming Eagle retimer at 100Gbps per lane, with the Blue Heron retimer beginning to contribute at 200Gbps per lane. Opportunities for the Toucan retimer are expanding with PCIe Gen 6 adoption, alongside Ethernets and UALink prospects for both Screaming Eagle and Blue Heron. Given the rapid evolution of scale-up architectures, the ability to support multiple protocols enables successful participation across these diverse frameworks.

Brennan also addressed two significant emerging growth areas. Progress continues on Active LED Cables (ALC) and OmniConnect gearbox solutions. The ALC solution leverages miniature light-emitting devices to combine the reliability and power advantages of copper connections with transmission distances up to 30 meters. Customer engagement is increasing, with plans to showcase ALC at the OCP conference in October. Initial revenue remains targeted for fiscal 2028. Simultaneously, customer interest in OmniConnect is robust. The OmniConnect SerDes and Weaver gearbox solutions address fan-out challenges arising from increasing memory bandwidth and capacity demands in next-generation AI architectures, particularly critical for inference applications. Brennan believes OmniConnect solutions could represent several thousand dollars of Credo product value per GPU, with revenue commencing in fiscal 2028.

In summary, Brennan characterized the first quarter as another strong period with robust customer engagement across all business segments. AEC continues to expand with existing and new customers and higher data rates. Both Retimer and optical DSP businesses posted record revenue, while the optical business's scope, coverage, and capabilities are broadening. He believes system-level approaches will grow increasingly important as AI networks advance to 1.6T and 3.2T solutions and scale-up architectures drive NPO adoption. AEC enabled Credo's current scale, and significant future growth remains in that segment. The key distinction today is the development of optical as another major growth engine from a larger business base and industry leadership position. Optical opportunities have expanded from DSPs and silicon photonics PICs to complete ZeroFlap Optics and NPO solutions, significantly increasing the product value opportunity as the company solves broader customer problems. AEC continues to grow, but optical is growing faster. The company remains confident in achieving its anticipated substantial growth for fiscal 2027 based on ongoing customer engagements and ramps across the entire product portfolio, grounded in the recurring theme of reliability.

Chief Financial Officer Daniel Fleming then reviewed the first-quarter results and second-quarter outlook. Revenue for the first quarter was $479 million, up 10% sequentially and above the high end of guidance, representing 115% year-over-year growth. This record quarter benefited from substantial year-over-year revenue increases from four US-based customers and marked the seventh consecutive quarter of triple-digit year-over-year revenue growth. In the fourth quarter, the top four end customers each contributed at or above 10% of revenue. Fleming noted that customer mix varies quarterly, with expectations of three to four customers exceeding 10% of revenue in the coming quarters and throughout the fiscal year, while the company continues diversifying its revenue base across hyperscalers, NeoCloud, and other customers.

The team achieved a 68% non-GAAP gross margin in the first quarter, at the midpoint of guidance. Non-GAAP operating expenses totaled $95.2 million, above the high end of guidance due to sustained R&D investments, representing a 16% sequential increase. First-quarter non-GAAP operating profit was $230.6 million, compared to $216.7 million in the fourth quarter, resulting in a 48.2% non-GAAP operating margin. Despite heavy R&D investment, the company again demonstrated significant operating leverage. Quarterly non-GAAP net income hit a record $236.3 million, up 4% sequentially from $226.7 million, and more than doubled year-over-year. Non-GAAP net margin reached 49.3%.

Operating cash flow for the first quarter was $90.2 million, down $92 million sequentially primarily due to working capital changes. Capital expenditures were $7.3 million, resulting in free cash flow of $82.9 million. The company ended the quarter with $764.3 million in cash and equivalents, down $679 million sequentially due to cash outlays for the DustPhotonics acquisition. Fleming emphasized the company's strong capital position to continue investing in growth opportunities while maintaining substantial cash reserves. Quarter-end inventory was $313.1 million, up $62.2 million sequentially.

Moving to guidance, Credo expects second-quarter fiscal 2027 revenue between $525 million and $535 million. Non-GAAP gross margin is projected between 67% and 69%, with non-GAAP operating expenses between $100 million and $105 million. Diluted weighted average share count is estimated at approximately 200 million shares. These expectations reflect the current tariff regime, which remains subject to change. As fiscal 2027 progresses, the company continues to anticipate an inflection in the second half of the year, driven by over $600 million in optical revenue, with ZeroFlap Optics, silicon photonics PICs, and optical DSPs each contributing more than $100 million. Full-year total revenue growth is projected to exceed 85% year-over-year. Non-GAAP gross margin for fiscal 2027 is expected to be broadly in line with fiscal 2026 levels. Non-GAAP operating expenses are expected to increase approximately 55% year-over-year, notably below revenue growth, as the company continues investing in R&D to support new product development and capture significant growth opportunities. Consequently, non-GAAP net margin is anticipated to approach 50%.

