Credo Technology (CRDO.US), the designer of high-speed copper cables, optical modules, DSPs, and silicon photonics for AI data centers, released its latest quarterly results and forward guidance after the US market close on Tuesday. The company posted fiscal Q1 FY2027 revenue of $479.0 million, up 114.7% year-over-year, slightly ahead of the $471.7 million Wall Street consensus that had been repeatedly revised higher recently.

Credo's fresh earnings and outlook, combined with AI leader Anthropic's new multi-billion-dollar cloud computing agreement and robust semiconductor export data from South Korea, underscore that global AI compute demand remains in a large-scale expansion cycle that is far from peaking. For fiscal Q2, the company guided revenue in a range of $525 million to $535 million, with a midpoint of $530 million representing 97.8% year-over-year growth and topping analysts' expectations of $516.5 million. However, GAAP gross margin contracted 290 basis points year-over-year and 370 basis points sequentially to 64.5%, with the Q2 guidance midpoint dropping to 63.9%. The relatively soft margin picture partly explains why shares still moved lower after hours, plunging more than 10%, even as core metrics and revenue outlook came in strong.

Despite the post-earnings selloff in Credo shares, global investors' risk appetite toward the AI compute supply chain remains sturdy following the July deleveraging and de-crowding episode. As of September 1, the Philadelphia Semiconductor Index stood at 11,288.61 points, up roughly 59.4% year-to-date. The index had pulled back nearly 29% from its June peak to a July 29 low, then reclaimed the 20% technical bull-market threshold from that low by mid-August. South Korea's KOSPI index hit 6,835.80 points, up 62.2% this year, and after a 22% July slump, it rallied 22% above its July 30 trough on August 13, formally re-entering a technical bull market. In other words, July was more of a position unwinding in crowded AI trades than a fundamental topping signal in the AI capex cycle.

AI infrastructure demand is increasingly converting from hyperscaler budget intentions into multi-year compute capacity lock-ups. Market research firm TrendForce projects major cloud providers' capital expenditures to grow 98% in 2026 and another 50% in 2027, with DRAM and NAND combined rising from 47% of their capex in 2026 to 68% in 2027. Server DRAM contract prices are expected to accumulate roughly 270% gains in 2026, enterprise SSDs around 235%, and HBM contract prices could still climb 70% to 140% in 2027. According to media reports, Anthropic signed a cloud computing agreement worth approximately $35 billion with Nvidia-backed Lambda, covering roughly 350 megawatts of capacity in Texas; earlier, it also inked a six-year, $45 billion deal with Nscale for about 460 megawatts of Vera Rubin compute lease arrangements. South Korea's August exports surged 68.7% year-over-year to $98.26 billion, marking a 15th straight month of growth and beating the 62.6% estimate, with chip exports reportedly hitting a record $46.65 billion, roughly triple the year-ago level.

Revenue nearly doubled, yet shares initially "disconnected": Credo's copper cable and AI interconnect bull case is hitting a margin speed bump. Credo's core revenue generation primarily comes from AECs (active electrical cables) and high-speed optical interconnect chips inside data centers, with Nvidia being one of its largest customers—though it would be inaccurate to characterize the company simply as a "copper cable supplier inside Nvidia's NVL racks." Credo's SEC filings describe AEC applications focused on server-to-top-of-rack (ToR) high-speed switching, leaf-spine-router connections, and "GPU to network switch" links. More precisely, the physical cables typically connect NIC, DPU/SuperNIC ports on GPU servers or accelerator trays to ToR or back-end switch ports, rather than directly into bare GPU chips.

Credo's longest 7-meter ZeroFlap AEC enables high-speed connections between GPUs and network switches as well as servers and ToR switches inside liquid-cooled AI racks. The company recently showcased Nvidia's Rubin-architecture NVL144 rack product and next-gen AI GPU cluster—the Rubin Ultra NVL576 architecture—fully wired with 1.6T ZeroFlap AECs. Beyond AECs, Credo also offers PAM4 DSPs for 400G/800G/1.6T optical modules, ZeroFlap optical transceiver product lines with link telemetry, data center silicon photonics/PICs, PCIe 6.0 and CXL 3.x retimers, PCIe AECs, OmniConnect and Weaver memory fan-out chips for memory expansion and scale-up/scale-out, SerDes chiplets/IP, microLED active cables up to roughly 30 meters, and the PILOT link diagnostics platform.

Latest earnings data shows Credo's fiscal Q1 revenue rose 114.7% year-over-year to $479.0 million, beating the $471.7 million analyst estimate by 1.5%; adjusted EPS surged 130.8% to $1.20, exceeding the $1.17 consensus by 2.6%. Fiscal Q2 revenue guidance came in at $525–535 million, with a midpoint of $530 million representing 97.8% year-over-year growth and 2.6% above the $516.5 million consensus. But GAAP gross margin contracted 290 basis points year-over-year and 370 basis points sequentially to 64.5%, with the Q2 guidance midpoint falling to 63.9%. So the after-hours share decline is not about weakening AI demand, but rather investors now demanding "high growth and margin quality" to materialize simultaneously.

