Optical Computing is Serious Business

By John MacWilliams | December 16, 2014

The new iNEMI-MIT consortium on silicon photonics demonstrates that optical computing is serious business.

Intel LightPeak

Intel LightPeak

Fiber optics have had a long, uphill battle in our industry, with millions invested and dot-com busts, but now FO appears in launchpad mode in the near future. The newly formed Photonic Systems Manufacturing Consortium (PSMC) is funded through The National Institute of Standards and Technology (NIST), with an objective to analyze the various component parts of future silicon photonics systems; develop roadmaps for critical subsystems, packaging, and interconnects; and identify technology roadblocks that must be addressed. This will feed into a larger national effort, the goal of which is to lay the groundwork, with regional foundries and academic research, for a high-volume US silicon photonics manufacturing industry by the 2020s.

This will be accomplished by a global electro-optics R&D effort and component supply chain, which will include various efforts to pursue next-generation electro-optical systems.

Input is needed from OEM and EMS firms; semiconductor, IC packaging, waveguide and optical fiber, optical printed circuit, connector, and cable manufacturers; and smaller, specialty suppliers of optical subsystems, including aerospace/military fiber optic connectors and systems. This is a long list, but all stakeholders will eventually be affected by this emerging technology.

Of particular interest are those suppliers who have a vision toward high-volume manufacturing of future electro-optical systems based on low-cost silicon photonics technology (the integration of wafer/panel-processed SiPh integrated circuits packaged in advanced waveguide/interconnect SiP systems). This will break through to achieve Gb/Tb performance in data center, datacom, and high-performance computing applications. System scope is from chip to SiP to electro-optical systems to I/O ports and LAN cables.

One of the challenges facing this effort is the fragmented, proprietary (if not highly successful) individual efforts at work in the industry, which include active optical cable assemblies (AOCs), integrated optical engines, and a wide array of fiber optic interconnect products. They provide an excellent framework, but companies will need to see beyond their own protected IP  to the broader market opportunity that will benefit many.

The scope for connectors in this project includes optical mediums including direct chip-attach waveguides; interposers; above-board and embedded PCB glass and polymer waveguides; board-level optical engines for AOC and CPU optical signaling; printed circuits (rigid and flex) with optical capability; optical printed circuit, board-to-board, backplane, and other optical connectors; and cable assemblies.

We have identified more than 15 interconnect applications to be roadmapped. There is a call for white papers that deal with specific challenges and explain potential pathways to achieve HVM of silicon photonics systems through all levels of packaging – from the chip to I/O. We also need to develop parameters for the product/technology roadmaps for the following categories: On-chip OPCBs, connectors, and cables; chip package board-to-board connectors and cables; interposer/socket chassis-to-chassis connectors and cables; I/O connectors and cables; intra-system cables; and fiber/waveguides.

If you are interested in participating in the roadmap or white papers, please contact John MacWilliams at [email protected].

Visit the PSMC webiste.  

John MacWilliams
Get the Latest News
On TTI’s podcast the Distribution Download, our interconnect expert Bob Hult discusses the fiber optics.
eBook 2024 Bright Ideas