Apple A-series chip

Also known as: A18 · A19

On this page 7

An Apple A-series chip is the system-on-chip (SoC) that Apple designs in-house for the iPhone and for many iPad models, integrating the CPU, GPU, Neural Engine, image signal processor, media engines and memory controller on a single die. In mobile hardware it is the reference example of vertical integration: the same company defines the silicon, the operating system and the developer frameworks that run on it.

The line began with the A4 in 2010 and has received a new generation almost every year since, with recent members such as the A18 and A19 each shipped in a standard and a Pro variant. Apple licenses the Arm instruction set architecture but designs its own cores, and the chips are manufactured by TSMC on its most advanced process nodes.

How an A-series SoC is organized

An A-series chip is organized as a set of specialized processing blocks sharing one memory system. Each block handles the work it does most efficiently, and the operating system schedules tasks across them so that the phone stays responsive without draining the battery.

The CPU follows a heterogeneous big.LITTLE-style design with a small number of performance cores and a larger group of efficiency cores. Recent generations use two performance cores and four efficiency cores, a layout Apple has kept since the A11 Bionic while widening each core and raising clock speeds. The GPU, the Neural Engine and the media blocks all read from the same unified memory, which avoids copying data between separate pools.

The following table lists the main functional blocks and the work each one takes on.

BlockPrimary roleTypical configuration
CPU clusterGeneral application and system code2 performance plus 4 efficiency cores
GPUGraphics, games and compute shaders5 cores (standard) or 6 cores (Pro)
Neural EngineMatrix math for machine learning16 cores in recent generations
Image signal processorCamera noise reduction, HDR mergingTied to the camera pipeline
Media enginesHardware video encode and decodeHEVC, H.264, ProRes on Pro models
Secure EnclaveKeys, Face ID and payment dataIsolated coprocessor with own memory
Main functional blocks of a recent A-series chip

Milestones across A-series generations

Several A-series generations introduced features that later became industry norms in mobile SoCs. The A7 in 2013 was the first 64-bit processor in a mainstream smartphone, and the A11 Bionic in 2017 added the first Neural Engine and the first GPU designed entirely by Apple.

Later chips followed TSMC process transitions closely. The A14 Bionic was among the first consumer chips on a 5 nm-class node, and the A17 Pro moved to a 3 nm-class node in 2023 together with hardware-accelerated ray tracing and a USB 3 controller. The table below summarizes selected milestones.

ChipYearNotable change
A42010First Apple-branded SoC, iPhone 4 and first iPad
A72013First 64-bit smartphone CPU, Secure Enclave
A11 Bionic2017First Neural Engine, in-house GPU
A14 Bionic20205 nm-class node, 16-core Neural Engine
A17 Pro20233 nm-class node, hardware ray tracing
A18 and A18 Pro2024Second-generation 3 nm, tuned for on-device AI
A19 and A19 Pro2025Neural accelerators inside GPU cores on Pro
Selected A-series milestones

Exact transistor counts and clock speeds vary by generation and are not always published, so comparisons should rely on measured benchmarks rather than headline numbers.

Standard and Pro variants

Since the A16 Bionic, Apple has used product segmentation within a generation. A chip may stay in the lineup for a second year in a cheaper model, and from the A18 onward both the standard and Pro iPhone models received chips of the same generation with different GPU, memory and I/O capabilities.

The differences matter for buyers and developers because they affect sustained graphics performance, camera recording formats and data transfer. The comparison below reflects the general pattern rather than every individual model.

AttributeStandard chipPro chip
GPU core count5 cores6 cores
Memory capacityLower tierHigher tier in the same year
USB data speedUSB 2 classUSB 3 class with suitable cable
Pro video formatsLimited or absentProRes and Log recording
Thermal headroomThinner devices, earlier throttlingBetter cooling in Pro chassis
Typical differences between standard and Pro chips

How the A-series compares with other mobile SoCs

The A-series competes with chips such as Qualcomm Snapdragon, MediaTek Dimensity and Google Tensor, but the business model differs. Apple builds chips only for its own devices, so it can spend more die area on large caches and wide cores without needing to hit a price point for many phone makers.

Historically, A-series chips have led in single-core CPU performance, which governs how quickly apps launch and web pages render. Android flagships have narrowed or closed the gap in multi-core and GPU tests in several recent cycles, so leadership varies by workload and by generation.

AspectApple A-seriesMerchant SoCs such as Snapdragon
CustomersApple devices onlyMany phone manufacturers
Cellular modemSeparate chip, in-house or third partyOften integrated on the SoC
Software stackiOS and Metal, tightly tunedAndroid with vendor drivers
Design priorityWide cores, large cachesBalance of cost, power and features
A-series compared with merchant mobile SoCs

The related M-series chips for Mac and iPad Pro share the same core designs but scale up core counts and memory bandwidth, so the A-series is effectively the smallest member of one architecture family.

Limitations and common misconceptions

The most frequent misconception is that peak benchmark scores equal real-world speed. A phone has a thermal budget of only a few watts, so a chip that scores well in a short test can throttle under sustained load such as long gaming sessions or 4K video export.

  • Process node names such as 3 nm are marketing labels and do not describe a physical feature size.
  • Neural Engine TOPS figures are not comparable across vendors, because precision and test conditions differ.
  • Battery life depends on the whole system, including display, modem and software, not on the SoC alone.
  • Long software support comes partly from the chip, but iOS feature cutoffs are policy decisions as well.

Some newer features, notably on-device generative AI, depend on memory capacity as much as on the Neural Engine, which is why older chips with less RAM may be excluded even when their compute is adequate.

In simple terms

An Apple A-series chip is the brain of an iPhone. One small piece of silicon holds many parts. It has the main processor and a graphics unit. It also has a helper for artificial intelligence and parts for photos and video.

Apple makes both the chip and the software. So the two fit together, a bit like a tailored suit. Each new generation is usually faster and more efficient. Still, how a phone feels also depends on heat, battery and memory.

A practical example

A mobile game studio is tuning a 3D racing game for the iPhone. The team tests on a phone with an A18 Pro and on an older model with an A15 Bionic. On the newer phone the game holds 60 frames per second at high settings with ray-traced reflections, while the older device lacks hardware ray tracing and is set to a lower detail profile.

During a 30-minute stress test the newer phone drops from 60 to about 50 frames per second once it warms up, so the developers add a dynamic resolution mode that lowers rendering resolution slightly under heat. The result is a steady frame rate on both chips and fewer complaints about hot devices in player reviews.