The selloff likely reflects profitability pressure underneath rapid sales growth. GAAP gross margin fell to 64.5% from 68.2% in the prior quarter and 67.4% in the year-ago period. GAAP operating expenses more than doubled year-over-year to $188.4 million from $89.6 million. Credo remains a major beneficiary of AI data center spending, as these systems depend on high-speed connections among processors, memory, and networking gear. The company continues to lean into R&D investment, with R&D expenses climbing to $114.5 million from $52.4 million a year ago. Sales, general, and administrative expenses rose to $73.9 million from $37.2 million. GAAP operating profit rose to $120.7 million from $60.7 million in the year-ago quarter, though below the prior quarter's $155.8 million. Operating margin narrowed to 25.2% from 35.7% in the prior quarter and 27.2% a year earlier. Net income increased to $129.4 million, or $0.67 per share, from $63.4 million, or $0.34 per share, a year ago. Adjusted net income grew 140% year-over-year to $236.3 million. Adjusted operating profit came in at $230.6 million, up from $96.2 million in the prior-year period.

Credo's fiscal Q1 non-GAAP operating margin was 48.2%, up 510 basis points year-over-year but down 140 basis points sequentially; non-GAAP net margin was 49.3%, up 520 basis points year-over-year but down 260 basis points quarter-over-quarter. These margin figures collectively show that core adjusted profitability remains solid, but the marginal expansion is no longer continuously widening as in prior quarters. The company expects next quarter's GAAP gross margin between 62.9% and 64.9%, with a midpoint of 63.9% trailing the first quarter's 64.5%. GAAP operating expenses are expected to increase to $199 million to $204 million. Adjusted gross margin guidance is 67% to 69%, versus 68% in fiscal Q1; adjusted operating expenses are expected at $100 million to $105 million, above the prior quarter's $95.2 million.

The decade-defining "copper-optical synergy" growth story belongs to Credo. The expansion of Nvidia AI GPU clusters and Google TPU compute clusters serves as a key underlying demand driver for Credo. Beyond AEC copper cables, Credo supplies 400G, 800G, and 1.6T ZeroFlap optical modules, optical DSPs and silicon photonics ICs, OmniConnect memory and chip-to-chip interconnects, Ethernet and PCIe retimers, SerDes chiplets and licensing, microLED active optical cables, and the PILOT software platform for link health monitoring and failure prevention. A more accurate positioning for Credo, therefore, is a "chip-to-cluster, millimeter-to-kilometer copper-optical converged AI interconnect platform," rather than a pure copper cable vendor.

As GPU racks add copper connectivity demand and cluster expansion amplifies optical connectivity demand, Credo is positioned to capture incremental growth from both ends. Its value proposition extends beyond high-speed copper cables inside NVL racks to a complete product matrix spanning server networking, intra-data-center optical interconnects, PCIe, and memory expansion. CEO Bill Brennan stated in the latest earnings release: "Our product portfolio now spans connections from millimeters to kilometers, and spans both optical and copper connectivity solutions." The industry logic for intra-data-center high-speed interconnect is not "optics replacing copper," but rather a division of labor by transmission distance, power consumption, and network hierarchy: AECs suit intra-rack and short-reach inter-rack connections with low power, low cost, and high reliability; pluggable optical modules and optical DSPs handle longer-distance switch-to-switch, leaf-spine, and cross-rack connections; microLED active cables aim to cover the mid-range gap between copper and traditional optical modules. As GPU counts rise, short-reach copper connection ports grow in tandem; as clusters scale out, long-reach optical connections grow even faster.

Therefore, Citi's emphasis on AI capex expansion and Morgan Stanley's latest view that "AI data center copper and optics coexist long-term with a phased migration toward optical interconnects" jointly support Credo capturing both current AEC cash flows and future optical interconnect system upside, rather than betting on a single transmission medium. "Copper-optical synergistic explosive growth" is likely to become one of Credo's core valuation re-rating narratives over the next several years. BNP Paribas' $275 price target and the logic of "a potential growth market of at least $10 billion" essentially reflect Credo's evolution from an AEC-focused strength into an interconnect platform jointly driven by AECs, ZeroFlap optical modules, optical DSPs, ALCs, and OmniConnect.

Credo shares have climbed more than 43% this year, significantly outperforming the S&P 500. But with the stock still falling after solid earnings and revenue guidance, investment decisions can no longer ignore gross margins, customer concentration, and new product delivery in isolation. In fiscal Q1 FY2027, revenue reached $479 million, up 114.7%; adjusted EPS was $1.20, above the $1.17 consensus; fiscal Q2 revenue guidance was $525–535 million. The share price decline post-announcement—at one point plunging more than 10% to around $186—sufficiently reflects that market capital is no longer satisfied with fast revenue growth and quarterly guidance increases alone, but is beginning to scrutinize growth quality, product mix, and profitability. As Nvidia once again posted blowout earnings and exceptionally strong guidance, AI compute trading momentum is likely to expand beyond Nvidia GPU clusters into the entire AI compute supply chain: HBM/DRAM/NAND, CoWoS/3D advanced packaging, data center CPUs, high-performance networking infrastructure, high-speed copper cables/optical interconnects, and data center power infrastructure. Citi, Goldman Sachs, and Morgan Stanley all hold bullish views on Nvidia's next-gen Vera Rubin architecture demand, GPU cluster visibility into 2027 and beyond, and AI infrastructure expansion trends. Their consensus thesis centers on persistently strong AI compute demand, Vera Rubin ramp-up, and Nvidia's strengthening integrated hardware-software platform advantage. This also explains why 19 Wall Street analysts assign Credo an average price target of approximately $268.39, with the Street-high target reaching as high as $350.