Export Ban Opens the Door: Asian Companies Advance Competing AI Models to Rival Anthropic

Export Ban Redraws the AI Map in Asia

In a scene that exposes the fragility of relying on a single provider for critical technology infrastructure, the Trump administration's decision to ban the export of Anthropic's latest models outside the United States has prompted Asian companies to seize the opportunity — launching homegrown alternatives that are now competing for leadership in the rapidly growing AI market.

Fugu: The Japanese Model Flying Without Washington's Permission

Tokyo-based Sakana AI, a company specializing in the development of efficient AI models, has launched its new model "Fugu" — named after the Japanese pufferfish — asserting that it matches the performance of Anthropic's restricted models at the same tier. Founded in 2023 by three former Google researchers — Rin Ito, Leon Jones, and David Ha — the company focuses on models optimized for the Japanese language and local cultural context, operating efficiently with small datasets.

What sets Fugu apart technically is its design as an Orchestration Model, capable of directing and communicating with several other models via open APIs. CEO David Ha described this approach as "the next frontier in AI, beyond bigger models," arguing that dependence on a single supplier for national technology infrastructure is a risk that export restrictions have made plainly visible.

Although Sakana described the timing of the launch as "mere coincidence" with the enforcement of the ban, the company wasted no time capitalizing on the moment — explicitly advertising on its website "frontier capabilities without export control risks."

That said, the company was careful to strike a balanced tone, affirming that American models "remain important for Asia." Co-founder Rin Ito, in an article published in Project Syndicate, called on the US government to prioritize maintaining its closest allies' access to these technologies, warning against turning AI into "a technology to be monopolized rather than developed collaboratively."

China Doesn't Hesitate: Cybersecurity Tools with Strategic Clout

China's response, by contrast, was more direct and pointed. Cybersecurity firm 360 unveiled two AI models:

• Tulongfeng: A model designed for automated detection of software vulnerabilities, described as a direct competitor to Anthropic's cybersecurity-specialized models.
• Yitianqian: A tool targeting the automation of cyber defense and security incident response.

The company's founder, Zhou Hongyi, went beyond a technical announcement, raising concerns about what he called "one-way transparency" — a scenario in which some actors possess sophisticated vulnerability-detection tools while others are left defenseless. This framing positions these tools as national strategic assets rather than mere commercial products.

What Does This Mean for the Future of the AI Market?

Anthropic had been on an exceptional growth trajectory, with projected annualized revenues surpassing $4.7 billion by May 2026. But the export ban — issued just two weeks before this report was written — has laid bare the company's dependence on Asian markets, even if exact figures for that share have not been disclosed.

What now seems clear is that the Asian alternative is no longer a theoretical concept. Within just weeks of the ban taking effect, two companies on opposite ends of the Asian continent emerged to fill the gap it left. More importantly, these alternative models carry a genuine competitive advantage: they are trained on the nuances of local language and cultural context that American models cannot absorb with the same efficiency.

Even if the ban is lifted in the future, a critical question remains: will Asian customers who have discovered capable and reliable local alternatives return to dependence on American vendors? The US decision may have planted the seeds of a structural shift far deeper than was ever intended.

✦ بقلم فريق دروب أيديا