Our digital civilisation, if you can call it that, runs on just two numbers – 0 and 1. The devices we call computers run on vast strings of ones and zeros. How? By having electrical currents that are either flowing or not. The tiny electronic switches that decide whether they’re on (1) or off (0) are called transistors.
Once upon a time, these were tangible objects: I remember buying one with my pocket money in the 1950s for a radio receiver I was building. But rapidly they were reduced in size, to the point where electrical circuits using them could be etched on thin wafers of silicon. Which I guess is how they came to be called silicon “chips”.
Nowadays, a chip is a grid of millions, or even billions, of these tiny switches that flip on and off to process those ones and zeros – to store them and to convert images, characters, sounds, whatever – into billions of binary digits. In the 1960s, Gordon Moore, the co-founder of Intel, an early chip manufacturer, noticed that every year the company was able to double the number of transistors it packed on to a given area of silicon. And since computing power seemed to be correlated with chip density, he formulated Moore’s law, which indicated that computing power would double every two years – a compound annual growth rate of 41% – which kind of explains why the A15 processor in my Apple iPhone (which has 15bn transistors) has vastly more computing power than the room-size IBM computer I used as a student.
Inescapably, then, computers need chips. But what that increasingly means is that nearly everything needs chips. How come? Because computers are embedded in almost every device we use. And not just in things that we regard as electronic. One of the things we learned during the pandemic was that cars and tractors need chips – simply because their engine-control units are basically small, purpose-built computers. Once Covid-19 hit car sales, semiconductor manufacturers switched their production lines to serve other – much bigger – customers. And then, as things started to return to normal in 2021, car manufacturers discovered that they had slipped to the back of the semiconductor queue – and their production lines ground to a halt. Similarly for microwave cookers, washing machines and refrigerators.
Making chips is a phenomenally demanding and expensive business. It requires enormous capital investment, fanatical levels of cleanliness and quality control and formidable amounts of expertise. At the moment, most of those ingredients are concentrated in one organisation – the Taiwan Semiconductor Manufacturing Company (TSMC), which was founded by Morris Chang in 1987 as a silicon “foundry” – that is, a contract manufacturer. In other words, it makes the chips that other organisations (Apple, Qualcomm, Broadcom, Arm and Nvidia, for example) design. And without TSMC, these corporations would have great difficulty turning their leading-edge circuit designs into products.
In the decades when the west was still high on the globalisation drug, the fact that things upon which we relied were manufactured elsewhere didn’t seem to bother us. Apple could cheerfully boast that while its phones were designed in California, their processors were made in Taiwan and they were assembled in China. And the fact that there was no real alternative to TSMC in the west likewise seemed unproblematic.
But that was then and this is now. We’re heading for cold war 2.0. Tensions between the US and China are manifest and increasing. The Americans have imposed a wide range of swingeing export controls on tech products, including a measure to cut China off from semiconductor chips made with US tools anywhere in the world. The aim is to slow down Chinese progress towards high-end chip fabrication. At the same time, there’s a genteel stampede under way to build high-end chip-fabrication plants in the US and continental Europe. TSMC is even building one in Arizona, possibly lured by more than $50bn in incentives offered by the US government. But it won’t be ready for prime time for quite a while.
Underpinning all these developments is a geopolitical nightmare. China, like the west, lacks its own high-end chip-fabrication capability and TSMC is in Taiwan, which the Beijing regime regards as part of the motherland – and just across a narrow body of water. As the Economist put it, TSMC makes 84% of the most advanced chips and “were production at TSMC to stop, so would the global electronics industry, at incalculable cost. The firm’s technology and knowhow are perhaps a decade ahead of its rivals’, and it will take many years of work before either America or China can hope to catch up.” In that editorial, the paper described Taiwan as: “The most dangerous place on Earth.” If it turns out to be, then it won’t matter how many billion transistors there are on our phones.
What I’ve been reading
Artificial flavour
Love and Loathing in the Time of ChatGPT is a clever and interesting essay by Ali Minai on the 3 Quarks Daily platform about interacting with that program.
Forging ahead
The subject of Adam Shatz’s article Beyond Borders in the London Review of Books is Adolfo Kaminsky, possibly the most accomplished forger of the 20th century.
That petrol emotion
The Gasoline Car Review is a nice satirical blogpost on buying a fossil fuel-powered vehicle – in this case a Mazda – by Geoff Greer on his website.
