At what has now become the Linley Mobile and Wearables Conference, Linley pointed out that wearables are growing at 38%. However, there was a big caveat on Linley's slide, that this assumes "Apple releases a popular smart watch in 2017" followed by "popular Android watches in 2018." So that is a bit like saying that the market will grow fast if it has fast-growing products. I don't think Apple releases numbers for the Apple Watch, but all the estimates are that the market has saturated. The kind of people interested in a $350 watch have already got one, The way Apple only does annual product releases means that there tends to be a big spike in new sales (especially for iPhones) after each new release and then the shipments slowly decline until the next release. So it's possible that the next watch will cause new customers to purchase, or drive an upgrade cycle for existing Apple Watch customers. Or it's possible that none of that happens. I remain a skeptic on that 38% number, but as a person in the semiconductor ecosystem I'd love to be proven wrong. Pankaj Kedia of Qualcomm, the following day, said that watches are growing faster than any historical product. In three years they have shipped 10 million, then 25 million, and now 50 million. Of course, watches are a lot cheaper than smartphones, so in dollar terms this is less dramatic. Plus the silicon content is nowhere near as high so it isn't driving the technology roadmap of the foundries in the same way, with all the implications for advanced node capabilities in design tools. He also pointed out the operating system transition. In the first year, all 10 million smartwatches ran a real-time operating system (RTOS). The next year, sales went to 25M, but RTOS use fell to 8M. Now it is down to 7M. Almost all smartwatches run a high-level operating system, Linux or Android, since it is so much easier to program. Pankaj said that he was at the Linley Conference two years ago and Qualcomm was about to start shipping its first product. Now they are shipping over 100, so basically one new product per week over that two-year period. When he says "they," he means Qualcomm customers are shipping wearable products containing Qualcomm silicon. He pointed out that there are really two kinds of wearables. general purpose, such as watches or goggles, and special purpose such as step trackers. Qualcomm has produced two chips for the market, the 2100 that it announced in February and the 1100 announced in May. Both chips contain LTE modems and GPS functionality. The 2100 contains a quad-core ARM ® processor, the 1100 just a single core. These chips have gone into over 100 products in over 60 countries. Over 80% of Android smartwatches are based on Qualcomm, including the first LTE connected kids watch. They have a strong presence in GNSS (basically GPS but supporting all four satellite systems). That is a lot of functionality in each of those chips with multiple radios, high-end operating systems, and powerful processors. There are clearly a lot of products being designed. Just what volume they will ship in remains to be seen. Shipping 50 million products is an achievement, and it is still early. But it's not the smartphone market where annual sales are over 1.4B units per year, 20X and probably 200X in terms of the dollar value of silicon content. Meanwhile, things are not all going well. You've probably already heard that Intel recently recalled its Basic Peak watches since, in their words, they were getting so hot that it "could result in burns or blisters on the skin surface." Ouch. Pat Gelsinger, when he was Intel's CTO, pointed out that if we kept increasing clock rates then the power density would equal a rocket nozzle. I guess Intel wanted to prove him correct. Previous: CDNLive India Keynote![]()
![]()
↧