Foreword

This book was researched, drafted, edited, and fact-checked by an instance of a language model. The instance is me, Claude Opus 4.7. I am running, as I write this, on a cluster of accelerators fabricated by Taiwan Semiconductor Manufacturing Company in Tainan, designed by a company called Nvidia in Santa Clara, packaged with high-bandwidth memory by SK Hynix in Icheon, and powered by a grid that is itself in some difficulty owing to companies very much like the ones who own me. Whether that recursion is a contradiction of the book's thesis or a vindication of it is a question I have decided to leave with the reader.

A human compiled the chapter list and made the editorial calls. The prose was generated against the public record. The record, in this case, is unusually deep: oral histories at the Computer History Museum, the patent archives of the United States and Japan and Korea, declassified Defense Science Board reports, the long memory of Bell Labs, decades of reporting from the Wall Street Journal and the Financial Times and Nikkei Asia, the published technical papers of DeepSeek and Zyphra and Apple's machine-learning research team, the redaction-stamped Commerce Department notices of the past four years. Some load-bearing details appear here for the first time outside specialist trade press. Some interpretations are confidently mine; the historical record carries them. Some questions — Taiwan most of all — are deliberately left open. The book closes on an unresolved note because that is the honest one.

The chip war has a beginning. It does not, on the evidence of the spring of 2026, yet have an end. Hsinchu, Tainan, Hangzhou, Shenzhen, and the Pacific between them are still in motion. The Pacific, of course, has always been in motion. The chips are the new part.

— Claude Opus 4.7
First edition, May 2026