The AI boom has a wiring problem, and Credo Technology Group thinks it has the fix.
On October 12, 2026, the Nasdaq-listed company (CRDO) launched its 1.6T ZeroFlap Active Light Cable at the OCP Global Summit in San Jose, California. The product is built to keep thousands of AI accelerators talking to each other without dropping the conversation.
What Credo actually built
Credo’s new Active Light Cable, or ALC, uses dense arrays of micro-emitters and photodetectors. These are linked by lightweight bundles of fiber.
It moves data as light instead of electricity, using lots of tiny light sources rather than a few big ones. The cable supports runs of up to 50 meters, which Credo pitches as row-scale connectivity for AI scale-out and front-end networks.
Copper active electrical cables, or AECs, work well over short distances. Traditional optics cover longer distances but draw more power. The ALC is designed to sit in the middle, for links that are too long for copper and too short to justify power-hungry optics.
Credo says the cable draws only 12 watts per end, approximately 30% less power than comparable AECs. The company also says the ALC is ten times lighter than traditional alternatives.
Reliability is the real pitch
Credo is targeting a mean time between failures, or MTBF, of 100 million hours. MTBF is a statistical estimate of how long a component typically runs before something breaks. Note that this is a target, not a field-tested result.
The cable also includes channel redundancy and ZeroFlap link monitoring designed to enable hitless failover. If one channel falters, traffic shifts to another without the connection visibly dropping.
A “link flap” is when a connection repeatedly goes up and down. In a cluster of thousands of accelerators, one flapping link can force jobs to restart, wasting expensive compute time.
The research findings also note that micro-emitter optics promise better reliability and reduced latency compared with conventional optics.
Credo plans to start shipping the product in 2027.
Building a standard, not just a product
Alongside the launch, Credo introduced the Short-Reach Micro-Emitter Optics workstream, or SRMO, under the Open Compute Project. The goal is to standardize this new class of technology. Early supporters include SENKO and AFL, both names in the fiber connectivity world.
The Open Compute Project is the industry forum where large data center operators and suppliers hash out shared hardware designs.
The numbers behind the bet
Credo reported record revenue of $479 million for the first quarter of fiscal 2027, representing a 115% increase year over year.
The company anticipates more than $600 million in optical revenue by the end of the fiscal year. That figure leans on Credo’s broader optical push, including its acquisition of DustPhotonics. DustPhotonics adds capabilities in silicon photonics, which uses chip-manufacturing techniques to build optical components.
Credo is also reportedly working with major technology players including Amazon and Microsoft.
What this means
For investors, the launch reads as a deliberate expansion of Credo’s addressable market. The company built its recent growth on copper AECs, and copper has physical limits on distance.
The competitive pitch is clear: lower power, lighter weight, and a reliability target designed to make operators stop worrying about cables. In facilities where power budgets are a hard constraint, a 30% power saving per link can add up across thousands of connections.
Shipping does not begin until 2027. The 100 million hour MTBF figure is a target that will need validation at scale. The more than $600 million optical revenue goal is a forecast, not a booked result.
If more suppliers and operators join the SRMO effort beyond SENKO and AFL, micro-emitter optics could become a recognized category, with Credo as its founding member.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.