Artificial intelligence firm Anthropic has published a policy paper outlining two contrasting futures for the global AI landscape by 2028, warning that today's decisions on compute access could determine who leads the next generation of artificial intelligence. [1] The report, released as US President Donald Trump visited Beijing, urges Washington to tighten controls on advanced chip exports and crack down on what it calls “large-scale distillation attacks” by Chinese AI labs.
In its more than 2,000-word analysis titled “2028: Two Scenarios for Global AI Leadership,” Anthropic describes a future where advanced AI systems become a “country of geniuses in a data center”—able to perform complex intellectual tasks in science, cybersecurity, and engineering at or beyond human expert levels. Whoever controls such systems, the company argues, will wield tremendous economic, political, and military power.
Chips and Compute at the Center of the Fight
Anthropic claims that China’s AI sector remains constrained mainly by limited compute rather than talent or research capability, and says Chinese firms have stayed competitive by exploiting export-control loopholes and using distillation techniques to imitate leading US models. [2] The company previously accused Chinese AI firms DeepSeek, Moonshot AI, and MiniMax of conducting coordinated extraction campaigns against its Claude model.
The paper sketches two broad outcomes:
Scenario A: The US and allies tighten controls, block smuggling and distillation, and maintain a lead, securing democratic norms for AI.
Scenario B: Weak enforcement allows China to close the gap, leading to authoritarian influence over global AI infrastructure and governance.
A Narrowing Window for US Policy
Anthropic urges policymakers to close loopholes around chip smuggling, overseas data center access, and model extraction, while accelerating US AI adoption worldwide. It frames 2026 as a “breakaway opportunity” for American AI, warning that the window for decisive action is narrowing.