I just closed out the first day APEC 2017 (Applied Power Electronics Conference) in Tampa after spending last week at OFC (Optical Networking and Communication Conference & Exhibition) in Los Angeles and I’m contemplating the GaAs market and the overall compound semiconductor landscape. If I were to personify the GaAs market and its relationship to the compound semiconductor industry, the feelings would be characterized by confusion and uncertainty. Confusion because the GaAs industry is trying to digest changing demand, filled with the uncertainty of charting a satisfying course forward.
For more than a decade, GaAs device revenue has been in lockstep with developments in cellular terminal applications, with some contributions from other wireless applications. As cellular terminals grew in quantity and sophistication to both enable and support the onslaught of data traffic, the GaAs market flourished. As I’ve mentioned in other blogs and presentations, this growth engine is slowing and even with data traffic continuing to grow, the GaAs supply chain is working diligently to keep revenue essentially flat.
Even though I’m only one day into APEC, the broader compound semiconductor industry is on a different trajectory than GaAs. At OFC, many discussions centered on how quickly system level data rates will be increasing past 400 Gbps to 600 Gbps and even 800 Gbps. To achieve this system level performance, the “lanes” that multiplex to the system level data rates are climbing from 25 Gbps to 50 Gbps, with higher rates envisioned with improvements to modulation schemes and silicon processors. GaAs still plays a role in this market segment, but as data rates increase, InP and SiGe technologies are becoming much more significant and there is even the specter of CMOS for some of the electrical applications in this segment.
One common thread at both conferences is the growing importance of data centers, along with providers like Google, Facebook and Amazon. Because of the scale and data traffic, fiber offers benefits versus copper, even in short reach intra-datacenter applications. The scale and loss inherent in data centers using existing technology also provides an attractive target for the power electronics industry. While silicon currently dominates this industry, GaN and SiC are starting to capture market share and the market opportunity for GaN could potentially become larger than in RF applications.
The future looks bright for the compound semiconductor industry, but the technology mix is changing. GaN, SiGe, InP and SiC all appear destined for strong revenue growth. In comparison, GaAs seems destined for much slower growth. So, is this a case of “you can’t go home again” where the GaAs industry has to accept the consequences of change and move on with a modified role? The GaAs supply chain has been very resilient over the years, evolving in the face of challenges to become the dominant compound semiconductor technology. Emerging 5G standards and architectures do represent a wildcard and a potential growth engine for GaAs and the entire compound semiconductor industry. Perhaps the message is change is inevitable, but you ignore history and the path to the present at your own peril.
I’m planning to expand my coverage of these compound semiconductor technologies and market applications in 2017. It will be very interesting to follow the developments in these dynamic segments, so stay tuned!