Apple’s power in the mobile industry has given it a kingmaker position, able to make and break its suppliers with its changing supply chain decisions. Its desire to assert more and more control over its components triggered the collapse of Imagination Technologies, has landed Qualcomm in a costly legal battle, and now threatens to help Intel dip permanently below Samsung in the global semiconductor tables.
Apple boosted Intel’s fortunes in 2016 when it decided to use the company’s LTE modems in some iPhone models, introducing it as a second source alongside Qualcomm. But now it is reported to be planning to dump Intel x86 processors in the Mac family, as early as 2020, and replace them with an ARM-based chip designed in-house.
This move would be a logical one, emulating what Apple did years ago when it adopted an inhouse processor for the iPhone, replacing its original Samsung offering. That gave it full control over the availability, performance and pricing of the most expensive iDevice component.
Apple has continued to bring components inhouse, most recently its graphics processor unit (GPU), replacing Imagination’s technology with its own and triggering the sale of the UK graphics specialist. A big concern here is how Apple’s inexperience in building components like CPUs and GPUs will influence the video experience. Reports suggest Apple intends to reinvent the Mac OS to align more with the iPad and iPhone – potentially meaning a complete app overhaul. Although this scenario is mere speculation for now.
So the Mac move would follow the pattern of seizing control of key elements where it can, and otherwise introducing second and third sources to reduce its dependence on any single vendor, and to improve its ability to negotiate pricing and availability. So while it has failed to excise Samsung from its supply chain, especially as a source of memory chips and displays, it has actively supported alternative vendors such as Toshiba and Sharp.
On April 2, Intel’s share price fell by more than 6% on the reports that it was to be the latest casualty of Apple’s quest for control, and the disproportionate effect its buying decisions have on the whole industry. And some analysts believe that Intel could also lose the modem business only two or three years after it won some of those slots away from Qualcomm.
That is because Apple is reported to be, belatedly, addressing one of the negative aspects of its decision to use its own mobile processor. That prevented it adopting a highly integrated processor/modem system on chip (SoC), which could have saved power consumption and space. It has continued to use separate third party modems, from Qualcomm and now also from Intel (via its acquisition of Infineon’s wireless business, which supplied modems for the first two iPhone generations).
Now that may change, as it becomes more urgent to achieve the high levels of integration which Qualcomm’s customers, in particular, have been able to harness to reduce cost and power consumption. Price pressures are mounting even for the iPhone, and there are more connectivity and processing options to pack into a smartphone (vision processors, new WiFi radios and so on). That may mean Apple is working on an all-in-one SoC.
Jim McGregor, principal analyst at chip research firm Tirias Research, told EETimes: “I think there is a double threat to Intel. Intel will also likely lose Apple’s modem business once Apple integrates a modem into its mobile SoCs.” He thinks that will be in the “same time frame” as the chance of supplier for Mac processors, with both possibly happening next year.
Sources told Bloomberg that Apple would adopt a new Mac processor – currently being developed under a codename of Kalamata – in stages from 2020. It would be expected to introduce the new chip first in its ultraslim Mac Air notebooks, whose small size and long battery life put it in a similar category to larger mobile devices like the iPads. In the past, there have been rumors that Apple might adapt its iDevice processor for its mobile notebooks, though the Bloomberg sources say Kalamata will be a separate family which will gradually be introduced to the larger professional and desktop Mac models.
However, convergence of the product ranges would be a logical step as PCs get more mobile and smartphones gain PC-like power. “We think that Apple is looking at ways to further integrate their hardware and software platforms, and they’ve clearly made some moves in this space, trying to integrate iOS and macOS,” Shannon Cross, an analyst at Cross Research, told Bloomberg. “It makes sense that they’re going in this direction. If you look at incremental R&D spend, it’s gone into ways to try to vertically integrate their components so they can add more functionality for competitive differentiation.”
If Apple does make the complex transition from x86 to ARM, taking advantage of its architectural ARM licence (which allows it to build its own customized designs on the cores), it will be a boost for the entire ARM ecosystem. This will not be seen in terms of units, since the Mac accounts for less than 1% of sales in a PC space that is declining anyway. But it will help make the case that anything Intel can do, ARM can do too these days.
The company, now owned by Softbank of Japan, has always stayed ahead of Intel in its power efficiency, but has often struggled to catch up in sheer horsepower. That has been changing, with the introduction of 64-bit cores and of ARM-based processors for high end devices and even for servers. A top end Mac running on an ARM chip, and achieving the same performance as the x86 models, would be an important proof point for ARM licensees.
While the ARM instruction set still lacks some of x86’s performance and features, increasingly the micro-architecture is more important than the instruction set in driving capabilities. Tirias believes Apple will use its Metal 2 application programming interface, and its Core ML machine learning framework, to offload tasks to GPU-based accelerators, to make up for any shortcomings on the central processor.
Apple already uses ARM-based coprocessors in some Mac models – the MacBook Pro and iMac Pro – to handle specific functions like security optimally.
Apple has been using Intel processors in Macs since 2006, when it switched from the PowerPC platform, and in 2017, the company accounted for about 5% of Intel’s sales, though only 1% of its profits, according to Bloomberg.
Apple is also continuing to bring its two main product families closer together on the software side. The user interfaces for the Mac and iPhones have become more similar and they have also recently started sharing some lower level capabilities such as file management. Apple has an internal software project codenamed Marzipan, which is expected to be released later this year and will allow iOS applications to run on Macs.
Even if Intel manages to keep the modem slots for a while – perhaps because Apple first uses the rumored SoC to sideline its new enemy, Qualcomm, further – the loss of the Mac processor would be a psychological blow. This would come at a time when Intel is still battling to make up for the decline of the PC sector, once its mainstay, as well as ARM attacks on its core server business. And it would largely negate the revenue impact of winning some modem deals.
It will also be interesting to see whether Apple is actually able to design a modem for its inhouse SoC. The modem is a challenging component to create from scratch (though the same was said about the GPU, and Apple replaced Imagination more quickly than most thought possible). However, Qualcomm tends to say its modem expertise is, above all, what brings it success, and it has certainly been the hardest aspect of the San Diego company’s platform for others to challenge. Many have tried and failed, including Qualcomm’s recent suitor Broadcom, and Nvidia. Others, like MediaTek, are in the market but have lost share as Qualcomm has surged ahead with very advanced products for LTE-Advanced and 5G.
If Apple does go inhouse, rather than buying modems from a new supplier, it will take yet another major customer out of the market. The other top three smartphone makers – Samsung and Huawei – both have significant internal modem and SoC products.