Lesson 3 — The Background You Need

Understanding the Complex: How Do Chips Actually Work?

---

Before we can understand how chips are made, we need to understand three things: what silicon is and why it's special, what a transistor does, and what Moore's Law predicted about how that would change the world. None of these require mathematics. All three are accessible with the right analogies.

If you already have a background in electronics or physics, you can skim this lesson. If not, it will give you everything you need to follow lessons four through six.

---

Silicon: the element that made the digital world possible

The periodic table contains 118 elements. Most of them are useless for building electronics. A few — copper, gold — conduct electricity beautifully. Others — rubber, glass — don't conduct at all. Silicon sits in an interesting middle position: it's a semiconductor, which means it conducts electricity under some conditions and doesn't under others.

That conditionality is everything.

Silicon makes up about 28 percent of the Earth's crust, mostly in the form of silicon dioxide — what we call quartz or, in impure form, sand. But the sand on a beach is nowhere near pure enough for electronics. Chips require silicon so pure that the concentration of unwanted atoms — aluminum, iron, phosphorus — must be below one part per billion. Achieving that purity is itself a major industrial process: heating silicon dioxide with carbon, then purifying it repeatedly until only silicon atoms remain, arranged in a perfect crystal lattice.

The result is a cylinder of silicon crystal called an ingot, typically around 300 millimeters in diameter, which is then sliced into thin discs called wafers. Each wafer will eventually contain hundreds of identical chips.

---

The transistor: a switch smaller than a virus

A transistor is, at its heart, an electrically controlled switch. When voltage is applied to one of its terminals (the gate), it allows current to flow through. When that voltage is removed, the flow stops. On and off. One and zero. This is the physical foundation of binary computation.

The first transistors, built at Bell Labs in 1947 by John Bardeen, Walter Brattain, and William Shockley, were roughly the size of a human hand and were made of germanium. They replaced vacuum tubes, which performed the same on/off switching function but were large, fragile, and consumed enormous amounts of electricity.

The key innovation that enabled modern computing was miniaturization. Smaller transistors switch faster and consume less power. By the 1960s, engineers had learned to build transistors directly into silicon wafers using a process called planar fabrication — essentially drawing the transistor on the surface of the silicon rather than assembling it from separate components.

Today's chips contain transistors measured in nanometers. A nanometer is one billionth of a meter. A human hair is about 70,000 nanometers wide. A single red blood cell is about 8,000 nanometers across. A modern transistor — the kind in Apple's M-series chips or Nvidia's AI accelerators — is about 3 nanometers in its key dimension. At that scale, individual transistors are smaller than most viruses.

There are approximately 19 billion transistors in Apple's M3 chip. In a space the size of a thumbnail.

---

Moore's Law: the prediction that shaped an industry

In 1965, Gordon Moore — then an engineer at Fairchild Semiconductor, later co-founder of Intel — published a short paper in the trade journal Electronics that contained an observation and a prediction. His observation: the number of components on a chip had been doubling roughly every year since integrated circuits were introduced. His prediction: this trend would continue for at least another decade.

He was off about the pace slightly — the doubling period settled at roughly 18 to 24 months rather than 12. But the trend held for much longer than he predicted, and it became known as Moore's Law.

The practical consequences were staggering. If the number of transistors on a chip doubles every two years, and if each transistor is cheaper and faster than the last, then computing power per dollar roughly doubles every two years. This is why a smartphone today is more powerful than a supercomputer from the 1990s, while costing a fraction of the price.

Moore's Law was never a physical law — it was an observation about the pace of engineering progress, sustained by the extraordinary collective effort of the semiconductor industry. And it has been slowing. The doubling period has stretched. The physics of miniaturization are approaching hard limits — at some point, transistors made of silicon atoms cannot be made smaller, because there won't be enough atoms to constitute a device.

Whether Moore's Law is "dead" is now one of the central debates in the industry, and we'll address it in detail in Lessons 8 and 9.

---

Putting it together

Here is what you now know: Silicon is a semiconductor — a material whose electrical conductivity can be controlled. Transistors exploit this property to act as switches. Miniaturizing those switches — fitting more onto a chip, making them switch faster, consuming less power — has driven 60 years of computing progress. And the machinery required to print those switches onto silicon has become, over time, extraordinarily sophisticated.

The next three lessons dig into exactly that machinery: how transistors are formed in silicon (Lesson 4), how they are patterned onto the surface using light (Lesson 5), and why the geopolitics of where this happens has become a matter of national security (Lesson 6).

---

Next lesson: From Sand to Transistor — the step-by-step process of turning purified silicon into a working semiconductor device.

---

Reading time: approx. 8–9 minutes