Why Semiconductor Development Matters
People have been admiring Apple’s recent stock price run—at the beginning of 2019 Apple hit a year low and has since skyrocketed 65% to its highest point in history.
Reflecting on Apple’s strategy to get here, which I wrote about in 2015, I found this gem from John Gruber when he linked to my post in Daring Fireball:
Bluetooth sucks. In my opinion it’s the single-biggest problem with Apple Watch. Would be interesting if Apple created its own better-than-Bluetooth wireless protocol. Proprietary, of course.
The smart money is on Bluetooth dying a slow and gradual death. Last month, I covered how Ultra Wideband will subsume all short range wireless interaction models, from high bandwidth transfer to short range to micro-location in a piece entitled ‘The Internet’s Next Act’.
When a new technology is adapted by Apple it’s typical that others follow.
Two weeks back at the Pixel launch event Google adapted a new term called ambient computing. The lack of UWB in Google’s new phone will keep it 2-3 years behind Apple in how peripherals will work alongside a phone in the future of computing. In other words nothing has changed. From my 2015 piece ‘On The Future of Apple and Google’:
The entire platform is not necessarily years behind iOS, but elements of it are—and that means its aggregate user experience is. Though Google is ahead in services, these weak links hold Android back. Apple devices are now optimized at the hardware level to degrees we have never seen before. The iPhone 6 has two NFC chips and two chips responsible for gyroscopic movement and device acceleration. Apple is simply years ahead when it comes to system integration
Google is still 2-3 years behind.
Ambient computing depends on peripheral devices being connected to a server (your phone) through a high bandwidth, high quality connection that can – in the future – understand situational awareness and sense direction.
As we see new peripherals emerge, whether they be smart glasses, earpods, or standalone devices connected wirelessly like watches, all will move to communicate with new chips.
In 2020 you will see Google and Samsung embrace this strategy, integrate ultra wideband, and even consumerize it in marketing materials. They have no choice.
The fascinating think around the economics of all this, is the technology can be integrated for around a dollar—comparable to the price of Bluetooth today.
Clearly Apple has been an outlier around vertical integration at the system and chip level. Intel never succeeded with that strategy because they approached it from the excellence of manufacturing and couldn’t get distribution or fit (or build compelling products) in smart watches or elsewhere.
There are exceptions, and some exemplary companies have mastered being vertically integrated down to deeper levels. For example Square has built a 20 person chip team that builds Secure Enclave ASICS and customized tablets running a different version of Linux. Although they did this after achieving product market fit in point of sale. Tesla is also destroying everyone in the car industry at the chip and system level.
The key to understanding modularity of SoCs (literal integrated system-on-chips) and systems that are comprised of many chips, is to understand and model attributes like power profiles and intelligent hand off of different radios for different jobs to be done.
These jobs don’t exist yet. Because the use cases don’t exist yet. But I would argue we have a good idea of what they may entail.
We all know how it is with software development. You can iterate. Hardware is different. You should not mass produce something when you don’t have a good idea how it will be used. But a directional idea, that might suffice.
In order for future ambient peripherals to come online, a complex series of networking enablement needs to happen, paired with software that can secure, scale, and manage new wireless interaction models. And it needs to be built and distributed into the core device (the phone) ahead of time.
John Gruber was right in 2015. Bluetooth sucked. And while it did, things were being done to advance what it should actually be. 802.15.4z Ultra Wideband is however an open standard—not proprietary. Open wins when interoperability must exist.
The 2019 version of “proprietary” is something different—it’s called a head start. As ambient computing develops – and as new interaction models emerge in tomorrow’s connected world – system-on-chips will continue on the “race to peace” perpetuated by today’s smartphone wars.