To download the latest issue
Jun 19th, 2014
How to turn MEMS, 3D stacking technologies and CMOS processes into optics?
Silicon photonics is getting more and more momentum with a very strong market pull linked to the increase of the flow data in data centers as well as the strong need to keep the energy consumption of data centers as low as possible.
And this is a field which is mixing optics, CMOS, MEMS and 3D stacking technologies. Silicon photonic is a business opportunity for different types of players: OSATs, MEMS players, semiconductors companies… Because it involves different challenges for packaging, optical alignment, electronics integration… there are needs for very diverse technologies and thus different packaging / micro machining / manufacturing approaches.
So MEMS, CMOS and 3D technologies are now again combined in order to empower optics into silicon (does it remember you something that happened between 1998 and 2001?). The big difference with the optical telecommunication bubble is… well… clearly not in the financial valorization of the companies: 2013 has been the year of more than $1B spent on acquisition of Silicon Photonics companies. The main difference compared to 2000 is clearly the market traction: nobody knows how to solve the growing problem of the increase of the data flow and the energy consumption in data center. A solution has to be found before 2018 and we are convinced at Yole Développement that photons will replace electrons and Si photonics will be the platform in the mid-term to solve this transition. But we are 4 years ahead to the real market start.
Also, these 4 years are needed to solve the remaining technical issues, which are still very important. Over the past years, it is becoming clear that technical choices will be better than others for a successful commercial development of Silicon Photonics. It is clear now that the light source is a big integration challenge. As silicon laser is probably years-away before realization…, the different approaches are likely to be either attached laser (e.g. Luxtera) or (InP) wafer to wafer / die to wafer bonding followed by post-processing (e.g. Intel or Leti).
Over the past years, we saw a shift from monolithic integration for electronic – photonic to hybrid integration as critical dimensions are very different. Today, favored approach seems to be two-chip hybrid integration (Cu-pillar from STMicroelectronics for example) as semiconductors and photonics critical dimensions are likely to be at least one order of magnitude different.
Apart from these technical choices, acquisitions and consolidation of the industry continue. Acquisitions are generally under the roof of modules/systems makers that enlarge their technologies portfolio as it can be faster and less money-consuming than R&D investments. We also see big players such as Intel having both approaches: acquisitions and R&D while some others have decided not to invest into silicon photonics as they think future designs will be accessible via foundry services. As many IC foundries have started proposing an offer for Si photonics wafer foundry services, this also could open some acquisitions opportunities in the near future.
2018 will be the real start of the business in volume, we forecast the Si Photonics market for 4 different applications: HPC, future generation of optical datacenters, telecom & others (including sensors, medical, consumer …). We estimated silicon photonics devices market will grow from less than US$25M in 2013 to more than US$700M in 2024 with a 38% CAGR. In 2018, emerging optical data centers from big Internet companies (Google, Facebook …) will be triggering the market growth.
So you have now 4 years in front of you to develop your Silicon Photonic solution… or buy the companies that have started before you.
More information are available in our new report: Silicon Photonics 2014.
More OPTO & PHOTONICS news