CUPERTINO, July 11 — The technology at the heart of a widening legal battle between Apple and OpenAI traces its roots to a single chip released nearly a decade ago. Apple’s custom silicon, known collectively as Apple silicon, powers everything from the iPhone in a person’s pocket to the MacBook on their desk. The A-series chips first arrived with the iPhone 4 in 2010.
But the real breakthrough for artificial intelligence came in September 2017, when Apple released the A11 Bionic chip. It included the company’s first Neural Engine, capable of 600 billion operations per second.
That dedicated hardware, purpose-built for machine learning, set Apple on a path that would accelerate rapidly. By November 2020, Apple had introduced the M1 chip for Mac computers. Its 16-core Neural Engine could handle 11 trillion operations per second — a leap that put serious AI processing power into the hands of everyday users.
A history of AI on the device
The company’s work on artificial intelligence did not start with hardware alone. Siri, the voice-activated virtual assistant, launched on October 4, 2011, with the iPhone 4S. From the beginning, it relied on natural language processing and machine learning to understand and respond to users.
In June 2017, Apple introduced Core ML, a machine learning framework designed to let developers run AI models directly on Apple devices. That same year, the company also released Create ML, a tool that allows users to train machine learning models on their own Macs and iPhones.
Together, these tools and chips formed what the company calls a complete on-device AI pipeline. The lawsuit now before the courts alleges that OpenAI used stolen trade secrets related to these very technologies to develop its own AI gadgets. Apple’s trade secrets, according to the complaint, likely include proprietary algorithms, chip architectures, and software optimizations for AI inference — the process of running a trained model to make predictions or generate responses.
What is at stake
OpenAI’s hardware efforts remain largely undisclosed, with no specific product names made public. The company has signaled potential integration with robotics and edge devices, but the details are sparse. What is clear is that Apple sees its silicon as a crown jewel worth defending.
The M1 chip’s Neural Engine, capable of 11 trillion operations per second, represents years of engineering refinement. The A11 Bionic’s original Neural Engine, at 600 billion operations per second, was already a milestone.
The jump between those two figures shows how quickly Apple has pushed the technology forward. For users, the practical impact has been steady. Siri gets faster and more accurate.
Photos identify faces and objects without sending data to the cloud. Keyboards predict text with increasing precision.
All of this happens on the device itself, powered by the neural engines built into Apple silicon. The court will now have to determine whether OpenAI crossed a line. The answer could shape how both companies — and the broader industry — approach the next generation of AI hardware.
