• Skip to main content
  • Skip to search
  • Skip to footer
Cadence Home
  • This search text may be transcribed, used, stored, or accessed by our third-party service providers per our Cookie Policy and Privacy Policy.

  1. Blogs
  2. Breakfast Bytes
  3. Smartphones: Linley's Annual Review
Paul McLellan
Paul McLellan

Community Member

Blog Activity
Options
  • Subscribe by email
  • More
  • Cancel
Apple
Samsung
wearables
mobile
Smartphone
Huawei

Smartphones: Linley's Annual Review

2 Aug 2016 • 3 minute read

 Last week was the Linley Mobile Conference, although it is now the Mobile and Wearables Conference. As always, Linley Gwenap gave the opening keynote with his overview of the industry. One milestone that coincidentally was announced is that Tim Cook, Apple's CEO, announced the sale of the billionth iPhone at a company meeting.

The smartphone industry is very large, which makes it attractive. By 2020 there will be 1.95B units. Even though growth has slowed from the astronomical levels of a few years ago (remember, the iPhone was only introduced in 2007), it is still 6.5% between now and 2020. However, it has been a very difficult market to penetrate as a merchant semiconductor suppler. The three market leaders, Samsung, Apple, and Huawei/HiSilicon, all design their own application processors. The rest of the market is largely Qualcomm, MediaTek, and Spreadtrum. In fact, these three, along with the internally developed application processors, make up 98% of the market. Samsung and Huawei also make their own modems (Apple uses Qualcomm, although there is a rumored design win for Intel for part of the US market).

 Consolidation in the market seems to be complete. Intel has finally given up, although they still have LTE modems which are expected to show up in phones later this year. Plus LTE modems may become more standard on laptops, too. Marvell shut down their cellular business late last year. Broadcom tried and failed to sell their mobile business as a going concern. That followed earlier exits from the market by Freescale (now part of NXP), ST Ericsson, Texas Instruments, and probably a couple more that don't immediately spring to mind.

 In terms of unit shipments, Qualcomm is slightly ahead of MediaTek. Since they are strong at the high end of the market, they are presumably a long way ahead in dollars. Since most of the future growth in smartphones is expected to be at the low end, this should favor MediaTek and Spreadtrum. Qualcomm seems to have won back a good chunk of Samsung's business, which it had lost (last year's Galaxy S6 used no Qualcomm, but the S7 used about 30%).

Smartphones continue to be very important for the semiconductor ecosystem and for EDA and IP. That's because they are one of the products that pushes to the most advanced technology nodes as fast as they can (along with FPGAs and graphics). This means that the market is an early adopter of tools that support these processes with the complexity of double, triple, and maybe even quintuple patterning and other features that the designers cannot ignore.

 The low-end market players license their IP from 3rd party IP companies like ARM, Imagination, and Cadence. Qualcomm designs their own, as does Intel to the extent that it is still in the market. In the middle are the market leaders who license technology from ARM but build their own CPUs. Of course, that is expensive so it is only the largest players that can afford to get differentiation that way.

GPUs continue to get bigger. Apple's A9X (iPad Pro) and Samsung's Exynos 8 both use 12-core GPUs. This can improve power efficiency since the GPU can run at a lower clock rate. It also supports the increasingly high resolution of the high-end retina type displays. But it does require a lot of bandwidth. The A9X uses a 128-bit memory bus delivering over 50GBps. Another change is the move to cache-coherent interconnect that obviates the need for large amounts of data to be explicitly copied between CPU and GPU. ARM's Mali-G71 is the first cache-coherent GPU.

The transition to LTE for the air interface is in full swing, with the crossover between LTE (4G) and 3G happening this year. It is now 39% of units and 58% revenue. LTE should be half of all units by 2018, with a fast ramp in China.

The next big transition is to 5G. Deployment is likely in 2020, although some operators like Verizon are planning to roll out a subset of the capabilities (and no doubt will call it "5G" anyway). The rest of the first morning at the Linley conference was devoted to 5G. More about that coming soon.

Next: Chipworks Looks at Smartphones

Previous: CDNLive Boston