Skip to main content
Be Bitwise

ARM History & Design Philosophy

ARM Architecture Deep Dive · CPU Architectures track

3 min read

In this module

The Efficiency Champion

ARM is the most widely deployed processor architecture in history. Almost every smartphone, tablet, and smartwatch on the planet runs on an ARM-based chip. The architecture got there through a relentless focus on simplicity and power efficiency, where x86 chased complexity at any cost.

What RISC Means

RISC (Reduced Instruction Set Computing) is a design philosophy that favours a small set of simple instructions, each simple enough that a pipeline can finish roughly one per clock cycle. (Modern cores are far more complicated, with loads and divides taking many cycles, but the instructions stay regular.) Programs need more instructions to do the same work, but each one is fast and predictable. The simplicity makes the hardware much easier to optimise for speed and power.

From Acorn to Everywhere

ARM started life in 1983 at Acorn Computers in Cambridge. The original name was Acorn RISC Machine. The first chip (ARM1) came out of a tiny team and stood out for one thing above all: it used far fewer transistors than the competition.

In 1990, Acorn spun the processor team off into Advanced RISC Machines Ltd, a joint venture with Apple and VLSI Technology, and the acronym quietly changed meaning to match. (The company became ARM Holdings when it floated in 1998.) Then came the business decision that mattered most: ARM would never manufacture a chip itself. It would license its designs. That single choice reshaped the entire industry.

The Licensing Model

ARM doesn't make a single chip. Qualcomm, Apple, Samsung, MediaTek and dozens of other companies license ARM's designs, either dropping them in directly or building custom cores around the ARM instruction set. Compare that to x86, where you can count the manufacturers on one hand: Intel and AMD.

There are two kinds of ARM licence. A core licence buys a finished CPU design, such as a Cortex core, which the licensee drops into its own chip next to its GPU, modem and everything else. An architecture licence buys only the right to implement the instruction set, so the licensee designs its own core from scratch. Apple's M-series cores are built that way. Neither kind of licensee needs its own factory: the chips are made by foundries such as TSMC and Samsung.

The upshot is an enormous ecosystem. Different vendors tune their ARM implementations for different markets, from ultra-low-power IoT sensors through flagship phones to laptop-class silicon.

Why ARM Dominates Mobile

Battery-powered devices need performance per watt, not raw performance at any cost. ARM's RISC design philosophy delivers exactly that:

  • Simple instructions mean simpler decoding hardware, which means fewer transistors, which means less power.
  • Fixed-width instructions (32 bits in AArch64) make the front-end pipeline predictable and efficient.
  • Load/store architecture keeps the execution pipeline clean and regular.

These properties make ARM the natural choice when every milliwatt matters.

ARM Comes to the Desktop

For decades the conventional wisdom held: ARM owned mobile, x86 owned desktops and servers. Apple's M1 (2020) blew that wisdom apart. Built on ARM's AArch64 ISA, the M1 matched or beat Intel's best laptop chips while drinking a fraction of the power. The M2 and M3 kept the trend going, putting paid to the idea that ARM's efficiency had to come at the expense of speed.

The implications run deep. If ARM can hold its own with x86 on the desktop, the long arc of computing may bend toward architectures that prioritise efficiency over backward compatibility.