IP Core Licensing
Why most chip designers don't actually design a chip from scratch
Imagine building a car not by designing the engine, transmission and braking system from first principles, but by licensing proven, pre-engineered designs for each from specialist companies, and then focusing your own engineering effort purely on how they fit together and what makes your particular car distinctive. Most chip design in the world today works almost exactly this way.
An IP core is a pre-designed, pre-verified block of chip functionality, a processor core, a memory controller, a communications interface, that a specialist company designs once and then licenses to many different chipmakers, who each incorporate it into their own, larger chip design. The most famous example is ARM, whose processor core designs power the overwhelming majority of the world's smartphones, embedded devices and increasingly servers, licensed to companies ranging from Qualcomm to Apple to Samsung.
Licensing terms typically combine an upfront fee for the right to use a given IP core with an ongoing royalty paid per chip actually manufactured and sold, meaning the IP licensor earns recurring revenue tied directly to how successful its licensees' final products become, without itself manufacturing a single chip.
This model exists because designing a modern, competitive processor core from scratch is an enormously expensive, specialised undertaking that few companies can justify duplicating independently. Licensing a proven core lets a chip designer focus its own limited engineering resources on genuine differentiation, custom features, performance tuning, integration choices, rather than reinventing a component that dozens of other companies have already solved well.
Whenever a new chip announcement mentions an ARM-based design, or any other licensed core architecture, that detail signals the company chose to build on proven, licensed foundations rather than take on the cost and risk of designing that core capability entirely from scratch.