While early 2021 was more dramatic in terms of new co-packaged optics (CPO) products, the past year proved to be the period in which CPO took its first baby steps towards the data communications mainstream. Two important practical developments occurred. One, which we discuss in this article, is the OIF Co-Packaged Optics Framework. The other is Near-Packaged Optics (CPO Light as it has been called), which we will discuss in a later blog.
The OIF organization has put together a conceptual framework for taking CPO forward technologically and CIR thinks this will set the scene in terms of product development for years to come. Five aspects of the OIF that we believe will be especially important are set out below.
#1 The Framework itself: With the framework in place and broadly accepted there is now an established and centralized forum (the OIF) in which CPO technology can evolve and mature. Until now it seemed that different bits of CPO might evolve within several different groups with an obvious loss of efficiency. This does not mean that standards/MSA groups other than the OIF need to bow out of CPO development, merely that all these groups can now look to the OIF for leadership.
#2 Pluggability restored: Well not really, but with its heavy focus on the role of external pluggable laser in future CPO interfaces, some of the advantages of pluggability (i.e., modularity) is restored to the interface, partially blunting the nervousness in the datacom community at abandoning pluggability after so many years of good service. It now seems certain that when a CPO interface goes wrong, you won’t have to throw away a module costing several thousand dollars. Instead, you can just plug in a laser.
#3 Conserving the supply chain: In addition to concern over the potential demise of pluggability there has also been worries that the long separate photonic and electronic component supply chains would have to merge in some way in an era of advanced optoelectronic packaging. This implies new alliances, mergers, and acquisitions at a time when supply chain issues are messy enough for geopolitical reasons. For now, at least, the likelihood that the lasers in the CPO package will be separate devices may mean that laser makers can leave supply chain integration for a while longer. It also opens a new opportunity for laser makers in the form of the new kinds of lasers specially designed for the coming CPO package.
#4 Cooler: The single most important driver for CPO is to cut down on heat and – what is essentially the same thing – power consumption. Some vendors are saying that with CPO, perhaps 30% to 40% of heat can be reduced in a data center compared to high-speed pluggable modules. However, even with CPO cooling will be necessary and in the White Paper that has emerge from the OIF, there is much discussion about the various cooling mechanisms that might be employed for future CPO links.
#5 CPO modules: The beginning: Last, but very much not least, the Framework White Paper, makes mention of some kind of emerging CPO module MSA. This work has a long way to go, but hints at a day when a suite of CPO based datacom interfaces will emerge like what we have in the pluggable space today.
While none of these items — #1 to #5 — could be called major steps forward, they each, nonetheless, will make it a easier to conceptualize and manufacture actual CPO products ready for market. It may be a few years before CPO products are a data center staple, but the contents of the OIF CPO Framework, along with the Framework itself, are setting the stage for the next phase in data center optoelectronics.
See our latest report on Co-Packaged Optics that has just been issued.
