Geopolitical competition is increasingly shaping the global economy. And nowhere is this competition more intense than in the case of computer chips.
In October of 2022, President Biden launched America’s “chip war,” imposing restrictions on China’s access to advanced semiconductors and the equipment needed to make them. The goal was to slow China’s progress in artificial intelligence, which the Biden administration had identified as a critical strategic asset.
If anything, developments in artificial intelligence, or at least increasing levels of enthusiasm for the promise of AI among both big tech firms and investors, have only boosted the prominence of advanced chips and semiconductor technology as a focus for strategic competition.
Whilst Donald Trump’s administration has broadened the United States’ trade war with China, there is a strong vein of continuity dating back to President Biden’s agenda. Even if steps such as taking a stake in Intel, presumably to serve as some sort of US national chips champion, go a great deal further.
Bloomberg recently set out the stakes in play.
Most of the world’s leading semiconductor technology originates in the US, but today it’s Taiwan and South Korea that dominate chip manufacturing. China is the biggest market for the electronic components and has a growing desire to make its own chips.
The US is deploying export controls and import tariffs to contain Beijing’s ambitions. Leading Chinese tech companies including Huawei have been placed on a so-called US entity list, meaning American chip technology suppliers must get government approval to sell to these blacklisted companies. National security concerns are often evoked to justify the US restrictions, along with accusations that China doesn’t compete fairly in trade.
Biden’s government set aside huge sums of federal money to bring home physical production of the components and reduce what it saw as a dangerous reliance on a few facilities in East Asia. Trump has said he doesn’t like the initiative, and his administration has slowed disbursement of the funds. Trump seems to prefer to rely more on tariffs to push foreign chipmakers to produce semiconductors on US soil.
The experts were first asked the really big question. Whether ‘US export controls on advanced semiconductor technology and equipment will contribute substantially to maintaining US technological dominance in the industry over the next ten years?’
This is, obviously, a tough one to answer, and, weighted by confidence, 57% of respondents were uncertain. Of those with a firmer direction view, disagreement was more common than agreement, but the issue is clearly not straightforward. As Ivan Werning of MIT put it, ‘In the short term, probably yes. Over 10 years having substantial impact is much less certain, especially due to the response and investment by China’.
Pinelopi Goldberg of Yale, one of the experts expressing agreement, noted that her own research had led to her changing her own view, ‘I used to be skeptical, but my own work on semiconductors has convinced me that past US export controls and restrictions on Chinese investments have been successful in slowing down China’.
There was more certainty on display when the experts were asked whether ‘US export controls on advanced semiconductor technology and equipment will substantially raise China’s presence in the industry over the next ten years’?
This time around, and again weighted by confidence, 5% strongly agreed, 58% agreed, 33% uncertain, and just 4% disagreed.
David Autor of MIT argued that ‘This has already begun and there’s no turning back. China’s new restrictions on Nvidia chips demonstrate that China has firmly decided to break free of U.S. technology dependency. China will pay a steep cost in the short run, harvest a huge strategic asset in the longer run’.
It does seem likely that if access to US technology is withdrawn, China will be incentivized to develop its own alternatives. One of Mancur Olson’s many crucial insights in understanding the economics of war and statecraft is that depriving a foe or potential foe of a so-called ‘strategic commodity’ rarely has achieved as much as is sometimes hoped. In reality, economies have strategic needs, but there is usually more than one way to meet them. For most goods, some form of substitute either exists or can be developed.
Recent reporting by the Economist from China suggests such adaptation is already well underway.
At a huge AI conference in Shanghai last month, the boss of a leading Chinese model-maker, MiniMax, predicted inference costs would continue to fall as they have done, by orders of magnitude. Mr Jiang of Yizhi Intelligence says that although Chinese AI development is frustrated by being denied the best foreign models (such as Gemini and GPT), domestic open-source models are improving rapidly.
Finally, the panel was asked whether, ‘In ten years, historians will judge that the US’s current use of sanctions, export restrictions and tariffs in critical sectors substantially improved the median American citizen’s welfare’?
Weighted, as ever, by confidence, 66% of respondents disagreed, and 14% strongly disagreed.
As many panellists argued, it is tricky to draw any sort of direct line from semiconductor exports to the median American’s welfare. And as Anil Kashyap of Chicago Booth put it, ‘It seems at least as likely that the disengagement and fracturing of trading relations makes us worse off’.
The experts then seem reasonably confident that the chip war and US export restrictions will spur Chinese domestic chip production, but on the biggest question – that of whether restrictions can cement US dominance for another decade – the jury is still out.
