Good article, some key points from POET perspective

In terms of technology, the biggest elephant in the room is the lack of a silicon-photonics laser integrated in the chip to get data off quickly, Gartner said.  “No one has developed that yet, and it is one of the holy grails of photonics.”

From the Dec 2018 company overview, p3 & CIOE presentation, p9:  First and only company to introduce true chip-scale integration and wafer-scale assembly and test of photonics devices.  Multi-chip module approach is described as True Platform Technology offering a design that is superior to traditional integrated silicon photonics chips.

The next evolution is to develop a 400G optical port over a single fiber across 500 meters at less than $1 per gigabit and with power [less than] 5 mW/Gb,” Yole stated. Terabit-per-second rates should follow.

From the Dec 2018 company overview, p11 & June 2018 technology background, p39:  BOM for complete 100G CWDM transceiver estimated at $100, translates to $1 per gigabit. OI architecture suitable for 500m - 10km devices.  Advertised as low(er) power, but unable to locate exact numbers on this.

“As we move from 400G to terabit speeds it gets challenging to get all that capacity off a line card or ASIC and you're going to need a different optics architecture to handle speeds like that.

The OI is advertised as speed agnostic with true silicon CMOS compatibility, therefore architecture already in place for terabit.

Regarding Cisco’s purchase of Luxtera, looking at the optical systems ecosystem again, we see Cisco is listed as system integrator, and currently buys from select module makers such as Luxtera, so this makes sense.  Luxtera buys from component makers, list of component makers below.  One would assume there is now a need for Cisco to buy directly from component makers....

The Optical Systems ecosystem (Dec 2018 company overview, p17)