Posted on by Newsdesk

Credo Breaks the Memory Wall Using AMD Programmable Logic-Based AI Accelerators

Credo Breaks the Memory Wall Using AMD Programmable Logic-Based AI Accelerators

Credo Demonstrates OmniConnect Running on AMD Versal™ Adaptive SoC Architecture to Offer System Integrators Terabyte-Class Memory Capacity and Bandwidth Without Waiting on Custom Silicon

SAN JOSE, Calif.–(BUSINESS WIRE)–Credo Technology Group Holding Ltd (Credo) (NASDAQ: CRDO), an innovator in providing connectivity at scale through fast, reliable, and energy-efficient system solutions, today announced that it is showcasing high-performance memory scaling for programmable logic-based AI accelerators this week at the 2026 OCP Global Summit in San Jose. Credo will demonstrate its OmniConnect Weaver memory fanout gearbox in a programmable logic-based environment, enabling high memory bandwidth and high memory capacity on an AMD Versal™ AI Edge Series Gen 2 adaptive SoC evaluation board. The system puts memory capacity and bandwidth previously only reserved for custom ASICs within the reach of programmable logic.

This press release features multimedia. View the full release here: https://www.businesswire.com/news/home/20261013236839/en/

At the 2026 OCP Global Summit in San Jose, Credo will demonstrate its OmniConnect Weaver memory fanout gearbox in a programmable logic-based environment, enabling high memory bandwidth and high memory capacity on an AMD Versal™ AI Edge Series Gen 2 adaptive SoC evaluation board.

At the 2026 OCP Global Summit in San Jose, Credo will demonstrate its OmniConnect Weaver memory fanout gearbox in a programmable logic-based environment, enabling high memory bandwidth and high memory capacity on an AMD Versal™ AI Edge Series Gen 2 adaptive SoC evaluation board.

News Highlights:

  • Credo and AMD have built a working emulation platform running OmniConnect Weaver memory disaggregation on AMD Versal AI Edge Series Gen 2 adaptive SoC architecture.

  • The platform enables terabytes-class of LPDDR memory with very high capacity and bandwidth to a programmable-logic XPU over VSR links, on standard organic packaging with no silicon interposer.

  • Application software runs unmodified as the expanded memory is transparent above the controller.

As AI inference workloads scale, memory bandwidth and capacity have become critical bottlenecks, often constraining performance more than compute throughput itself. Credo’s OmniConnect Weaver memory gearbox is designed to address this directly. The AMD Versal adaptive SoC platform provides a highly programmable environment with high performance for faster time to market. By integrating Credo’s OmniConnect technology, customers can break the memory ceiling with access to larger memory pools and increased memory throughput without fundamental changes to application software.

“Memory is what limits inference today, not compute, and the gap widens every generation,” said Vishal Shah, Vice President of Product, Credo. “By combining the versatile AMD Versal adaptive SoC platform with Credo’s OmniConnect Weaver memory expansion technology, we are enabling customers to build scalable memory architectures that can unlock higher performance and greater efficiency for next-generation AI systems.”

“As AI inference workloads scale, customers need a practical path to higher memory performance without sacrificing the flexibility that drove the need for a programmable logic design in the first place,” said Manuel Uhm, Director of Silicon Marketing, Adaptive and Embedded Computing Group, AMD. “This collaboration with Credo demonstrates how customers can unlock significantly more memory bandwidth and capacity on the AMD Versal AI Edge Series Gen 2 device while preserving the efficiency, power and time-to-market advantages of programmable logic-based designs.”

Credo will demonstrate OmniConnect Weaver on the AMD Versal AI Edge Series Gen 2 adaptive SoC evaluation board at the 2026 OCP Global Summit, Booth B2. More information is available at https://credosemi.com/products/omniconnect/.

About Credo

Credo’s mission is to transform connectivity at scale through fast, reliable, and energy-efficient system solutions. Our high-speed copper and optical interconnect technologies deliver industry-leading power and performance from chip to cluster to meet the ever-expanding data infrastructure demands of AI.

Our vertically integrated connectivity portfolio is comprised of our flagship purple ZeroFlap (ZF) Active Electrical Cables (AECs) and ZF optical transceivers; optical components including silicon photonics-based photonic integrated circuits (SiPho PICs) and DSPs; OmniConnect AI memory and chip-to-chip interconnect; and retimers for Ethernet and PCIe—supported by our PILOT diagnostic and analytics software platform. Credo innovations enable our customers to connect the systems that connect the world.

For more information, please visit https://www.credosemi.com. Follow Credo on LinkedIn.

Credo, the Credo logo and the color purple when associated with AECs are registered trademarks of Credo Technology Group Limited in the United States and other jurisdictions. All other trademarks referenced herein are the property of their respective owners.

Media Contact:

Kristin Hehir

[email protected]

Investor Contact:

Dan O’Neil

[email protected]

KEYWORDS: California United States North America

INDUSTRY KEYWORDS: Software Networks Hardware Artificial Intelligence Data Management Technology Apps/Applications Semiconductor

MEDIA:

Photo
Photo
At the 2026 OCP Global Summit in San Jose, Credo will demonstrate its OmniConnect Weaver memory fanout gearbox in a programmable logic-based environment, enabling high memory bandwidth and high memory capacity on an AMD Versal™ AI Edge Series Gen 2 adaptive SoC evaluation board.
Logo
Logo