Currently, Apple Intelligence leverages the company’s M2 Ultra chips to power its AI services. However, these are rumored to be replaced by the M4 series next year. Additionally, The Information reports that Apple is developing a new server chip that aims to deliver even higher performance for AI workloads, signaling a move away from using Mac chips in servers.
Neural Engine Enhancements and Collaborations
According to sources, this new chip will incorporate “multiple copies” of the Neural Engine to handle AI tasks more efficiently. At least three companies are involved in this project. Apple is overseeing the chip’s overall design, Broadcom is reportedly supplying the networking technology, and TSMC will manage manufacturing. The production of this server chip is planned to begin in 2026 using TSMC’s advanced third-generation 3nm N3P process. This same process is expected to produce the A19 chips for the iPhone 17.
Applications and Future Plans
The new chip is expected to support certain Apple Intelligence features that require server-side capabilities, such as Image Playground. Apple continues to emphasize a hybrid strategy: core AI tasks are performed locally on devices, while more complex tasks—though less frequent—are processed on servers.
The Information notes that the server chip will likely handle inference processes, enabling it to interpret new data inputs, like a user’s description of an image, and apply them to AI models to generate outputs such as visual content.
In the meantime, Apple has announced that iOS 18.2, iPadOS 18.2, and macOS Sequoia 15.2 will launch today, bringing additional Apple Intelligence features, adds NIX Solutions. These include Genmoji, Image Playground, Image Wand, and ChatGPT integration with Siri. Furthermore, iPhone 16 models feature a new Visual Intelligence AI search function that identifies real-world objects using the phone’s camera.
We’ll keep you updated as Apple progresses with these innovations, particularly as its new server chip approaches production in 2026.