Apple’s Failed Car Project May Shape Future AI Chips

Apple’s self-driving car project was regarded for years as one of Apple’s major mistakes, a decade long effort of private research and development and billions of dollars apparently lost but no vehicle was produced. New reports suggest that the AI foundation was not abandoned when the project died, it’s turning up in Apple’s next Mac chips: the new M7 and M8.

From the Road to the Chip Lab

When Apple officially shut down its car venture after about ten years of stealth efforts, the decision was ridiculed as a high priced flop. But according to the report, Apple did not abandon any of these plans, it just folded the automotive engineering group directly into its AI division. The team’s own self-driving software required huge amounts of real time machine intelligence, the kind of computing horsepower which turns out to carry over well from a car’s onboard systems to the chip in a laptop or phone..

Self-driving cars were previously identified by Apple executive Tim Cook as one of the most critical projects that the company has ever embarked on, and even publicly acknowledged self-driving cars as part of that vision as far back as 2017, a bold step from Apple to comment publicly on unreleased projects. So it was that early emphasis on autonomy that seems to have sprouted into the M7 generation.

AI Over Raw Speed

The key thing about the M7 and M8 is not their speed and improved efficiency, it is a reported shift in Apple’s chip strategy on workloads altogether, focusing on AI in particular. Instead of focusing only on traditional upgrades, these chips are designed for specific machine learning applications that are hard on current silicon, such as on device language and image processing.

According to Bloomberg’s Mark Gurman report, Apple had been preparing substantial improvements to the neural processing in the M7 class and decided those improvements were significant enough to justify moving up the entire next generation rather than finishing out the M6 lineup with Pro, Max and Ultra variants. That’s an interesting deviation from Apple’s traditional chip release schedule and highlights the increasing importance of AI in the Apple strategy.

The scale of the shift shows up in the numbers. The M7 will apparently have 240GBs worth of unified memory bandwidth, which will mark an enormous jump from the 153GBs of the M5. However, the jump will be even greater in the case of the M7 Ultra, which Mark Gurman of Bloomberg allegedly says will be a massive boost in AI capabilities for Apple’s silicon architecture, putting it closer to AI accelerators such as Nvidia’s Blackwell. There are also rumors that it will have up to 1.5TB RAM.

The M7 isn’t the end of the road for Apple, though, as the company has already said to be developing the M8, with even more AI-powered capabilities and plans to develop another processor codenamed Soko expected around 2028, reportedly built on TSMC’s upcoming 1.4nm manufacturing process alongside another chip effort referred to internally as Cardinal.

Why This Matters

The relationship between self-driving car projects and M7 ultra chips is not merely accidental in terms of time frame of development. The group of engineers working on self-driving cars’ ability to sense the environment, make split-second decisions and perform energy efficient parallel processing looks strikingly similar to the engineering required for running advanced AI locally in Mac or iPhone. Namely, the exact same problem solving capabilities which allow the car to detect the road now help the chip understand the language and images and contextualize them.

It’s a reminder that despite the high profile public announcement of the cancellation of a flagship project, Apple does not abandon R&D altogether. It’s possible the car never even got onto driveways, but the engineering DNA behind it is about to live inside millions of pockets and backpacks.

What Comes Next

Based on the timeline, the M7 chip is scheduled to come in the first half of 2027, M7 Pro and M7 Max are expected to be available by the end of that year, while M7 Ultra will be offered in 2028 alongside the first M8 chips. Whether this is an actual measly year over year bump remains to be seen, but if it does prove out, the buyers shouldn’t be surprised at Apple’s entire silicon course correction, and the ghost of Project Titan is doing much of the work behind the scenes.

TheTweaks Verdict

Apple’s decade-long self-driving car project was never actually about creating a vehicle. It was always an excuse to solve the most difficult problems in real time AI and that R&D does not just disappear because the project got cancelled. If these reports are accurate, the M7 and the M8 will not only be faster Macs but the first tangible dividend of a decade of “wasted” money that critics gave up too early. 

Our advice: Do not judge the next MacBook by its clock speed. Judge it by the amount of AI capabilities it can power without ever touching the cloud. It’s there that Apple’s true gamble is going down.