A deep dive into Apple’s most powerful mobile SoC yet
Before we dive deep, or at least as deep as we can, into the powerful new A18 chip that powers Apple’s powerful new iPhone 16 models, do bear in mind that the phone has only just launched and that there simply isn’t enough data to back up many of Apple’s many claims. That said, if we are to take Apple’s word on the updates – and we don’t see why we shouldn’t – the A18 is a significant upgrade over the A16 in the iPhone 15 and does, by itself, justify an upgrade to the newer iPhone.
Table of Contents
Performance claims
Overall, Apple claims that the A18 chip is up to 30 per cent faster than the A16 Bionic in the iPhone 15. Additionally, these chips can also deliver the same CPU and GPU performance as the A16 while consuming 30-35 per cent less power while doing so. This is in part thanks to a new manufacturing process as well as the efficiency improvements baked into the new chip architecture. We’ll go over those in more detail shortly.
Lastly, there’s now a much faster NPU. By some estimates, it’s about 58 per cent faster than the previous generation model, a necessary upgrade to support upcoming Apple Intelligence features.
Cores and clocks
In terms of core counts, the A18 and A16 have the same number of CPU, GPU, and NPU cores. We don’t have official confirmation of the clock speed of the A18’s cores, but they’re expected to be slightly faster than those on the A16.
As with the A17 Pro, the A18 Pro gets an additional GPU core for up to 20 per cent more graphics performance over the A18. Also, the iPhone 16 now gets 8GB of RAM, upgraded from the 6GB we saw on the iPhone 15.
Ray-tracing and mesh shader support
The biggest update to the GPU – for users coming from the iPhone 15 – is that it now supports hardware accelerated ray-tracing and mesh shaders. These updates dramatically improve visual fidelity and performance, resulting in a 40 per cent overall performance uplift for users coming from the previous phone.
With these updates, the A18-powered iPhone 16 is now a gaming powerhouse and should be able to run demanding AAA games like Death Stranding and Resident Evil without any issues. These games were previously restricted to only the Pro model iPhones.
New manufacturing process
The A16 Bionic was built on a TSMC 5nm process while the new A18 is built on a ‘second generation’ 3nm node. What do these numbers mean? The manufacturing process or node generally refers to the width of the smallest transistor on the chipset. The size of the transistor directly impacts the power consumption and heat output of the chip as the electricity has a larger distance to travel. Think of it this way: a longer wire has higher resistance and thus gets hotter.
Since the transistors on the A18 are smaller, the A18 can offer the same performance at a much lower thermal and power cost. This tracks with Apple’s claims about the performance improvements with A18.
New NPU
The NPU or neural processing unit is a dedicated processor like a CPU or GPU that’s designed for handling AI tasks. NPU performance is measured in TOPS (trillions of operations per second).
The A16 Bionic was rated at 17 TOPS while the new A18 chip is estimated to offer 35 TOPS of performance. Given that most of the new Apple Intelligence or AI features in the iPhone 16 are meant to run locally, i.e. on the iPhone, it makes sense that Apple has upgraded the NPU in this manner.
Pro features
The A18 Pro chip, present only in the iPhone 16 Pro models, now includes a 2x faster video engine for significantly faster video processing. The chip also has support for USB 3.2 Gen 2 (the A18 is still restricted to USB 2.0 speed) for even faster 10 Gbps data transfer.
The A18 chipset coming with the new iPhone 16 is a marked change from previous Apple upgrades in that the base model iPhones are now at near parity with the Pro models. Previously, the Pro models would be a generation ahead of the regular iPhone models.
The biggest upgrades have appeared in the GPU and NPU department, likely owing to Apple’s focus on Apple Intelligence features that require significantly more powerful hardware to run.
