Co-Packaged Optics vs Pluggable 1.6T: Who Pays When the Laser Dies

The co-packaged optics debate revolves around the shift in the industry's packaging approach for optics, with a focus on whether hyperscalers or other entities bear the operational consequences. The transition from pluggable to co-packaged optics has shown notable differences in failure mode and cost implications. Pluggable optics are easily replaceable, consuming around 12-15 watts for 800G, while co-packaged optics integrate the photonic engine with the switch ASIC, resulting in a lower power consumption for the engine at 5-7 watts but a higher overall system-level saving due to the removal of the retimer and package-to-front-panel path.

However, the failure model of co-packaged optics is different, as a failure in the integrated engine could lead to a more costly chassis replacement instead of a simple field-replaceable unit swap. This change in failure mode carries significant implications for the operational rhythm and cost structures of different network environments.

At 1.6T, pluggable modules are projected to draw 25-30 watts, with the front-panel power budget of a 1U switch reaching its limit. This situation is expected to drive the adoption of co-packaged optics due to physical constraints. As the industry moves towards standardisation and multi-vendor optical engines, the transition from proprietary interfaces to open standards is anticipated to occur within one product cycle, ultimately affecting the pricing and availability of network services.