Acacia Casualty: Back to Si Photonics Hype

In addition to the start of out-of-whack valuations for optical component companies resulting from the convergence of two massive market forces, the initial overexcitement about Acacia Communications in general, and its focus on “silicon photonics” (“SP”) specifically (despite arguably, the superficiality of any positive impact on the firm’s cost), unfortunately made it more necessary for vendors to include that capability in their marketing efforts, regardless of any strict definition of the term. In several cases, such a transition would occur irrespective of their previous public stances on the technology. It also seems to be ignored that there are doubts that a high-volume producer of SPs, Luxtera, is even profitable, albeit Intel, regardless of actual shipments, helped to artificially drive PSM4 prices at the 100G rate down. Furthermore, at least for a while beforehand, it became fashionable to look at “SP” critically and realistically in the industry, such as with coupling loss problems at Luxtera, which we will consider in greater detail below.

Followers of our separate LinkedIn company blog site would have read the post about a couple of weeks ago: “Did Acacia Give the Whole Story on ADVA?” We addressed the implication at a conference “that Acacia only stressed a loss of market share by ADVA Optical Networking for the reduction in purchases from Acacia, and not that the customer had decided to buy discrete components for integration at a substantially lower cost.” Once more, the impact of “SPs” on Acacia’s cost structure is open to question.

Turning our attention to Luxtera, at OFC 2015, it almost got laughed out of the room at the Rump Session. While we thought some of the criticism toward the supplier was somewhat unfair, it still reflected a willingness to be truthful about the future prospects of “SPs”. Now the industry has returned to a reluctance by optical engineers to speak negatively about such solutions because of potential adverse ramifications on the perceptions of their companies.

Even more worrisome is the increasingly haughty attitude in the marketplace over the nebulous nature of the phrase, “SP.” At Session 3 of the OIDA Workshop on Manufacturing Trends for Integrated Photonics in March, this writer wrote in that other blog: “I was impressed with at least the notion of using the original definition of silicon photonics by Sicoya of a monolithic approach at 100G (and it claims it can go to higher rates)…” When I suggested that the vendor should not be so generous to its competitors and highlight the importance of the achievement in avoiding a hybrid approach, it was met with indifference. While it is true that customers do not care what something is called, as long as it works and is cost-effective, the lack of concern about the precise truth of the meaning of a term in a technical session was ironically indicative of the lack of progress by such technology in general because of its inherent, immutable limitations.

For example, much of silicon photonics is accomplished with edge coupling of the optical signal. However, it can be a challenge to overcome problems, such as with thermal performance and topography restrictions as well as with optical testing. In addition, high-resolution fabrication is necessary.

Luxtera’s use of a grating coupler is easier and more practical in that it permits wafer-scale optical testing, eliminates the need for facet polishing, and allows for complete flexibility on the placement of the coupler. Conversely, the grating coupler has low efficiency, operational bandwidth constraints, and strong polarization dependence. Also, something like an angle-polished fiber is a necessity. Moreover, the center wavelength frequently cannot be perfectly controlled resulting in substrate scattering and mode mismatch.

Regarding the last point, the coupling loss can be much greater than is the case even with the edge approach. As much as 2-4dB can be lost with the grating coupler. In fact, that is why Luxtera only does a 500-meter solution because it simply cannot get enough power out of its devices to go, say, 10 kilometers.

Returning to Acacia, despite the greatly reduced amount of hoopla, the impact on messaging throughout the optical space with the wildly successful IPO at the beginning, is clear. Its stated mission has been the “siliconization of the optical interconnect.” Acacia insisted that its “SP” was a critical and unique differentiator. The company compared itself to Intel, paradoxically, because again, that giant chip player never even came close to reaching its promises of impacting the market with “SP.”

[written by Mark Lutkowitz]