We’ve just published our Wireless Connectivity Market Share for 2018, showing that Broadcom remains the biggest retailer of radio chips and that MediaTek’s plan to diversify is working, but the numbers sidestep the increasing problem of how we should be counting these chips anyway.
At Gartner we segment wireless chips by function – sorting out market share by Wi-Fi, Bluetooth, GNSS, and so forth, and building a forecast split along the same lines. However, the “combination” category is becoming an increasing problem as one chip is increasingly asked to do several things.
In phones we don’t see any single-function chips at all these days – tear down a smartphone and you’ll see a Wi-Fi/Bluetooth combination at best, often nothing at all as the wireless connectivity gets built into the SoC. We occasionally see discrete wireless chips in laptops, and set-top boxes, but everywhere else the combination chips are taking over.
Wi-Fi/Bluetooth is the most common combination, but every year we see a greater variety. Bluetooth and 802.15.4 are being pushed together for companies that don’t want to count out ZigBee (or Thread), while putting LPWA and GNSS onto a single chip makes for a cost-effective asset tracker.
GNSS is a fine example. We’ve never had more products capable of positioning themselves using satellite technology but despite that we’re forecasting a decline in GNSS chip shipments over the next couple of years, as the functionality gets integrated into combination silicon.
We can’t track every possible combination, especially as the more-esoteric options proliferate, and things are going to get worse if we consider a software defined future. 5G networks are already using cloud radio access networks (C-RAN) which sees radio protocols decoded in software running on generic hardware, so we have to ask: how long will it make sense to track semiconductors by the radio protocols they support?
For the moment we can confidentially split our semiconductor market share and forecast by radio protocols, but it seems those dividing lines won’t be around for much longer. Anyone with opinions on how we might fashion our wireless semiconductor market analysis in the future is welcome to get in touch or pitch in below.