At the recent OIF 448G AI Workshop in Santa Clara, Mike Klempa of Alphawave Semi, who also serves as the OIF’s PLL Interoperability Chair, called for a broad, collaborative industry approach to define and implement 448G interconnect standards. Speaking during the SerDes panel, Klempa highlighted the architectural and analog front-end trade-offs associated with scaling to 448G speeds in support of advanced AI workloads.
Klempa noted that while technical challenges remain—particularly in choosing appropriate modulation schemes and defining interoperable SerDes specifications—the path forward hinges on industry-wide participation. Key focus areas include topologies ranging from chip-to-chip to long-reach links, all of which require careful optimization and alignment.
The OIF has already kicked off a framework document to guide the development of 448G standards across multiple application spaces. Klempa stressed the importance of input not just from SerDes and connector vendors, but also from material suppliers and hyperscaler end users, who are driving demand for millions of ports. The collaborative momentum, he said, is essential to ensuring interoperability and accelerating time to deployment.
🔑 Key Points:
• OIF is exploring multiple application scenarios: long-reach, chip-to-chip, and chip-to-module links.
• Klempa discussed trade-offs in SerDes architecture, analog front ends, and modulation schemes.
• The OIF has initiated a framework document process to align the ecosystem.
• Participation is encouraged from vendors, end users, and system architects to ensure real-world readiness.
• Keynotes at the workshop included insights from major hyperscalers driving port deployment demand.
Want to be involved our video series? Contact info@nextgeninfra.io
https://ngi.fyi/oif448-oif-michael
