Credo Technology Sees $1 Billion Quarterly Run Rate by Early Fiscal 2028 as Optical Ramp Accelerates Growth to 30% Sequential by Year-End
Goldman Sachs Communacopia + Technology Conference, September 10, 2026
Credo Technology Group's management used its appearance at the Goldman Sachs Communacopia + Technology Conference to lay out one of the more aggressive growth trajectories in the semiconductor connectivity space, with CFO Dan Fleming detailing a sequential growth path that accelerates sharply through the back half of fiscal 2027. Having already guided to 11% sequential growth into the second quarter, Fleming told analyst Jim Schneider that the company expects roughly 20% sequential growth into the third quarter and 30% into the fourth. "That sets you up where a $1 billion quarter is within our sights within early fiscal '28," Fleming said, a disclosure that puts a concrete timestamp on what has been a blistering scale-up since the company's IPO, when revenue sat below $200 million.
The acceleration comes just a week after Credo raised its fiscal 2027 growth guidance to 85%, while maintaining its target of $600 million in optics revenue for the year. CEO Bill Brennan framed the current base of more than $1.3 billion entering fiscal 2027 as still early innings, noting the company has already strung together six consecutive quarters of 20%-plus sequential growth.
ZeroFlap Optics: The xAI Origin Story Behind Credo's Reliability Pitch
The most detailed disclosure of the session concerned the origin and mechanics of Credo's ZeroFlap optical product line, which Brennan traced back to early work with xAI. According to Brennan, xAI was struggling with "link flaps" — connections that don't hard-fail but disconnect and reconnect, corrupting data across a cluster and forcing a full reset — as it built out a liquid-cooled data center. xAI asked Credo to extend its active electrical cable family from 3 meters to 7 meters so it could cover every connection in the row with copper rather than optics, enabling what Brennan called a "ZeroFlap cluster." The payoff was significant: Brennan said customers went from bringing clusters up in six to eight weeks to five days, and pushed uptime to 99.9%, a difference he quantified as worth roughly $1 billion in revenue per month of compute.
Oracle subsequently approached Credo with a related but distinct problem: its NIC-to-first-switch connections exceeded 7 meters and couldn't be redesigned with copper. That led Credo to build a custom DSP capable of continuous, real-time telemetry across all six sub-links between a GPU and a switch, measuring eye height, SNR, post-FEC bit error rates and histograms to create what Brennan described as "a yellow, almost check-engine light for the link" rather than the industry's traditional binary green-or-red status. The system also identifies dust-induced multipath interference on fiber plants and latent ESD damage before failure occurs. "We've accomplished the goal of being able to identify instabilities and mitigate proactively," Brennan said, adding that customer conversations around the product are expanding and the ramp this fiscal year is just the beginning.
Microemitter Cables (ALCs) Head Toward Production in Fiscal 2028
Brennan also provided fresh specifics on Credo's active limitless cables, its microemitter-based technology that has been in development for three years. The pitch is combining copper-like reliability and power efficiency with reach extended to 30 meters, up from the 7-meter ceiling of active electrical cables, while shrinking cable volume by 75%. Credo plans to demonstrate the technology at OCP next month, with qualification to follow and production targeted for fiscal 2028. Brennan said customer interest is high given that Credo intends to subject the new cables to the same qualification rigor — running live traffic across customer switches and NICs to intentionally break links and then harden them through firmware — that underpins its existing AEC and ZeroFlap Optics businesses.
Production Ramp: Hundreds of Thousands of Optical Units Monthly, Doubling Into Fiscal 2028
On the manufacturing side, Brennan reiterated that Credo expects to be producing hundreds of thousands of optical units per month by the end of the current fiscal year, with that rate doubling and then tripling by the end of fiscal 2028, and further expansion expected in fiscal 2029. Combined with the accelerating sequential growth Fleming outlined, the disclosures suggest optics is shifting from a nascent contributor to a primary growth engine faster than the market may have modeled.
Margins Hold at 68% Even as Optics Mix Shifts
Fleming noted that fiscal 2027 gross margin guidance of roughly 68% was reiterated without pushback for the first time in the company's five years as a public entity. "This was the first quarter where I didn't get a gross margin question on the earnings call because we were right at 68%, which is what we had alluded to," he said. Management attributed margin durability to vertical integration — Credo builds its own DSPs and silicon photonics PICs — and to premium pricing on differentiated features rather than commodity 1.6T transceiver competition. Brennan argued the combination of favorable ASPs and controlled cost of goods sold should widen gross margin as the optics business scales, rather than dilute it.
Operating Leverage: Revenue Growing 50% Faster Than OpEx
Fleming addressed the company's historical 2:1 revenue-to-OpEx growth ratio, noting that in fiscal 2027, with OpEx guided up 55% amid heavy optical investment, top-line growth is running roughly 50% faster than operating expense growth. He said the ratio could return closer to 2:1 as growth accelerates further into fiscal 2028, implying operating leverage remains intact even during a period of elevated R&D spend tied to the optics buildout.
Competitive Moat and M&A Appetite
On competition, Brennan downplayed the threat from new entrants in the AEC market, arguing Credo's differentiation comes from owning the full stack — SerDes, IC, system design, firmware and supply chain — which he called the company's "biggest competitive moat" and the reason it has never failed a customer qualification. He was more direct about optical transceivers, pushing back on the perception that the segment is commoditized: Credo's products are not targeting IEEE-standard 1.6T commodity transceivers but rather reliability-differentiated features that command premium ASPs.
On M&A, Brennan characterized all three recent acquisitions — a Canadian microLED team, CoMira for protocol and security, and DustPhotonics for silicon photonics PIC technology — as successful integrations of engineering talent rather than opportunistic deals. He signaled openness to further acquisitions within connectivity adjacencies but was explicit that any move outside core connectivity would need to support a differentiated system-level offering. "We don't draw bright lines around what we do and what we don't do," he said, drawing a loose comparison to Nvidia's move up the stack from semiconductors into systems.