Q&A Session Highlights

Tore Svanberg of Stifel congratulated the company on the record quarter and inquired about the optical business structure and revenue composition of the $600 million target, particularly regarding NPO and system-level initiatives. Brennan expressed satisfaction with the expanding product portfolio. The optical DSP business performs well at the component level. The DustPhotonics team has added significant momentum, with two major customer design wins for next-generation products already secured, ramping in fiscal 2028, potentially beginning later this fiscal year. ZeroFlap Optics continues making progress with multiple hyperscaler and NeoCloud customers. Regarding the broader optical opportunity, Brennan noted Credo's membership in the CPX consortium, viewing this as an important industry development addressing scale-up density with 10x improvements. The company plans to pursue this market on a technology-neutral basis, selling components where customers require and developing system-level solutions, with more details to come. He emphasized that fiscal 2027 is merely a stepping stone for the optical business. Market forecasts indicate the optical pluggable sector alone is expected to grow from 60 million units in 2026 to 175 million units by 2030, with additional growth from scale-up applications. The company expects sustained substantial growth through 2030 across the entire solution portfolio.

Quinn Bolton of Needham asked about the significance of Credo joining the Open CPX consortium and the types of solutions the company might offer. Brennan explained that scale-up opportunities involve form-factor changes, driven by fundamental requirements for higher-density packaging compared to scale-out networks. The industry is evaluating solutions like XPO and CPX as they relate to NPO and ultimately CPO, all addressing the need for 10x density improvements. Credo maintains a technology-neutral market approach, aligning with customer requirements and directions. For all NPO solutions, including CPX, the company will initially offer silicon photonics PICs, with consideration of complete optical engines as the portfolio expands.

Tom O'Malley of Barclays questioned supply-side challenges and capacity expansion, given contract manufacturer dependencies and reports of rising foundry prices. Brennan acknowledged the dynamic market environment, emphasizing substantial investments made 18-24 months ago in anticipation of current demand. The company is increasing working capital and supply chain investments, expressing strong confidence in the ability to supply growing product volumes in the second half and future fiscal years. Demand generation has also seen successful traction in marketing and customer partnerships. He stressed the long-term strategic approach, noting that while rapid ramps receive significant attention, the broader opportunity will transform the company's overall scale.

Sean O'Loughlin of TD Cowen asked about customer concentration details. Fleming disclosed that the largest customer contributed 33% of revenue, the second 28%, followed by 13% and 10%. The top three customers were consistent with the prior quarter, though ranking order shifted. The fourth 10% customer was different from the previous quarter's fourth customer. Regarding product diversification within these accounts, Fleming noted that hyperscalers are purchasing more than just AEC, with the overall product portfolio being strengthened and expanded.

Blayne Curtis of Jefferies questioned 1.6T timing in the AEC business and design progress compared to 800G. Brennan expects the AEC portfolio to begin ramping in a similar timeframe to ZeroFlap Optics and the broader market. He provided context on the AEC market within the pluggable transceiver space, which currently lacks a separate breakdown but represents the 1-7 meter segment within the overall market projected to grow from 60 million to 175 million units over four years. The company views this as a significant long-term opportunity. In the 800G segment, copper substitution for optical transceivers is evident in the first hop from GPU to switch, driven by reliability and power considerations. Brennan expressed optimism about 1.6T AEC opportunities, noting demonstration of elegant all-AEC solutions at OFC six months prior, with connections up to 6.5 meters. Contributions are expected in the second half of fiscal 2027, with larger scale ramps in fiscal 2028.

Joseph Cardoso of JPMorgan asked about positioning for inference opportunities. Brennan discussed the OmniConnect solutions expected to contribute revenue in fiscal 2028. Memory fan-out is a general issue limiting bandwidth and total memory deployment capacity for high-performance inference. OmniConnect comprises two parts: a licensed high-speed SerDes with minimal footprint, ultra-low power, and reach up to 10 inches, addressing XPU fan-out and GPU-to-memory distance challenges; and a gearbox solution called Weaver, connecting to embedded SerDes on XPUs and serving as an LPDDR interface. The first product will feature LPDDR5, with a second LPDDR6 variant planned, providing future-proofing when memory markets transition generations without requiring additional XPU tape-outs. The first customer, Positron, is performing exceptionally well, with capability to expand memory capacity to 2TB, significantly enhancing performance for frontier models. The roadmap includes bandwidth comparable to HBM5 while expanding capacity and eliminating reliability concerns associated with co-packaging XPUs and HBM.

Vivek Arya of Bank of America asked about AEC growth assumptions relative to the optical revenue projections. Brennan explained that growth is occurring across all products, with optical solutions entering their first year of significant commercial ramps. AEC will continue growing, but from a different base, having already more than doubled from fiscal 2024 to 2025 and more than tripled from 2025 to 2026. The company sees continued expansion in AEC opportunities, though overall growth rates will naturally slow compared to the large markets optical is entering. Long-term, he expects to see a healthy balance between copper and optical solutions as the company scales.

