As we reach the end of August and stand on the precipice of Labor Day Weekend in the US, it’s that time of year again; time for me to bemoan the passing of another summer in New England. With each year, summer seems to pass even more quickly and this one was no exception. It started out with a nice (but very long) trip to Hawaii for this year’s International Microwave Symposium. In addition to beautiful scenery, I caught up with some friends and of course, I saw the latest developments in the microwave industry. From there, it was a bit of additional travel around some very interesting project work and then off to vacation. My vacation travel party was a bit smaller than I would’ve liked. My daughter was only able to join me for a week because she now lives in South Florida (where she is just a bit too fond of reminding me that EVERY day is summer!), but as they say “a bad day on vacation is better than a good day at work”!
I don’t completely buy that, but what I do buy is the interest in 5G and what it will mean for GaN, as well as the entire compound semiconductor industry. Those topics were central to the theme of IMS2017 and the Strategy Analytics Semiconductor Component Applications Group addressed some of the issues, challenges and opportunities during our annual Lunch and Learn session at the conference (Which Semiconductor Technologies will Successfully Solve the 5G Conundrum?) I do have to say that despite summer winding down, it is always makes me happy to talk about 5G and GaN.
In addition to the group presentation that we gave in Hawaii, I just published my latest thoughts and forecast on the RF GaN market (RF GaN Market Update: 2016 - 2021). As a bit of a teaser from the report, RF GaN revenue grew nicely, again, in 2016. Defense and commercial applications for GaN will both continue to grow, but one segment will outpace the other. Not to understate the importance of the market, but even with double-digit growth the next five years, RF GaN revenue will still be less than $1 billion. This is not a slight, because it’s still a relatively new market and regular readers of this blog know that I am as bullish on GaN as anyone, but sometimes putting things into perspective is useful.
What is also useful, but very challenging, is trying to read the 5G tealeaves. The entire semiconductor industry wants the answer to the “5G technology conundrum”, because the vision of 5G has the potential to be a growth engine well into the next decade. However, the reality is that there will be technology winners and losers as the network architectures evolve and deploy.
There are many moving parts still in play, so the tealeaves have not settled just yet. It appears mobile broadband networks in the < 6 GHz frequency range and fixed broadband networks in the 28 GHz (and above) frequency range will be the first to deploy. This will be the necessary stake in the ground to announce the beginning of the 5G era. With South Korea serving as the host for the 2018 Winter Olympics, it seems a safe bet that they will have 5G capabilities for the games.
How does GaN figure into this? One of the fundamental building blocks of the 5G network architecture will be Massive MIMO antennas. The spectral efficiency, capacity and capability gains from these antennas will be central to 5G architectures. The industry is in the process of determining the sweet spot for the number of elements in these antennas. The obvious determining factor is output power. With regulations on the maximum transmit power for base stations (and user equipment), increasing the number of radiating elements decreases the transmit power of each element. Since the number of antenna elements may range from single-digits into the thousands, every power amplifier technology has a very vested interest in the “right answer”.
The right answer is liable to rely on more than just the output power. The vision of 5G incorporates multiple users in a 3-D volume and this requires antenna beams that can be steered in azimuth and elevation. This, in turn, requires sophisticated processing and DC power dissipation is a very significant issue. Ideally, every element in an array uses digital beamforming to get the most functionality out of the entire array, but this is approach dissipates a substantial amount of power. The silicon proponents argue that more elements is the right answer, because of the capability that approach would provide that could be enabled by the integration advantages of silicon. The GaN proponents are at the other end of the spectrum, arguing that the power handling, size, efficiency, frequency and bandwidth advantages of GaN make that the right choice, because fewer elements means less DC power dissipation and millimeter wave networks will depend on smaller packaging schemes to accommodate optimum element spacing. Since the silicon market is very large, 5G may be a small ripple, but to the < $1 billion GaN market, 5G could be a tidal wave.
