Apple Silicon
Apple Silicon is the name for the computer chips that Apple has designed itself since 2020 and builds into Macs, iPads, and iPhones. The chips combine the main processor, graphics unit, and memory connection on a single component and are considered especially energy-efficient.
Every computer contains a chip that does the actual computing work. For a long time, Apple bought these chips for its Mac computers from the company Intel. Since 2020, Apple has designed them itself, and these own chips are called Apple Silicon. The best-known ones bear names like M1, M2, M3, and M4. They are found in MacBooks, in iMacs, and in iPads. Related chips with names like A17 or A18 sit in iPhones.
The break with Intel and its consequences
The switch was a turning point for the computer industry. For decades it was considered a given that PC manufacturers source their chips from Intel or AMD. Apple showed that a device maker can also design the most important component itself. Intel thereby lost one of its most visible customers.
For Apple itself, the gain lay above all in energy consumption. A MacBook Air with an M-chip needs no fan and still runs a full workday on a single battery charge. Before, this was hardly possible for a device in this performance class. The reason lies in the design: the chips originally come from the mobile phone world, where energy efficiency is the top priority.
For investors and business journalists, Apple Silicon is also an example of a larger trend. Google, Amazon, and Microsoft too now design their own chips for their data centers. Whoever controls the chip themselves is more independent of suppliers and their prices.
Everything on one piece of silicon
Apple Silicon is what is known as a system on a chip. This means: components that sit separately on the motherboard in a classic PC are all located here on a single chip. These include the main processor, the graphics unit, and the working memory. Short distances mean lower power consumption and faster data transfer.
A key trick is called Unified Memory, meaning shared memory. In a normal PC, the graphics card has its own memory, separate from the working memory. Data has to be copied back and forth between the two. With Apple Silicon, the processor and the graphics unit access the same memory. The copying is no longer necessary.
In addition, there is the Neural Engine, a component specifically for AI computations. It only handles a certain type of task, but does so very efficiently. Exactly these kinds of calculations are needed for speech recognition or AI-based image editing. The chip itself, by the way, is not manufactured by Apple but by the Taiwanese contract manufacturer TSMC. Apple designs, TSMC produces.
From the MacBook to the local AI model
Anyone who buys a new MacBook or iPad today automatically gets Apple Silicon. In product descriptions, the chip appears as M3 or M4, often with additions like Pro, Max, or Ultra. These additions stand for more computing cores and more memory, not for a different technology.
In AI news, the term comes up for a particular reason. The shared memory makes it possible to run large language models directly on one’s own Mac. Such models need a lot of memory, and with a classic graphics card this is often in short supply. A Mac with 64 or 128 gigabytes of shared memory can handle models that some expensive graphics card fails at.
A common misconception: Apple Silicon does not make Nvidia graphics cards superfluous. When training large AI models in data centers, specialized graphics cards remain clearly superior. Apple Silicon scores in running finished models on a single device, efficiently and quietly. These are two different tasks.