Apple is preparing the A18 series chips for the iPhone 16 series models to be launched this year, and one of the focuses is to upgrade the neural engine, which is rumored to be even higher than the M4 chip, which is crucial for handling generative AI functions. If the iPhone 16 series models run AI models locally, the stronger neural network engine can better adapt to various types of AI tasks.
According to Wccftech, the iPhone 16 series is about to be delivered, and the full range of products is expected to use TSMC's second-generation 3nm technology, that is, the N3E process. According to supply chain sources, Apple expects shipments of the iPhone 16 series to be between 90 million and 100 million units, believing that the main driving force comes from generative AI and other upgrades to make new devices more attractive, and in order to better respond to the market's upgrade needs, Apple has increased the order size of the A18 series chips.
It is reported that Nvidia has accepted TSMC's offer of about 10% price increase for next year's 4nm order, while the price of Apple's 3nm order next year remains unchanged, mainly because the two sides continue to cooperate in Angstrom chips, TSMC gives price concessions, and the order price of other customers may not be determined until the end of the third quarter of this year.
It is understood that the A18 series chips may be divided into two models, one is the A18 for iPhone 16 and iPhone 16 Plus, and the other is the A18 Pro for iPhone 16 Pro and iPhone 16 Pro Max, which will be more powerful. Leverage TSMC's upgraded semiconductor process to bring better energy efficiency and performance to chips.
Sources say that the entire iPhone 16 series comes with 8GB of RAM, which may not be enough to run generative AI functions entirely on the device. The A18 series chips may be larger to accommodate the upgrade of the neural engine and the increase in cache, but the GPU will still maintain the same 6-core configuration as the existing A17 Pro, which limits the improvement of graphics performance to a certain extent.