Silicon Crossfire

A shaky truce in the US–China chip war has exposed the fragility of handshake diplomacy in the age of semiconductors.

In the 21st century’s hierarchy of strategic resources, semiconductors have replaced oil as the most contested commodity. They power the world’s most vital systems - from MRI scanners and fighter jets to smartphones and satellites. As digitalisation and automation advance, microchips are today cornerstones of both economic security and strategic independence.

The pandemic, followed by Russia’s invasion of Ukraine, laid bare a sobering truth that the global economy is dangerously dependent on an industry concentrated in a handful of places. Taiwan alone produces more than 60 percent of the world’s semiconductors and close to 90 percent of the most advanced ones. South Korea, Japan, China and America fill most of the remaining market. Such concentration leaves supply chains perilously exposed to pandemics, earthquakes and geopolitical sabre-rattling across the Taiwan Strait.

This vulnerability has set off a worldwide scramble to build domestic chipmaking capacity. America has rolled out lavish subsidies under its CHIPS and Science Act; the European Union has pledged billions to foster a ‘chips sovereignty’ of its own; Japan and South Korea are reinforcing their foundry capabilities. India, with its growing electronics industry and geostrategic ambitions, is pitching itself as a reliable alternative in an era of fractured supply lines.

Yet even as countries court new fabs and tout industrial strategies, an older battle has re-ignited in form of the ‘Chip War’ (to use the title of Chris Miller’s bestseller on the topic) - the technological cold war between Washington and Beijing, though now in a fresh and more convoluted phase.

In May, the two sides declared what looked like a modest truce. The Biden administration agreed to lift its ban on certain Nvidia chip exports to China, including the H20 - a deliberately watered-down processor designed to skirt earlier American export controls. In return, Beijing would resume exports of rare earths, minerals vital for electronics production and over which China holds a near-monopoly. The agreement, reportedly sweetened by revenue-sharing terms that would see chipmakers give the White House a 15 percent cut of sales to China, was greeted with relief in Silicon Valley. Jensen Huang, Nvidia’s boss, flew to Beijing in July, lavishing praise on China’s technological prowess.

The truce quickly crumbled. Rather than snapping up the newly approved H20s, Chinese authorities quietly told domestic firms to avoid buying American chips altogether, Nvidia’s and AMD’s in particular. The reasons appear two-fold. First, security concerns: Chinese policymakers fear the chips could contain hidden ‘kill switches’ or tracking capabilities, a suspicion bolstered by past moves such as banning Teslas near government sites and discouraging officials from using iPhones. Second, industrial policy: Beijing has poured billions into nurturing its own semiconductor sector, with Huawei leading the charge. While Chinese firms cannot yet match Nvidia’s most sophisticated models, they can compete with the mid-range H20.

From Beijing’s vantage point, the calculation is obvious. Why import an inferior chip that could be compromised when domestic alternatives, however imperfect, would strengthen local industry? For Nvidia and AMD, this is a nasty blow. China accounts for about 13 percent of Nvidia’s revenue and nearly a quarter of AMD’s. Both firms had banked on renewed Chinese demand to justify their concessions in the May accord. No sales mean no revenue, and no reason for Beijing to honour the spirit of the deal.

For Washington, the ‘mini-deal’ that Donald Trump championed now looks less like a clever compromise and more like a concession China never actually sought. It is also, characteristically, vague.

The fallout may yet force Beijing to soften. If Huawei and its peers cannot meet China’s domestic chip demand, particularly for artificial-intelligence workloads. China may have to turn back to Nvidia and AMD. American officials are betting on exactly that, hinting that more revenue-sharing arrangements could be floated if sales resume.

This dispute also carries lessons for the broader semiconductor race. First, that technology controls are blunt instruments. Efforts to hobble Chinese AI development by restricting chip sales have spurred, not stymied, China’s industrial policy. Second, supply chains built on mutual suspicion are brittle. If Taiwan’s chip foundries are the “single point of failure” in the current system, an emerging world of bifurcated tech ecosystems (one American-led, one Chinese-led) may prove even more fragile.

Meanwhile, the world’s appetite for chips is only growing. Cars are becoming rolling computers. Power grids and hospitals depend on sensors and processors. Artificial intelligence, with its voracious data-crunching needs, will require ever more sophisticated silicon. The more essential chips become, the more tempting it will be for governments to weaponise them.

For now, the silicon crossfire will continue and the rest of the world had better be ready for the shrapnel.