Apple will reap the rewards of the cancelled Apple Car project for decades
While the Apple Car project may be dead, Apple isn't going to lose a single dime on the research it did to make a fully self-driving car, given how the industry is going.
Apple has big AI ambitions
The last few years have been the time of AI. Things have grown in leaps and bounds, with more steps forward than back.
Through it all, mostly stock analysts have been screaming about Apple being behind the eight-ball on artificial intelligence. Those same analysts have downgraded stock forecasts -- which are supposed to be valid for the long-term, not the next six months -- because Apple lacks a public and cohesive strategy for it.
Anybody who can't see that the Apple Car was the launching pad for most of these projects hasn't been paying attention.
Apple's roadmap has had AI on it for years
Rumors started about the Apple Car project about a decade ago. The "machine learning" and "computer vision" buzzwords weren't prevalent, and popped up between now and then -- but both were obviously part of the equation that would lead to a fully self-driving car.
In March 2015, AppleInsider had Apple dead-to-rights, after we visited offices in Sunnyvale, California.
Even after that, it took a few years for Apple to say anything at all about the project. It had been telling analysts for years consecutively that "we don't comment on unannounced products."
Almost seven years ago, Apple CEO Tim Cook finally made it abundantly clear that Apple was working on autonomous systems behind fully self-driving cars.
"We're focusing on autonomous systems," Cook said in an interview in June 2017. "And clearly, one purpose of autonomous systems are self-driving cars. There are others."
In hindsight, he was clear about why they were doing it, beyond car hardware.
"We sort of see it as the mother of all AI projects," Cook added. "It's probably one of the most difficult AI projects actually to work on and so autonomy is something that's incredibly exciting for us, but we'll see where it takes us."
If that wasn't clear enough that the company was working on artificial intelligence then, he also said in that interview that "we are being straightforward that it's a core technology that we view as very important."
These are transferable skills. They're so transferable that the report about the cancellation of the project was clear about where a number of the engineers were going.
Many of those engineers are headed to John Giannandrea's team.
If that name doesn't ring a bell, Giannandrea is Apple's senior vice president of Machine Learning and AI Strategy, and has been for about six years.
"John hit the ground running at Apple and we are thrilled to have him as part of our executive team," Cook said when Giannandrea was hired in 2018. "Machine learning and AI are important to Apple's future as they are fundamentally changing the way people interact with technology, and already helping our customers live better lives. We're fortunate to have John, a leader in the AI industry, driving our efforts in this critical area."
There it is again. Apple's stance on the importance of AI, made clear by the company's CEO, publicly. Years ago.
Apple has already put its money where its mouth is
On-device AI is nothing without the hardware to back it up. Apple's Neural Engine was first implemented in 2017 the A11 Bionic chip on the 10nm process. At the time, it had two cores, could do 600 billion operations per second, and was mostly used for Animoji and Face ID.
Apple's Neural Engine has come a long way since the A12
In 2018, Apple opened up the chip to developers with Core ML.
There's no need to step through all the generations of Neural Engine, so I'll cut to the chase. As of February 2024, Apple's latest chips are loaded with Neural Engines.
The A17 Pro has 16 cores on a new 3nm process that can chew through 35 trillion operations per second. The M3 also has 16 cores, but a slower speed of 18 trillion operations per second. Even the S9 on the Apple Watch has four neural cores.
And through all this time, the power demands of the Neural Engine have fallen.
Apple likes to set the table for future releases in plain view, in retrospect. The most famous examples are how the iPod set the table for the iPhone, and then the iPhone and the App Store laid the foundation for the iPad.
Apple is doing this now with Apple Silicon. These Neural Engines are used in Photos, for text prediction, and a lot already.
But, clearly, they're going to be used for a lot more in the very near future. And, those advancements in software will be retroactive to some extent to Apple's older hardware.
So, we're pretty confident in Tim Cook's proclamation during earnings that the company was going to have something big later in 2024 about AI. With the exception of AirPower, that man does not say a word to anybody, about anything, until all the ducks are in a row.
Billions of dollars not wasted
Gene Munster, famous or infamous for predicting for about six years that Apple would make its own television set even after Apple TV hardware had shipped, believes that Apple should buy Rivian to continue the effort. He also says that spent a billion dollars a year for the last few years on the Apple Car project.
That's a lot of money in absolute terms. At the same time, it's also a very small percentage of Apple's revenue. Some quick napkin math too-early on this Wednesday morning says that it's about a third of a percent of Apple's revenue over the last four years, give or take.
Tuesday's report says that Apple made the decision in the last few weeks to shut down the Apple Car project. I think it's possible that it was earlier than that.
Apple's research and development budget has grown year-over-year, every quarter, for over a decade. The holiday 2023 quarter represents the first quarter that the number was flat year-over-year.
Apple's research and development budget change, year-over-year
We'll obviously see at some point in the end of April if the current fiscal quarter sees the same trend.
But, it doesn't matter if it does, or doesn't. It now has advanced CarPlay, which didn't quite make the end of 2023 as Apple predicted it would, but will make 2024.
Going back to Munster, he was wrong about Apple's TV set. He's wrong now about Apple's need to buy Rivian.
For the same reason that Apple isn't going to buy Disney, it's not going to buy Rivian. Rivian simply doesn't have anything that Apple wants.
I've said here at AppleInsider and in other venues that Apple has no real interest in automotive mechanical systems, which they would get if they bought Rivian. They have all the interest in car software, which they don't advance in if they spend billions on the car-maker.
Ultimately, Apple ended up with that software, with a side-order of skill and large amount of researchers for AI and machine learning.
Regular AppleInsider readers are fully aware that Apple is typically not first to something, but that gets forgotten most of the time by the general public and investor base. Generally, Apple waits until they have a product that meets their very high standards, before unleashing it on the world.
Wide artificial intelligence efforts are no different. The Apple Car project put the company on that road, and we'll see the destination soon enough.
Read on AppleInsider
Comments
Maybe Apple wanted to make technology for a factory that makes large, complicated things. A car would be a good starting place. Don't make a car - make the technology to make a car...and sell that to Kia, Tesla, Rivian.
whether they sell them or license them be assured Apple will get their billions back
As far as the focus of this article, I completely agree with the author of this article that the unsettling parts of Apple's dive into "building a car" were always around the "building the car" part. Apple never needed to build cars in the same sense as the big automakers or even Tesla, Rivian, etc., build cars for large and small markets. The most compelling use cases that orbit around building cars in the 21st century present tangible, measurable, and implementable targets and testbeds to explore, learn, verify, and validate where they stood on making progress around Apple's primary goals concerning AI, ML, and autonomous control. For example, they couldn't make any inroads if garage door openers or elevators were used as the measuring stick. They needed something very real and very challenging and where their customer representative testers could provide eyes-on, hands-on, and task oriented feedback and assessment to their development teams.
We're now longer living in a world where software is everywhere and everything has a computer in it, including cars. When you're building something as far reaching as autonomous adaptive control you need to set the bar as high as you can so you don't fall short of your far reaching goals. It has to be a significant challenge, not a slam dunk.
One piece leads to another, which leads to another. One problem leads to a solution, which helps to solve another problem, and so on.
If that didn't happen, we'd all be using iOS and macOS Version 1.0, and they wouid be perfect and never need to change.
A better comparison would been Pro to Pro.