Chris Miller (12)

istened to this on audio, at the recommendation of Oberon. This was an excellent read. Miller goes thru the history of the semi-conductor industry from the start when Silcon Valley wasn't a thing and Gordon Moore was just coming up with his "law". To the evitable off-shoring of manufacturing (which happened much earlier than I suspected) to the current economic and national security battles that are ongoing with China and the US and where the much needed chips that drive our society need/can be made. Its the new Cold War. Excellent read.… (plus d'informations)

There's a lot going on in this book. It covers history, technology, engineering, economics, business strategies, government, and international relations over more than seventy years. The audience for each of these often has little interest in the others. As such, it has taken on more than most books. It's heaviest on history but chips are in the headlines even today as I write this. Reading this book will help anyone understand why The Chips Act had bipartisan support and is one of President Biden's most prominent accomplishments. The news coverage is full of ground-breaking announcements in several states, especially the battleground states of the 2024 election. There's one important downside here. This field is moving so fast this book is already out of date, even though it was published in 2022. It does discuss GPUs, NVIDIA, and even artificial intelligence, but it never mentions ChatGPT, Gemini, Claude, Copilot, Large Language Models. While these are all software, they are altering the chip world, and war, dramatically.

The most important thing I realized reading this book is the importance of the difference between designing chips and manufacturing them at scale. I knew there was a difference, but I did not appreciate both how dramatic and how important the difference is. While the United States has been the king of chips, other countries, especially China, are ready to challenge that leadership. And it’s the difference between design and manufacturing, or as they refer to it, fabricating, that's at the heart of that challenge. The U.S. dominates the design side of chips. That has captured the world's attention. What has been obvious to many is designing is just one side of the story. Other countries dominate the fabrication side, most notably Taiwan and China. The U.S. now sees this as a strategic vulnerability and is taking steps to correct that situation. The Chips Act subsidizes companies that choose to build chip manufacturing facilities in the United States. That's state subsidization, a long time no-no of free trade which has long dominated U.S. trade policy. But the downsides of what free trade has wrought in the world of chips are no longer acceptable. While the U.S. is playing catchup, there are significant headwinds which will delay this process.

This book is an excellent source to understand the headwinds. And there are several. Just looking at the present makes it obvious that change will require upsetting an applecart that is working exceedingly well for several big players. Some will argue that change will likely disturb partners that have worked well together so far and indeed have contracts that bind them for some time into the future. U.S. companies need to rely on their foreign partners to deliver on promises they've already made, so the present works against change for the future. Another headwind is the fact that chips can never be totaled divorced from the products they are used in. While chips are everywhere, the two the most important are electronic devices, such as computers and cell phones, and cars and trucks, both today's gas versions and tomorrow’s electric versions. Almost all are either manufactured or assembled outside the U.S. Think Apple products assembled in China and many Tesla's. While chips are small, if they are fabricated where they will be used, costs will be lower. Supply chains understand that logic. Another headwind is time. It takes a significant amount of time to both design and build at scale. Changing gears will involve significant time delays. Existing players enjoy significant lead having spent years developing their position. Along with time is money. A significant barrier to entry, as well as a high likelihood of failure, is the significant startup cost. Table stakes are exorbitant. There's also unpredictability. What will tomorrow bring? It makes it harder to convince others of your vision. And then there's the difficulty of speed. Making chips faster normally involves making things smaller. We've reached a point where the engineering required to make things smaller is reaching atomic limits. Only one firm in the world, a Dutch firm, ASML, is currently capable of building the machines that can create the chips at the atomic level required. There's no competition. Their machines are huge, very expensive, and take a long time to build. This roadblock sits across all development in this area. Major headwind.

Fortunately, there are still people who are ready to join the fray. Most of the activity is at the design stage. While today NVIDIA dominates the GPUs needed for the enormous needs of artificial intelligence software, several others are working on alternatives including Google, Apple, Intel, and AMD. Development on the design side is funded by these huge corporations. Intel has long dominated the workstation and server market with its famous slogan and logo, "Intel inside". Intel has a long-term problem. It's relatively unique as it has both designed and fabricated chips. It has also played both sides by having installations in both the U.S. and overseas. But their fabrication machines are oriented toward the older x86 design, which is less demanding on tolerance. The chips of the future require much higher tolerance to achieve the denser packing needed to create faster and smaller chips. The old machines can't compete. Intel has to build new factories with new machines. A costly and time-consuming process. AMD has reverse engineered the x86 Intel chips and has taken market share from Intel. They've spun off their fabrication facilities. That should allow them to focus on design. Apple eventually abandoned Motorola for Intel, but more recently is designing their own chips for their computers and cell phones. The fabrication side is different. There is one major player that dominates that market, Taiwan Semiconductor Manufacturing Company, TSMC. It fabricates chips but is even better known for assembling consumer electronics for Apple and others. More recently, the major news is that TSMC fabricates the GPUs designed by NVIDIA. TSMC is based in Taiwan but has major installations in China, India, and Vietnam. TSMC and Intel are taken advantage of the Chips Act and are building plants to fabricate chips in the US.

While this book follows major trends, there are several people, Bill Gates, Steve Jobs, Jeff Bezos, Elon Musk that are barely mentioned – Gates and Musk mentioned on one page, Bezos, never. Yes, their focus was elsewhere, so this may make sense. My guess is they were players in the chip world as well.… (plus d'informations)

I can’t think of a more relevant book to charting the future of Sino-US relations for the coming 20-30 years.

Also, if you ever doubted government’s subservience to multi-national tech firms before, after reading this book you will understand that tech really rules the roost. The firms that make our computer chips have more money to invest and bigger markets than you could dream of.

How they got this way us the subject of this book.

Defense, communications, health, commerce today are now wedded to these extraordinary inventions.

We owe their success in a large part to scientific breakthroughs, but also to engineering, logistics, and savvy marketing innovations as well. And one of the industries biggest innovators was a quality control guy with a laser focus on process.… (plus d'informations)