A few weeks ago on the The New York Times’ Ezra Klein Show, Dan Wang, a technology analyst at Gavekal Dragonomics, outlined how China has come to dominate the solar industry, despite the U.S. — and countries in Europe — doing a lot of the research and development work behind it. China waited for the science to mature, then focused on scaling up production by building up its solar manufacturing capabilities, and Wang spelt out how this could be emblematic of a wider issue for the U.S. “I fear that it is going to happen in different industries if the U.S. doesn’t think much more about building out the labor force to have a very good manufacturing ecosystem to be able to scale up the next solar technology,” he said.
The end result in the case of solar is that currently China controls around 80 percent of the value chain, which means solar subsidies in the U.S. are likely to flow to China. At the same time, “[China] installed almost three times the volume of solar capacity between January and the end of April than in the same period in 2022, and is on track to add more panels this year than the entire total in the U.S.,” according to Bloomberg.
For anyone who believes the U.S.’s apparent lead on artificial intelligence development could be overtaken, solar is a useful demonstration of what’s possible. And there are good reasons to believe that the same thing could happen again.
One of the most high-profile predictions along these lines was made by Lee Kai-Fu (李開復), the former president of Google China and author of the book AI Superpowers: China, Silicon Valley, and the New World Order. The central contention of that book is that as AI transitions from a “discovery” phase into an “implementation” phase — as happened with solar in the past few decades — China has a number of fundamental advantages over the U.S. and others that position it to become the dominant force in what looks set to be the dominant technology of the 21st century.
Lee’s key intervention is that while the U.S. might still have an edge in having more top-level AI engineers, that advantage will ultimately be outweighed by China’s superior access to the data on which AI systems can be trained to recognize patterns. On top of its large number of citizens, China’s top apps — like WeChat (微信) — offer a wider range of services than the most high-profile U.S. apps — like Facebook or Twitter — and working with weaker privacy norms and laws are able to gather more and better information on users. Lee says that alongside China’s ruthlessly competitive entrepreneurial environment and a pro-active government overtly attempting to be a world leader on AI by 2030, this fundamental data advantage will set the country apart.
Of course, even within the framework set out above, there are grounds to disagree. Part of Wang’s point on how China came to dominate solar was that the U.S. had been complacent with its early advantage in the field. On AI, there has already been a dramatic indication that this is not the case. The U.S.’s export sanctions on advanced semiconductors heading for China were targeted at China’s AI industries and, according to Ian Hogarth, author of the annual State of AI report, they “mean China is not likely in a position to race ahead of [high-profile U.K.-based AI project] DeepMind or [U.S.-based] OpenAI.”
However, there are those who believe that in the long term semiconductors won’t hold China back, and Wang is actually one of them. On the same episode of the Ezra Klein Show he addressed the issue. “It is certainly the case that U.S. export controls can hobble a lot of China’s most advanced technology companies,” he said. “But over the longer run, does this decelerate or accelerate China’s ability to manufacture more advanced chips? I expect that it will accelerate.” Though he added that he wasn’t sure about this prediction — “I could be wrong on that” — he also provided the logic for his argument: “Before the sanctions came in, Chinese companies were uninterested in buying obviously inferior Chinese chips. Now they have no choice but to buy them and that has given them an ‘enormous push.’”
The takeaway: Just because the U.S. might want to stop China leaving it behind on AI as it did on solar, doesn’t mean that it can. And if that’s right, and so are Lee’s ideas about China’s fundamental advantages, then the AI Superpowers author’s view is that AI tends toward “winner-takes-all” because more money and more data generated by AI success in turn helps to improve the AI. In this case that “winner” would be China.
There are a lot of “ifs” here. And it should be noted that both of the thinkers cited above want to see cooperation between China and the U.S. Wang, for instance, points out that on solar China’s manufacturing power has “been good for the climate movement where the breathtaking drops in solar costs has fallen something like 97 percent over the last few decades.” Not everything is a zero sum game.
But with happy cooperation looking unlikely at the moment, it’s hard not to notice warning signs for the U.S. Last year’s State of AI Report, for instance, noted that “Chinese institutions are producing 4.5 times as many [AI] papers than American institutions since 2010, and significantly more than the U.S., India, UK, and Germany combined. Moreover, China is significantly leading in areas with implications for security and geopolitics, such as surveillance, autonomy, scene understanding, and object detection.”
Solar, then, proves China can dominate key sectors that determine the trajectory of the entire planet. And there are certainly some good reasons to believe that could happen again with AI.
Image: Photo by Google DeepMind on Unsplash
Leave a Reply