Neural Engine — Apple's Self-Driving Spinoff

Apple spent roughly a decade and billions of dollars building a self-driving car that never shipped.

The company hired automotive engineers, acquired autonomous vehicle startups, and positioned the effort as a moonshot that would remake transportation. Then it quietly killed the project in 2024.

The story that followed was neat. It was a failed moonshot with an unexpected legacy. The chip architecture Apple developed to power self-driving inference became the Neural Engine. Now central to every iPhone, iPad, and MacBook.

Except the entire narrative depends on one unstated claim. That claim is that Apple's goal was ever actually to build and sell a consumer vehicle. This assumption crumbles if you look at what Apple actually does. The company did not invent the smartphone. It did not invent the smartwatch.

The phone was the lever for the App Store, the services ecosystem, the lock-in. The watch was the same lever for health data, biometric harvesting, and deeper integration into your day. In both cases, the hardware was subordinate to the extraction mechanism. The self-driving program was structured identically. Apple was not trying to beat Tesla.

It was building the infrastructure to process sensor data locally, to own the software stack, to control what your car's AI could and could not see. A finished vehicle was never the real target. The Neural Engine was.

If this is true, then nothing here was accidental. The program did not fail and accidentally produce something useful. It succeeded at exactly what it was built to do. It generated the architectural control Apple needed.

The question this opens is not about Apple's strategy. It is about how you evaluate your own work. Do you know what you are actually optimizing for? When you say you are pursuing something and it does not materialize, did you fail or did you succeed at something else entirely, something you never named? Most people cannot answer this honestly.