Gazettabyte is asking industry and academic figures for their thoughts after attending ECOC 2022, held in Basel, Switzerland. In particular, what developments and trends they noted, what they learned, and what, if anything, surprised them.
In Part 3, BT's Professor Andrew Lord, Scintil Photonics' Sylvie Menezo, Intel's Scott Schube, and Quintessent's Alan Liu share their thoughts.
Professor Andrew Lord, Senior Manager of Optical Networks Research, BT
There was strong attendance and a real buzz about this year’s show. It was great to meet face-to-face with fellow researchers and learn about the exciting innovations across the optical communications industry.
The clear standouts of the conference were photonic integrated circuits (PICs) and ZR+ optics.
PICs are an exciting piece of technology; they need a killer use case. There was a lot of progress and discussion on the topic, including an energetic Rump session hosted by Jose Pozo, CTO at Optica.
However, there is still an open question about what use cases will command volumes approaching 100,000 units, a critical milestone for mass adoption. PICs will be a key area to watch for me.
We’re also getting more clarity on the benefits of ZR+ for carriers, with transport through existing reconfigurable optical add-drop multiplexer (ROADM) infrastructures. Well done to the industry for getting to this point.
All in all, ECOC 2022 was a great success. As one of the Technical Programme Committee (TPC) Chairs for ECOC 2023 in Glasgow, we are already building on the great show in Basel. I look forward to seeing everyone again in Glasgow next year.
Sylvie Menezo, CEO of Scintil Photonics
What developments and trends did I note at ECOC? There is a lot of development work on emergent hybrid modulators.
Scott Schube, Senior Director of Strategic Marketing and Business Development, Silicon Photonics Products Division at Intel.
There were not a huge amount of disruptive announcements at the show. I expect the OFC 2023 event will have more, particularly around 200 gigabit-per-lane direct-detect optics.
Several optics vendors showed progress on 800 gigabit/ 2x400 gigabit optical transceiver development. There are now more vendors, more flavours and more components.
Generalising a bit, 800 gigabit seems to be one case where the optics are ready ahead of time, certainly ahead of the market volume ramp.
There may be common-sense lessons from this, such as the benefits of technology reuse, that the industry can take into discussions about the next generation of optics.
Alan Liu, CEO of Quintessent
Several talks focused on the need for high wavelength count dense wavelength division multiplexing (DWDM) optics in emerging use cases such as artificial intelligence/ machine learning interconnects.
Intel and Nvidia shared their vision for DWDM silicon photonics-based optical I/O. Chris Cole discussed the CW-WDM MSA on the show floor, looking past the current Ethernet roadmap at finer DWDM wavelength grids for such applications. Ayar Labs/Sivers had a DFB array DWDM light source demo, and we saw impressive research from Professor Keren Bergman’s group.
An ecosystem is coalescing around this area, with a healthy portfolio and pipeline of solutions being innovated on by multiple parties, including Quintessent.
The heterogeneous integration workshop was standing room only despite being the first session on a Sunday morning.
In particular, heterogeneously integrated silicon photonics at the foundry level was an emergent theme as we heard updates from Tower, Intel, imec, and X-Celeprint, among other great talks. DARPA has played - and plays - a key role in seeding the technology development and was also present to review such efforts.
Fibre-attach solutions are an area to watch, in particular for dense applications requiring a high number of fibres per chip. There is some interesting innovation in this area, such as from Teramount and Suss Micro-Optics among others.
Shortly after ECOC, Intel also showcased a pluggable fibre attach solution for co-packaged optics.
Reducing the fibre packaging challenge is another reason to employ higher wavelength count architectures and DWDM to reduce the number of fibres needed for a given aggregate bandwidth.