Bottom line: A new policy paper analyzes two scenarios for 2028: in the first, democratic countries preserve their AI leadership through tightened export controls and faster adoption; in the second, autocracies take control. Computing power is decisive — the US must defend its technological lead.
A new analysis sheds light on intensified competition between the US and China for leadership in artificial intelligence. While democratic countries currently hold a substantial advantage in computing power, experts warn of the risks of squandering this edge.
The US and its allies find themselves at a critical moment in AI competition with China. A new paper sketches two future scenarios for 2028, when transformative AI systems are expected to be available.
In the first optimistic scenario, the US has successfully preserved its computing advantage. The government has tightened export controls, blocked China’s illegal distillation attacks, and accelerated AI adoption in democracies. In this world, democratic countries set the rules and norms for AI and can cooperate constructively with China on security issues.
The second scenario paints a bleak picture: policymakers leave loopholes open, Chinese AI labs quickly catch up to the frontier and even overtake America. Here, autocracies shape AI norms, and advanced models enable automated mass-scale repression.
The critical component in AI competition is computing power (compute), on which models are trained. With their innovative technology ecosystem, democracies currently hold a substantial advantage. Chinese AI labs under control of the Communist Party of China (CPC), however, are not far behind in model intelligence — through talent, exploiting control loopholes, and large-scale distillation attacks that illegally extract innovation from American companies.
The central message is clear: America and its allies do not have to squander their current advantage. It is about not making it easier for the CPC to catch up.
Source: www.anthropic.com