Asian AI labs fill the gap after Anthropic export ban

The Trump Administration imposed restrictions on Anthropic’s Mythos and Fable 5 models in mid-June 2026, blocking non-US access to the company’s frontier offerings. Within two weeks, Asian AI companies moved to fill the resulting gap.

Sakana AI, a Tokyo-based startup co-founded by former Google researchers, announced Fugu — a model it claims delivers “frontier capability without the risk of export controls.” The company positioned Fugu as a hedge rather than a direct replacement, designed for organizations that need to preserve optionality across multiple AI systems. Separately, Chinese firm 360 released Tulongfeng and Yitianzhen, a pair of AI-powered cybersecurity tools framed explicitly around preventing “one-way transparency” in AI access — a reference to the asymmetry created when US models are unavailable to non-US actors.

The broader signal is market fragmentation along geopolitical lines, a pattern that has been anticipated but is now accelerating. As the TechCrunch reporting notes, local alternatives “trained to better understand local language and nuance, are already filling the gap” — and the concern is that enterprise relationships built around US models may not fully recover once local providers establish a foothold.

For product managers working on AI-powered products distributed globally, the practical question is whether their model dependencies create regional compliance or access risk. Teams that chose US frontier models based purely on capability benchmarks now have a reason to audit that decision against geopolitical exposure. The Sakana framing — Fugu as a hedge, not a replacement — offers a model for how PMs can approach the conversation with stakeholders about AI infrastructure diversification without triggering a full model migration.