How Apple's failed self-driving car project built its most powerful AI chips

Apple's self-driving car initiative is remembered as one of the most secretive and most talked-about projects in the company's history. After years of work, the program was officially shut down and no Apple-branded vehicle ever hit the road. But details that have recently emerged show just how critical that "failed" project turned out to be for the company's current AI hardware strategy.

In the early stages of developing the autonomous driving platform, Apple's engineers confronted a hard truth: for a vehicle to perceive and interpret its surroundings in real time, it needed on-device AI processing power far beyond what existing hardware at the time could offer. Relying on cloud-based processing carried an unacceptable latency risk for driving safety.

That need pushed Apple to develop its own custom AI processor architecture from scratch. The chip designed for the self-driving car was never completed and never reached production, but the engineering knowledge and design principles gained along the way were not wasted.

According to details in Apple senior editor Mark Gurman's latest "Power On" newsletter, that technical foundation built for the self-driving car project fed directly into what became the company's dedicated AI processing unit, the Neural Engine. Present in virtually every Apple device today, that component was built on architectural principles inherited from the earlier car project.

The Neural Engine first reached consumers with the iPhone X and the A11 Bionic chipset. At the time, its primary uses were relatively modest by today's standards: powering Face ID's facial recognition, real-time facial expression tracking for Animoji characters, and general computer vision tasks — a far cry from today's large language models and generative AI applications.

But that first step laid the groundwork for everything that followed. Apple has steadily expanded the Neural Engine's processing capacity with each new chip generation, and the component now forms the backbone of on-device AI features ranging from text suggestions to photo analysis, voice command processing and live translation.

Industry analysts say the story offers a striking lesson about Apple's approach to research and development: a project's commercial failure doesn't mean the technology it produced was worthless. Even though the self-driving car project is publicly regarded as a "failure", the value it added to the company's chip design capabilities is felt today across billions of devices.

The episode also helps explain why major technology companies are pouring huge investment not just into software, but into custom hardware, in the AI race. While most rivals have focused on cloud-based AI processing power, Apple's on-device processing strategy carves out a different competitive space, built around privacy advantages and low latency.

Experts expect the engineering legacy left by Apple's self-driving car project to keep shaping the company's future AI hardware plans. Future chip generations are expected to house even more powerful AI processing units.

In the end, a self-driving car project that never made it to the road continues to shape the technology world through an indirect but lasting legacy — concrete proof that even a seemingly failed R&D effort can lay the groundwork for one of a company's most valuable future assets.