Apple on Tuesday said that the forthcoming top-of-the-line iPhone X smartphone will feature a chip custom-built for handling artificial intelligence workloads.
The dual-core “A11 bionic neural engine” chip can perform 600 billion operations per second, Apple executive Phil Schiller said at the inaugural launch event at the company’s new Apple Park headquarters in Cupertino, California.
The biggest thing the chip can do is enable fast face recognition for the Face ID authentication feature for unlocking and making purchases on the iPhone X.
The news isn’t a complete surprise. Bloomberg reported in May that Apple was developing an AI chip for the iPhone, but it wasn’t clear if the chip would be ready for this year’s crop of iPhones.
And there is a more general move in the industry toward customization of silicon to accommodate the needs of AI as it becomes more common in software. Alphabet’s Google has already designed two generations of chips for handling AI computing workloads in data centers. Microsoft is developing an AI chip for a future version of its HoloLens mixed reality headset.
Having the new specialized chip on board the iPhone will mean the main chip will have less work to do, which should improve battery life. Otherwise tasks like ongoing object recognition during video captured through the phone’s camera, for example, could quickly burn through the battery.
Plus, in the near future, more mobile devices than the iPhone could contain processors that are tuned for AI, if recent comments from Google’s Android chief, Dave Burke, are any indication.
“Over time, we expect to see DSPs [digital signal processors] specifically designed for neural network inference and training,” Burke said at the G I/O conference this year.
While today’s news will certainly rightfully grab headlines, Apple has previously packed iPhones with silicon that could potentially be used for AI. The iPhone 7 contains a field-programmable gate array, which Intel and Microsoft have both explored for the acceleration of AI tasks. Apple had never previously included an FPGA in an iPhone, Forbes reported.
Apple is no stranger to silicon development. The company already commands respect with its A-series processors in iOS devices, and adding a component in the phone that’s dedicated to certain types of computations isn’t such a huge departure.
In June, Apple introduced a library of tools called Core ML, which one AI practitioner described as a “dead giveaway” that Apple is preparing to launch an AI chip.