Sebastien Naji of William Blair asked about PILOT and the competitive advantages of accumulating link-level telemetry data. Brennan explained that the PILOT software platform is integral to the ZeroFlap offering. Custom DSP design enables telemetry on every XPU-to-switch link, with six total links monitored. This provides real-time, richly detailed telemetry at the SerDes level, including eye height and SNR post-histograms. This monitoring enables detection of link integrity degradation. Existing industry mechanisms operate on a green/red light principle, whereas Credo adds a "check engine" yellow light capability that identifies issues and enables mitigation. PILOT supports various response options, from orderly GPU removal on suspect transceivers to centralized network-level approaches. The comprehensive datasets collected drive better solutions and optimization of next-generation products. Other capabilities include detecting even minimal ESD damage on transceivers, potential sources of future failure, and contamination on optical fiber infrastructure, including dust particles causing multi-path interference. PILOT's network iteration capability enables proactive issue identification and mitigation. The dual objectives are faster time-to-revenue, with cluster bring-up in 5-6 days rather than 6-8 weeks, potentially representing hundreds of millions to billions in opportunity cost per month of delay; and post-deployment uptime targeting 99% or higher. This provides a significant differentiation advantage at the cluster level.

Karl Ackerman of BNP Paribas asked about customer dynamics in the standalone DSP and PIC businesses as hyperscalers increasingly customize optical transceivers. Brennan noted that optical components are a vital part of short and long-term business, serving as the path into mainstream standard-based pluggable markets. Having DSP and PIC with system-level design enables competitive performance, power, and yield outcomes. Module customers often work directly with hyperscalers, who actively specify components for integration into modules. Hyperscalers play a significant role in both component sales to module customers and in the ZeroFlap module business. Long-term, he anticipates continued complementarity between component and module sales, with broad market needs served by components and specific high-reliability segments served by new transceiver products.

Vijay Rakesh of Mizuho asked about fiscal 2028 outlook across the four business segments. Brennan discussed the ALC position within the portfolio, noting the initial product uses MicroLED technology to deliver AEC-equivalent reliability and efficiency at extended reaches up to 30 meters. ALC represents the third differentiated pluggable transceiver solution in Credo's product strategy, following AEC and ZeroFlap Optics. The natural evolution is application to scale-up opportunities, potentially addressing reliability, availability, and maintainability challenges that have hindered market adoption. He views ALC as a multi-billion dollar opportunity, with scale-up representing an equally significant complementary opportunity. The company aims to target multiple multi-billion dollar TAMs, with current pluggable opportunities reaching substantial scale and further expanding with ALC on the fiscal 2028 timeline. Growth depends on successful execution across customers. OmniConnect similarly represents a multi-billion dollar annual opportunity. Expecting substantial growth in fiscal 2028 and beyond.

Mark Lipacis of Evercore ISI compared current optical market development with the historical approach to AEC. Brennan reflected on the AEC journey, noting that customers initially approached the company with ideas for innovations like telemetry and rack-level system enhancements. Microsoft was the first AEC customer, driven by functionality including lossless failover to backup ToR ports. This momentum built over years as data rates increased beyond DAC signal integrity capabilities and cable form factors expanded. The differentiated go-to-market approach includes rigorous certification testing beyond typical customer requirements, validating complete systems including switches and NICs at real rates. The key difference for ZeroFlap Optics in the AI era is the fundamental nature of AI networks compared to front-end networks. AI clusters have tens or hundreds of thousands of interdependent links where link flaps can significantly impact cluster performance, with GPU utilization potentially declining over 10% or approaching 20%. This creates an ideal opportunity to apply the AEC playbook to an area lacking innovation. The market is moving faster than AEC adoption due to the clear, existing pain point, with technical network teams welcoming these solutions.

Suji DeSilva of OTH Capital asked about competitive dynamics and differentiation. Brennan discussed the go-to-market strategy and vertical integration across the entire technology stack, from SerDes through DSP to PIC. This vertical integration provides advantages in system-level optimization and COGS. The company expects advantages at both COGS and ASP levels compared to standardized market solutions. Credo has become an innovation leader in optics, balancing proprietary development with open participation in standard organizations. The challenge lies in continuously innovating faster than competitors, with confidence grounded in the comprehensive technology stack that produced past success in AEC.

Christopher Rolland of Susquehanna asked about DSP and PIC integration, specifically inquiring whether the two Dust design wins include DSP integration and market economics. Brennan noted that the component market is highly competitive, limiting ability to discuss specific customers. The first two significant design wins do not include DSP, leaving upside for future co-marketing efforts. The Dust team has achieved excellent momentum with major industry customers in a short period, validating the leadership position of their technology. Long-term opportunities to combine DSP and PIC provide significant value for the customer base, representing a highly promising development for the overall portfolio.