Why XPO Resonated at OFC 2026

· Source: Big Data & AI News - EE Times · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure · Depth: Advanced, medium

Summary

eXtra-dense pluggable optics (XPO) emerged as a key solution at OFC 2026, held March 15–19, addressing critical challenges in next-generation AI infrastructure, including bandwidth density, power efficiency, thermal management, and serviceability. Unlike co-packaged or on-board optics, XPO maintains the operational benefits of pluggable systems while significantly increasing density, enabling over 200 Tbps within a single open rack unit with cold-plate liquid cooling. Over 100 companies have joined the XPO multi-source agreement (MSA), with more than 10 vendors demonstrating full retimed (FRT), half retimed (LRO), and linear (LPO) XPO modules. The XPO module, measuring 60.8 mm (W) × 111.8 mm (L) × 21.3 mm (H), features a "belly-to-belly" paddle card design with a central cold plate for high-power components, supporting flow rates from 0.35 LPM to 0.7 LPM. This design improves thermal performance, reliability, and enables 4x higher front-panel density compared to OSFP modules, reducing switch rack footprint by approximately 75%.

Key takeaway

For CTOs and VPs of Engineering scaling AI infrastructure, XPO presents a compelling solution to overcome bandwidth, power, and density bottlenecks. Your teams should evaluate XPO's ability to deliver 4x system-level density and improved thermal management, which can significantly reduce infrastructure costs and enhance reliability in hyperscale AI data centers. Consider integrating XPO to future-proof your optical networking architecture and simplify deployment.

Key insights

XPO offers a scalable, power-efficient, and high-density optical interconnect solution for next-generation AI infrastructure.

Principles

Method

XPO modules integrate high-power components facing a central cold plate for liquid cooling, using 48V power distribution and edge-connector paddle cards to achieve high density and thermal stability.

In practice

Topics

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Architect, AI Hardware Engineer, MLOps Engineer

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Big Data & AI News - EE Times.