02-12-2021 | | By Robin Mitchell
Recently, Apple revealed that they will be using TSMC to fabricate their first in-house 5G modems that will be used in iPhones from 2023. Why is Apple manufacturing their own 5G modem, what details do we know about Apple’s developments, and how Apple’s in-house semiconductors trend shows where the electronics industry is heading.
For the past few years, Apple has quietly designed their own 5G modem to use in future products. Currently, Apple uses Qualcomm products for mobile connectivity (as do most major manufacturers), and the development of their own 5G modem would see Apple remove their need for Qualcomm products. Considering that Qualcomm is one of the industry’s leading designers in cellular technology, one would wonder why Apple is going to so much effort to create their own modem.
For Apple, creating their own modem has two significant advantages, with one advantage already being realised in their custom M1 processor. The first advantage is that Apple removes their dependency on Qualcomm technology, which costs Apple royalties for using its chips and requires that Apple purchases licenses for patented technologies held by Qualcomm. Qualcomm has been widely criticised for this practice of the No License, No Chip policy.
The second advantage to Apple creating their own modem is that Qualcomm products are designed to cater to a wide range of customers, and this generally means that there is wasted silicon. Wasted silicon often contributes to increased energy consumption which shortens battery life in portable devices. As such, Apple creating their own modem will only contain the essential circuits they require, which minimises energy consumption and thus increases battery life.
This is one of the most important reasons for Apple M1’s success; it has one of the industry’s greatest (if not the highest) performances per watt. As such, portable devices using the M1 last significantly longer on battery power than other mobile devices with the same performance characteristics.
So far, very little information has been released on Apple’s 5G modem developments, but what we do know is that TSMC will be the foundry to manufacture them. It has also been stated that the new modem will begin mainstream production in 2023, where Qualcomm products will only make up 20% of Apple products (i.e. Apple will produce 80% of their own modems). Furthermore, it has also been said that the new chip, called A17, will utilise TSMCs 3nm process, which provides a 30% yield and 15% power consumption improvement compared to 5nm devices.
Will Apple still be required to pay Qualcomm royalties after the acquisition? While their technology will be entirely in-house developed around Intel technology (after Apple acquired Intel’s mobile technology division for $1 billion in 2019), Qualcomm still has patent rights that are integrated into current cellular standards (this is another area of contention because such standards should be entirely open-source and free for any manufacturer to use), and as a result, they may still make money from modems.
One of the most significant driving factors behind companies producing their own custom silicon products is the need for low energy devices. Traditionally, electronics have been constructed from discrete parts individually sourced that each has its own unique function. For example, a computer would be built by connecting a CPU to memory chips, I/O controllers, and various other devices. Even though most of these parts could be from the same manufacturer, they would all be in their own IC packages.
While this construction method is suitable for large bulky devices connected to a permanent power source, it is far from ideal for creating portable devices that need to use a battery. This is because having designs spread out across multiple chips introduces additional resistance in traces and wires and unnecessary circuits inside chips that may not be needed in the design.
By combining all the major circuitry into a single chip, designers can reduce the distance between active components and remove unused features. Furthermore, intelligent energy-saving solutions can be incorporated into each subcircuit on the chip that enables and disables areas of the chip not in use.
Another major advantage to having a custom chip is that combining all the circuitry onto a single silicon chip significantly reduces the size of the final product. If many millions are being manufactured, then significant cost savings are made by reducing the size of the PCB, reducing the number of components needing to be soldered, reducing the number of separate manufactures used, and keeping the supply chain closer to the company.
As technology progresses, it increasingly makes more sense for companies to start designing their own silicon devices.