Professor Chris Miller Assesses Emerging Trends in the U.S.-China Technology Competition
By Nayan Seth, MGA 2024 Candidate at The Fletcher School
The U.S.-China technology competition forms a significant part of the great power rivalry between the world’s two biggest economies. The United States export controls on advanced semiconductors, which were first imposed in October 2022 and strengthened a year later, are one of the biggest flashpoints in an increasingly tense relationship.
On January 30, 2024, Chris Miller, Associate Professor of International History and Co-Director of the Russia and Eurasia Program at The Fletcher School, joined the Fletcher Eurasia Club lunch seminar series to analyze the chip rivalry, emerging trends in the U.S.-China technology competition, and the role of artificial intelligence (AI) in the Russia-Ukraine war. Miller is the author of the award-winning book, “Chip War: The Fight for the World’s Most Critical Technology” (2022).
Highlighting the geopolitical significance of semiconductors in turbocharging AI development, Miller said that countries identify access to them as one of their core challenges, “China, for example, has set a goal of doubling computing power in its country by 50%. Over the next year, India and the UK have committed to spending billions of taxpayer dollars, building up sovereign supplies of GPUs to provide their own national AI training and ecosystems.”
At the same time, nations and companies that play an oversized role at various nodes in the current supply chains are “trying to consolidate their position of strength.”
“There’s one company in the world, Taiwan Semiconductor Manufacturing Company (TSMC), that manufactures around 90% of the most advanced chips, including basically all chips used for training AI systems.” Companies like Nvidia design almost all of the most advanced chips.
Miller focused on the strategic nature of chip-powered AI applications for intelligence and military purposes, calling it “a moment of inflection.” He cited the ongoing war in Ukraine, where both sides are attempting to “innovate very rapidly in the use of drones on the battlefield.” He also pointed out that Russia, while being “reckless,” has shown willingness to take more tech-related risks in the war.
“Russia’s doing things that are reckless, but nevertheless, willing to take risks, something that the U.S. military is not willing to undertake,” Miller said, adding that Moscow’s policy “imposes interesting challenges for Western policymakers.”
“They have to, on one hand, not lose the race, but on the other hand, act responsibly. I think it’s a complex trade-off.”
In his view, governments worldwide understand the potential of this tech advancement and are attempting to acquire it. “The race to produce computing power is partly about giving you a better Chat GPT. Governments also think it’s core to their ability to compete in intelligence collection and analysis, and in military power.”
Given the high stakes, in 2022, the United States led an international effort to “throw sand in the gears of China’s AI development” for defense applications by imposing export controls on advanced semiconductors. According to Miller, China understood the vulnerability well before the export controls were deployed and is striving to achieve a higher level of self-sufficiency in advanced chip manufacturing.
“The challenge is that it’s still somewhat behind about five years in terms of basic manufacturing capabilities. But even further behind, perhaps seven or eight years behind, in terms of manufacturing at scale. So right now, all of China’s most advanced AI systems are trained on Taiwanese-manufactured and U.S.-designed chips.”
Regarding the likely implications of various countries sparking a subsidy war to localize the semiconductor supply chains, Miller argued that countries are “afraid of each other’s subsidies.”
“People, in theory, will like a level playing field, but there’s no political strategy to get there. And no one wants to be in a situation of everyone else subsidizing when they are not subsidising, and therefore becoming more dependent over time.”
He believed that it was China that fired the first subsidy shot. “In 2014, China started subsidizing to the tune of tens of billions of dollars a year, then the Taiwanese noticed. In 2015, there was already a discourse emerging in Taiwan about the risks to Taiwan’s industry due to China’s subsidies, and then the Japanese noticed.”
On the semiconductor export controls against China, Miller said that the United States hopes to disrupt China’s gains in AI development. “The U.S. is betting that by controlling access to cutting-edge chips, it’s going to slow development of new leading-edge models, thereby slowing China’s ability to take advantage of it.” But he cautioned that while China is behind in the race, success is not guaranteed.
“This will naturally push Chinese firms to rely more on open source models and maybe fine-tune with their own data. And if your goal is dissemination as rapidly as possible, that’s actually a more effective and rapid route than trying to train as big a model.”