America is leading the way on frontier AI development, but a major infrastructure bottleneck means that China could pull ahead even if its models are less capable.
Citadel Securities says there’s a risk that the US gets tripped up by constraints in its own power grid, with the energy bottleneck emerging as an obstacle right as China is poised to massively expand domestic generating capacity.
Nohshad Shah, head of EMEA fixed income sales for Citadel Securities, wrote on September 26 that, in general, while China might have less capable models, more robust physical infrastructure could be the secret to its success against the US.
“China does not need the world’s best model in every domain if it can combine a slightly less capable (and much cheaper) one with more factories, robots, vehicles, drones, and industrial equipment,” Shah wrote in a blog post.
He highlighted the difference between two key terms that may determine the AI race’s overall winner: development and deployment. The former is driven by the labs building the models and the latter hinges on a country’s ability to put AI to work throughout the broader economy.
Energy is the big bottleneck. Shah said that the possibility is looming that the US lacks the power supply to support a rapid expansion of AI capabilities.
Local opposition to data centers has prompted some Wall Street analysts to warn of risks to their AI outlooks, particularly as politicians double down on advocating for stricter regulation ahead of the November midterm elections.
All the while, China continues to grow its power supply.
“China will add almost six times as much power-generation capacity as the US over the next five years, whilst America remains constrained by grid connections, permitting, and local opposition to data centres,” he wrote. “That creates an uncomfortable possibility: America develops the better models…but China has more places to run it.”
Shah predicts that the US can still come out ahead, but to do so would require more political support and looser regulation. He made a similar case in August, arguing that the next big AI bottleneck would be permission, rather than any physical asset.
He added that the recent AI doomsday warnings are not only an overreaction, but they could end up holding the US back from in the race with China.
“The mistake is to tell the public that AI may eliminate their jobs (or indeed humanity itself!) and then ask the same public to provide the land, electricity and permits required to build it,” Shah said. “The industry has spent several years making the strongest possible case for why AI is powerful…and the weakest possible case for why ordinary people should want it.”