Lesson 11 — What Are You Taking Away? — Course Close

Understanding the Complex: How Do Chips Actually Work?

---

We started this course with a crisis — a global car shortage caused by a scarcity of chips that cost less than a dollar each. We end it somewhere more fundamental: with an understanding of what chips actually are, how they're made, and why their production has become one of the defining geopolitical contests of our time.

Let's take stock of what you've learned.

---

The physical reality

A chip begins as sand — silicon dioxide, one of the most common compounds on Earth. Purified to extraordinary levels, it becomes a crystal of near-perfect silicon. That silicon is then doped with phosphorus and boron, forming p-n junctions that can be switched on and off with applied voltage. Those switches — transistors — are printed onto the wafer surface using light-based lithography, layer by layer, in a process that can involve over a thousand steps.

Today's transistors are 3 nanometers at their most advanced, roughly 15 silicon atoms across in their critical dimension. A modern chip contains 19 billion of them in a thumbnail-sized area. Each switches millions of times per second, consuming fractions of a picojoule of energy per operation.

The machines that print these circuits use extreme ultraviolet light — photons generated by vaporizing tin droplets with lasers, in a plasma hotter than the surface of the sun. The only company in the world that builds these machines is ASML, in Veldhoven, Netherlands.

---

The geopolitical reality

The physical marvel of chip manufacturing rests on a supply chain of breathtaking fragility. ASML's machines go to a handful of companies, the most capable of which is TSMC, operating on an island in a contested geopolitical strait. The software to design chips comes from three American companies. The chemical precursors come from hundreds of highly specialized suppliers, each essentially irreplaceable in the short term.

This concentration emerged not by design but by the logic of specialization. Each link in the chain invested decades of expertise and billions of dollars to become irreplaceable. The result is extraordinary efficiency — and extraordinary vulnerability.

Governments around the world are now attempting to address this vulnerability, with subsidies, export controls, and industrial policy. These efforts will produce results over decades, not years.

---

The central insight

If this course has a single most important idea, it is this: chips are invisible infrastructure. Like electricity grids, water systems, and shipping lanes, they are essential to modern civilization, and we notice them primarily when they fail.

The 2021 chip shortage made that infrastructure briefly visible. The US-China chip war has made it visible again. Understanding that chips are not just consumer gadgets but the physical substrate of the digital economy — of AI, of telecommunications, of medical devices, of military systems — changes how you read the news, how you understand government policy, and how you think about technological change.

---

Where to go from here

If you're curious about the AI systems that depend so heavily on advanced chips, the How Does an LLM Work? course in this series traces the path from transistor to transformer — from the silicon substrate to the language models that have captured public imagination. It's a natural companion to what you've learned here.

The What Is AGI? course engages the question of where AI might ultimately be heading, and how the "compute race" — the scramble for chips — fits into the broader trajectory of machine intelligence. Understanding that chips are the rate-limiting resource for AI development gives that discussion a materiality it sometimes lacks.

The series also includes Quantum Computing, which examines the technology that may eventually supplement or transform conventional chip-based computing — another angle on what "after silicon" might mean.

Each of these courses explores a different facet of a world that is being fundamentally reshaped by technologies that, like chips, most people use every day without understanding. The more you know about how any of them work, the better equipped you are to think clearly about the choices — technological, political, and ethical — that these technologies present.

---

A brief reflection

We are at a peculiar moment in the history of computing. The transistor, invented in 1947, has been relentlessly miniaturized for 75 years. The pace of that miniaturization is slowing. The geopolitics around chip manufacturing are intensifying. New architectures — chiplets, neuromorphic systems, eventually quantum — are beginning to emerge.

Whether the next 75 years of computing progress will resemble the last 75 is genuinely unknown. What is certain is that the answers to that question will be worked out in fabs, in research labs, in government ministries, and in corporate boardrooms — and that the stakes are enormous.

You now have the tools to follow that story as it unfolds.

---

Reading time: approx. 7–8 minutes