Skip to the document
Madhuopen lab
Outcome School · Inference and serving14 min read

by Amit Shekhar · 28 June 2026

How does a Google TPU work?

How a Google TPU works. We will also see what a TPU is, why Google built it, how it is different from a CPU and a GPU, and how it makes machine learning fast.

2,766 words#ai#system-design4 recall cards

How does a Google TPU work?
Read on Outcome School ↗then come back to lock it in
Before you read, guess

What fundamental concept must be grasped before understanding how a TPU functions?

Ten seconds, a guess, then read — a wrong guess still makes the answer stick.

What this article covers

  1. What is a TPU
  2. Why Google built the TPU
  3. A quick refresher: CPU and GPU
  4. The one operation that matters most
  5. The big idea: Systolic Array
  6. How data flows through a TPU
  7. The full journey of a TPU computation
  8. Why a TPU is so fast and power efficient
  9. Where TPUs are used
  10. Limitations of a TPU

The article lives on outcomeschool.com. Read it there, then come back: the tutor in the margin has read it and will answer questions, and the questions below check what stayed.

Before you go

In one sentence, what was this chapter about?

From memory, without scrolling up. Writing it is what makes it yours; the grade is only to show you what you had.

How sure?