Apple, TensorFlow, and Amazement
Apple is playing chess while everyone else plays checkers.
This is an article about customers and is not technical.
Google announced that they’re making major updates to how developers understand and debug deep learning for the programming language… Swift.
Let’s go backwards
Things that are important to clarify up front:
1. Swift is a programming language developed by Apple 5 years ago to better build iOS apps. It’s open-source, and now you can build for Android with it and build with it in the cloud. Developers largely love it.
2. AI is normally done in a language called Python.
3. TensorFlow is a code library (think an encyclopedia but for code) for building deep learning algorithms. It’s built by Google.
4. Google pays Apple over $9B per year to be the default search engine on iPhone. This is not actually relevant but is a staggering statistic.
When asked about the original iPhone, Phil Schiller, Apple’s VP of Global Marketing, told the reporter that the iPhone was a beautiful device that distributed Apple’s state of the art software.
Yes, Steve Jobs invented the modern computer mouse and the Apple II was industry-defining, but Apple’s differentiator was always the software. Steve Jobs said it himself. “People who really love software need to make their own hardware.” Software comes first.
When you look at why people buy iPhones, they buy because of iOS. Integrations between their Apple Watch and iPhone and AirPods and Mac. The apps, the interface, or iMessage. Beautiful hardware is a reality, but it’s not the differentiating factor. It’s “iOS vs Android” and “Google Pixel 3 vs Samsung Galaxy Note,” not “iPhone XS vs Google Pixel 3.”
This differentiation is even easier seen in the tablet space. It’s “iPad vs Android tablet.” I bet right now you can’t even name one Android tablet. It’s just “Android tablet.” They look the same, the hardware runs similar enough, but you can’t imagine buying a tablet that isn’t an iPad. That’s because of the software.
Think people buy a Mac because it looks $1,000 better than the PC equivalent?
I think I’ve proven my point.
Apple built Swift so that people could build great apps. Great apps in Swift get deployed to iPhone which further differentiates iOS which keeps iPhone competitive vs Android, iPad industry-leading and Mac, well, Mac.
So, why would it matter to the non-AI developer that Google is building huge feature sets into its AI development software? Apple uses software to sell hardware.
Apple built Swift! Software written in Swift will run better on an Apple device than any other. Building AI in Swift will be easier on a Mac than on a PC. This is a huge deal because of the Apple customer.
If you like having state the art apps, those apps will have advanced AI, and you’ll stick with Apple because “the apps just work better on iPhone.”
If you like building state of the art apps, those apps will have advanced AI, and you’ll build with Apple products because “it’s just easier to build apps on Mac.”
Swift for TensorFlow is huge because it’s a step backwards in time for Apple. Back to where they started - with creatives. This time, the creatives are the developers, not the artists.
In order to have that feeling of the Apple product feeling more natural, you have to buy the Apple product. In locking the creatives into their network, it locks their primary customer base into their network. Those two locks create the recurring revenue stream for the not-so-distant future.
AI isn’t going away and so neither is TensorFlow, but iPhone will… sometime.
Swift for TensorFlow is the next generation furthering of what Apple has always done. Use software to sell hardware. The difference this time, is that Apple is positioning itself to creatives and to consumers, no matter the hardware.