I like that article very much, because it describes technologies and challenges in very clear and understandable way.

As of whether POET's optical interposer will be a compelling technical solution, I'd say it is an answer to the following challenge, quoted from the article:

"New and innovative ways to manufacture connectors between fibers and from fiber to transceiver are needed to meet growing customer demand for both precision and low component price. However, very few manufacturing techniques are close to being inexpensive enough."

Well, the optical interposer's manufacturing technique is inexpensive enough. The article goes on and mentions wafer-scale integration, which is the way POET is going.

The article also mentions the interconnections between the components of a rack, which should be optical rather than copper. Here POET's GaAs platform, resp. the active optical cable (AOC), could come into play again. As we are all aware of, the integrated VCSEL transceiver has been shelved for the time being. However, with a strategic partner with big pockets and/or with cash-flow from interposer-based products, it could be revived again to solve the fourth-terminal problem.