Opinion Last year, the US Army War College published a paper suggesting that the Taiwanese government might give TSMC’s chip fabs their own self-destruct systems in case China invaded. At the time, China said it had no interest in TSMC, thus defusing the Strangelove scenario. Now, It’s talking about changing its mind. It might want TSMC very much indeed.
This change of heart was signalled in May by a speech from a top Chinese economist, who proposed that should China be on the sticky end of the sort of sanctions the West has doled out to Russia, it should invade Taiwan pronto and nab TSMC’s chip-churning capability sharpish. Suddenly, the American plan is back in the news.
Cold war coming around again?
Stanley Kubrick’s 1964 black comedy film masterpiece Dr Strangelove has the quintessential Cold War plot. A mad American air base commander initiates a surprise strike on the Soviet Union in expectation of rapid victory. Not only does it not work, it triggers global destruction. Couldn’t happen in 2022, right?
The film still has fans who think otherwise, and are keen to steal its ideas. Kubrick engineered his armageddon by a massive automatic planetary self-destruct system triggered by just such a surprise attack.
Admittedly, blowing up TSMC wouldn’t immediately destroy the planet. But it would plunge the chip industry into its own nuclear winter: TSMC has more than half the global market. Given how unpleasant the current silicon supply chain crisis has been for multiple industry sectors, the consequences would be unknowable but extreme.
How bad could it be?
Students of the last great multi-region civilization collapse tend towards the idea that no one factor causes it, but multiple blows combined. Climate change, warfare, plague, say. Add in tech meltdown? We might not care for that experiment.
It remains unlikely that TSMC will be wiring up high explosives to its IoT factory control systems. The trouble is, it doesn’t need to. All chip fabs have a potent auto-destruct system fitted as standard: extreme environmental fragility.
A saboteur with access and a handful of dry dirt could bring a line down for months: with a little more planning and pinches of the right chemicals, it could be for a lot longer. That’s without knocking out key components that only one company in the world can make. Fragile is far too butch a word.
It’s not as if you can send in the People’s Liberation Army to take over. The only people who could pick up the pieces afterwards would be the people whose country you just invaded. You can’t shoot them, and you can’t tell whether fab problems are sabotage or science, so the normal coercive tactics of violent invaders won’t work. There’s even a lesson from history.
After Germany invaded the Netherlands in 1940 and took over Philips’ giant Eindhoven radio tube factory, saboteurs ensured there was never a single full day’s production.
Some of the tricks were beyond ingenious: a small cylinder of chlorine was set up to slowly leak into the rooftop air intake of a machine that flushed the radio tubes with nitrogen during production. The finished tubes tested perfectly as they came off the line, but the tiny traces of chlorine gradually contaminated the electrodes so they failed in service some time later. Imagine what that sort of subtle thinking could do to a 4nm fan.
If the Chinese government thinks it could take TSMC by force – it can’t, no matter what. It would be utterly destructive to the Chinese economy just as much as elsewhere. Yet as we know all too well, self-destructive madness is no barrier to desperate autocrats. If critical parts of the global economy are dependent on multi-billion-dollar systems that can be broken with a hard stare, that’s a vulnerability that needs to be addressed, and fast.
There’s a second twist in the Philips story. The British Chain Home radar system, on which the RAF depended, itself depended on one uniquely high performance tube, the EF50. These were provided by the main UK tube maker, Mullard. Supplies were thought safe. But while the tubes were labelled Mullard, the company couldn’t produce the design reliably. The tubes were quietly shipped in from the company’s owner, Philips. Who made them in Eindhoven?
The penny dropped just in time. A huge shipment of essential components and tooling was rushed out of Eindhoven just as the Germans poured in. Philips’ management and key staff were evacuated by battleship, and UK production started up in time to see off the Luftwaffe.
You can’t rush a chip fab anywhere. You can take the lesson that distributed production is the best defence against supply chain disruption, even by Nazis. You can therefore see encouraging the world’s most important chip makers to set up shop abroad not just as sound economics in the traditional sense, but critical to national and global security. It’s clear that this lesson has not been lost to America and the EU, both of whom are soliciting top-notch foreign fabs, even if it’s less clear that everyone sees the big picture.
Britain, alas, seems destined once again to blithely assume that it’ll all work out in the end, with no detectable economic or political strategy for attracting fabs to its green and pleasant land. The best on offer is a slow recognition that it might not want one of the few sizable silicon produciont facilities to be sold to China. That’s something. Not much, but you take what you can.
Of all the possible mega-disruptors to the global economy, the shock loss of great chunks of silicon production is among the easier to foresee and to make practical plans to prevent. It would take but a modicum of agreement among the major nations to ensure, if not full cooperation, then at least shared goals and mutual encouragement to make it happen. We can at least manage that. Can’t we? ®
It was discovered after the end of the Soviet Union that a Strangelovian automated armageddon system had indeed been constructed by Russia – Dead Hand. It remains in use.