The power of the chip

Look at your little fingernail, and try to imagine it covered in billions of tiny circuits. Then try to imagine each of those circuits – known as transistors – endlessly opening and closing at blistering speed. This is the mindboggling technological feat that powers your mobile phone, your computer, your car, and even your country’s national security systems. Welcome to the world of semiconductors.

“All digital computing, all software, all data storage are just billions and billions of ones and zeros [generated by] these tiny chips,” Professor Chris Miller, a historian at Tufts University and author of the award-winning Chip War said in a recent Found In Conversation podcast. “And each one of those transistors in your phone, for example, is smaller than the size of a coronavirus. They're measured in nanometres or billionths of a metre.”

The first ever chip had four of these transistors, Miller explains. But on a new iPhone, “just the primary chip will have 15 billion transistors. From four to 15 billion has been the rate of technological progress, and it's a rate that's unparalleled in any other sector of the economy.”

Much of this progress is owed to efficient global supply chains and cooperation between major companies in different corners of the globe. Yet that could all now be threatened by geopolitical tensions as governments are increasingly seeking to develop domestic supplies.

Unsurprisingly, something so advanced is also devilishly complex – and expensive – to manufacture. Each stage of the process requires very sophisticated tools, with the best ones often made by only one or two companies globally.

Take lithography, which uses huge machines to beam ultraviolet light on to silicon in order to carve out the microscopic transistor circuits. The most advanced lithography machines weigh over 200 tonnes and can cost as much USD200-250 million. That creates pretty hefty barriers to entry, and more than 90 per cent of this market is controlled by Dutch company ASML.

Similar concentration is seen in other parts of the chip production chain.  The devices that deposit thin film onto the silicon are almost exclusively manufactured by the US’s Applied Materials. And because these tools and machines need to be assembled together in one place for the chips to be produced, chip manufacture can only take place in large, expensive foundries that only companies with economies of scale can afford to build and maintain. Which is why chip manufacture is dominated by one company: Taiwan Semiconductor Manufacturing Co Ltd (TSMC).

"Today, there are only three firms that are anywhere close to producing cutting edge processor chips, there's TSMC in the lead, Samsung in South Korea behind them, and then Intel of the US about two generations [behind],” says Miller. “These are the only three companies that are anywhere close. There are no new entrants in this business, nor is there any risk of new entrants.”

He estimates that a single new facility could cost around USD20 billion. With technology moving so fast, it would only be able to produce the most advanced chips for up to three years before it’s overtaken by the next round of innovations.

Geopolitical hurdles

As semiconductors are vital to our every day, and as they are used by governments – including in the military – the dominance of certain companies and countries in the manufacturing process causes problems. Governments don’t want to be beholden to each other for chips, nor subject to supply chain issues (such as those seen during the Covid pandemic, when chip shortages nearly stalled the auto industry).

“Because of the role of chips in AI, in particular, governments are focused on access to the most advanced chips and cutting off the rivals from getting access – partly for commercial reasons, but more importantly for the military and intelligence ramifications,” says Miller. “If you think of the ways that AI is going to transform the way you compute, it's going to have even more dramatic ramifications for the ways that intelligence agencies and militaries compute and that that's why we've seen a lot more political interference in chip supply chains in recent years.”

Alert to these problems, governments across many of the world’s largest economies have pledged funds and incentives to promote chip manufacture. The European Chips Act aims to mobilise over EUR43 billion in public and private investment into the sector by 2030, while the US has unveiled USD52 billion in subsidies for semiconductor production and research and China has made expansion of domestic production a key pillar of its “Made in China 2025” strategy.

In many cases, however, the extra investment has been accompanied by more hostile measures. The US is restricting advanced semiconductor exports to China and Beijing has launched a security review into US-based chipmaker Micron Technologies. Geopolitical tensions are thus a major concern. But ultimately, the industry thrives because it is a global business built on international cooperation.

“A third of new processing power each year comes from Taiwan, if that were to disappear, good luck making not just a smartphone or PC, but a car or an airplane, or even dishwashers, which don't require sophisticated chips but do require a lot of chips,” says Miller. “[But] TSMC only functions thanks to imports of materials from Japan, the US and Europe, spare parts for the machines come from abroad and the energy required to operate the machines is extraordinary.”

Miller expects appetite for chips to grow over the coming years, led by AI, auto and datacentre sectors. To meet that demand and continue to deliver innovation, governments and businesses will have to work together:

“There's a balancing act to be struck by government funding for the basic science … and companies scaling it up.”

If you would like to hear from more experts on understanding the modern world, listen to the Found in Conversation Podcast.