Monday, February 6, 2023

Chip War

Over the last several decades, silicon computer chips have shifted from being parts used by fringe, nerdy, hobbyist groups to becoming integrated into almost everything we do. This is even more the case in the military. Where once bombs had to be launched by the dozens to even have a limited chance of hitting a target, today's precision-guided (thanks to onboard chips that make mammoth calculations) missiles are neccessary to win wars. Chips have become a matter of national security. Chip War makes this case to the reader, and provides details on the industry's evolution, with a focus on the countries that play a role in shaping this industry.

I thought the book was fantastic. I now understand this industry much better than I did before I read it. The history is interesting, though I have no hope of remembering any of the principle actors that pushed the leading edge (and still do so) to make chips ever smaller and thus more powerful (though paradoxically, using less actual power to do so!).

The book laments some of the loss of leadership in this industry that the US has suffered, as other countries' governments have subsidized manufacturing growth, which has eventually led to increased foreign know-how and eventual leadership in some subsectors. The best example of this is Taiwan Semiconductor (TSMC), which is the largest logic foundry in the world now, and which makes the chips used in the most cutting edge devices. While chips are still mostly designed in the US, should something happen to Taiwan (say, a Chinese takeover of some sort), there could be dire consequences for the West. Not only does the West lack the manufacturing capacity to sustain its chip requirements should TSMC fall into enemy hands, but it lacks the know-how that company has built through years of besting the competition.

On the other hand, US companies still play lead roles in the chip manufacturing process. For example, capital equipment for many chip-manufacturing processes are still provided by US companies (including to TSMC) and chip-design software is controlled by a US oligopoly. Intel is also embarking on a monster capital program to be able to better compete with TSMC in the next generation of chips.

There seem to be lots of potential for moats throughout this industry vertical, as some specific processes have consolidated into just a few companies. But the risk to these moats also seem high, as constantly changing technologies mean you can't just coast in order to maintain your moat.

Intel is a great example, as at one time it had a massive lead in revenues, providing it a lower R&D and marketing spend per unit. But now they're playing catch-up!

Government subsidies may also re-shape the industry as countries may be willing to accept losses in order to have a shot at longer-term leadership. The US may also be falling into this category; as the thesis that this industry is vital to national security grows, taxpayer dollars may be given to US companies in order to keep manufacturing technologies within US borders.

If you have any interest in this industry, Chip War is a must-read.

No comments: