Semiconductors, explained: why a chip the size of a fingernail runs the world · Pulse Beat

Almost every object with a battery or a plug now contains one or more chips. Cars, phones, washing machines, hospital scanners and the servers behind every app all depend on them.

Almost every object with a battery or a plug now contains one or more chips. Cars, phones, washing machines, hospital scanners and the servers behind every app all depend on them. When chip supply wobbles, whole industries stall, which is why semiconductors have become a topic of news and national policy. Here is how they work and why they are so hard to make.

What a chip is

A chip, or integrated circuit, is a tiny slab of material, usually silicon, carrying an enormous number of microscopic switches called transistors. A transistor is a switch with no moving parts: a small electrical signal decides whether current flows. Combine billions of them, wired in the right patterns, and you can store data, add numbers and run programs.

Silicon is a semiconductor, a material that conducts electricity under some conditions and not others, which can be controlled precisely by adding tiny amounts of other elements. That controllability is what makes transistors possible.

Why smaller has meant better

For decades, manufacturers have shrunk transistors. Smaller transistors switch faster, use less power and can be packed in larger numbers on the same chip. This long trend is often called Moore's law, after an observation about how the number of transistors on a chip tended to double over a couple of years. It was never a law of physics, and keeping the pace has become steadily harder and more expensive as sizes approach the scale of atoms. Progress now comes from several directions, including new transistor shapes, stacking chips and designing chips for specific jobs.

How a chip is made, in broad strokes

Making a modern chip is among the most complicated manufacturing processes ever developed.

Design. Engineers describe the circuit using specialised software, and verify it extensively, because mistakes after manufacturing are very costly.
Wafers. Ultra-pure silicon is grown into a crystal and sliced into thin round discs called wafers.
Patterning. Light, passed through a stencil called a mask, prints the circuit pattern onto a coating on the wafer. This step, photolithography, is repeated many times, layer upon layer. The most advanced machines use extreme ultraviolet light and are produced by very few companies in the world.
Etching and adding materials. Chemicals and gases remove or deposit materials to build the structures.
Testing and cutting. The wafer is tested, cut into individual chips, and packaged so it can be connected to a circuit board.

A factory for this, called a fab, costs a very large sum to build, needs extreme cleanliness, and takes years to get right. Even small contamination can ruin a batch.

Why the supply chain is so concentrated

Different companies specialise in different steps.

Design companies create the chip but may own no factory. They are often called fabless.
Foundries manufacture chips for others.
Equipment makers build the machines used in fabs, a field with a handful of leaders.
Materials suppliers provide wafers, gases and chemicals.

At the most advanced level, only a few firms can manufacture, and they depend on equipment from only a few makers. This concentration means efficiency, but also vulnerability: a disruption from a natural disaster, a shortage, or a political dispute can ripple through many industries. That is why several governments are funding new fabs and research, aiming for a more spread-out supply. These efforts take years, and their results are still uncertain.

Different kinds of chips

You will hear a range of terms.

CPU (central processing unit). A general-purpose processor, good at many tasks one after another.
GPU (graphics processing unit). Originally for graphics, built to do many simple calculations in parallel, which also suits training and running AI models.
Memory chips. Store data temporarily or permanently.
Specialised accelerators. Chips designed for a particular job, such as AI inference, video encoding or networking.
Microcontrollers. Small, cheap chips inside appliances, toys and sensors, often forgotten in the news but essential.

Not every chip needs the newest technology. Cars, for example, use many older and simpler chips, which is why a shortage of cheap parts can stop production of an expensive vehicle.

Why AI made chips a headline

Training large AI models requires an immense amount of computation, and the chips that do it efficiently are in high demand. This has made the design and supply of AI processors a competitive area and has affected which companies have the resources to build the largest systems. It has also raised questions about energy use, since large data centres consume a lot of electricity.

What this means for careers

Chips create demand for many roles beyond the obvious ones: circuit and systems designers, verification engineers, process and materials scientists, equipment and automation engineers, software for design tools, and supply-chain specialists. If you study electronics or electrical engineering, physics, materials science or computer architecture, this is a field with real depth. Even for software developers, understanding how hardware works, memory, caches and parallelism, helps write faster programs.

Reading chip news sensibly

Nanometre names are marketing to a degree. A "3 nm" label is no longer a literal measurement of a feature, and comparing across companies needs care.
Capacity announcements take years to matter. A new factory announced today produces nothing for a long time.
Shortages and gluts alternate. Demand cycles can swing from scarcity to oversupply.
Security and policy matter. Export rules and trade disputes affect who can buy what.

In short

A chip is a vast arrangement of tiny switches made through a process that pushes physics, chemistry and engineering to their limits, by a small group of specialised companies. Understanding that explains why they matter, why shortages hurt, and why countries are paying attention.