Asian startups are accelerating AI innovation as export restrictions reshape the global market.
Asian tech companies are starting to fill a vacuum left by U.S. export restrictions on leading-edge AI systems as the global artificial intelligence race enters a new phase. Weeks after tighter export controls clipped access to some of the most powerful American AI tools, fresh models from Chinese and Japanese startups were emerging. The establishment reflects how swiftly dominant regional players are reorienting to the changing geopolitical landscape in which they compete in artificial intelligence.
New AI Rivals Emerge in Asia
A new artificial intelligence system known as Tulongfeng was reportedly launched by Chinese cybersecurity company 360 Security Technology on Wednesday. It said the model is capable of going head-to-head with US-developed AI systems that are more focused on cybersecurity. Tulongfeng is designed to search for software vulnerabilities and provide support in cyber defense operations.
Sakana AI, a Tokyo-based startup, introduced this week Fugu, a frontier artificial intelligence model named after the Japanese blowfish. Fugu says the firm can compete with top-level artificial intelligence systems and also manage scheduling of tasks between multiple externally hosted models via application programming interfaces, or APIs. Both launches come as the balance of power is reshaped with export bans on advanced AI technology.
Export Controls Reshape the Artificial Intelligence Market
Recent U.S. rules cover Anthropic’s most advanced models, such as Mythos and the limited Fable 5 model. The rules have limited access to these systems for customers outside of the United States, leaving many of these customers in limbo.
Export restrictions have spurred Asian firms to hasten work on developing local substitutes, industry observers say. Firms and governments accustomed to a world dominated by American AI writing systems are busy assessing homegrown alternatives, because doing so is expected to mitigate geopolitical exposure. The curbs were introduced just weeks ago, but market reaction has been quick.
Sakana AI Says Timing Was Coincidental
A Sakana AI spokesperson said that Fugu was not postponed because of the recent U.S. export restrictions. The company says it’s worked on the model for over a year, but most of that foundational research has been available in academia.
However, Sakana admitted that the timing has shed extra light on its newest offering. In its promotional material, Fugu claims that the model can deliver sophisticated functionality without entailing the unpredictability around export restrictions. The company maintains its launch of Fugu stemmed from its own technology roadmap — not geopolitical circumstances.
Focus on Japanese Businesses and Government Agencies
Founded in 2023 by a team of Japan-based and former Google DeepMind researchers to focus on generative AI systems more efficient in terms of Japanese language, culture, and business objectives.
Fugu, which the company claims is targeted mainly at Japanese enterprises and government bodies needing tighter control over their artificial intelligence landscape. These groups would lessen their vulnerability to shifts in global policy by depending on homegrown systems. Sakana executives have stopped short of claiming that Asian markets are turning their backs on American AI providers.
U.S. Models Still Remain Important
Sakana officials continued to say that U.S.-developed artificial intelligence systems remain essential across Asia. Executives at the companies contend that the current backdrop should not be treated as a new normal in which US tech will not be used.
Instead, they frame the situation as a diversification campaign. Organizations now prefer access to multiple AI providers instead of relying solely on one company or country. This new approach represents a flexible take on nascent concentration risk in a rapidly evolving industry with global implications.
Collective Intelligence Becomes a Strategic Goal
David Hay, co-founder and Chief Executive Officer of Sakana, was among those who recently stated that orchestration models are the next milestone stop on the AI roadmap. Orchestration systems use many smaller and specialized models rather than one large model to finish complex tasks.
Ha characterized this approach as a pragmatic form of protection against interruptions in access to any one provider. Such examples, he added, were the recent restrictions on exports that illustrate how quickly access to advanced technology can be turned off. Fugu is a model created by the company that enables models to be used within this multi-model framework, allowing organisations to draw on different capabilities offered by various AI systems.
China’s 360 Takes a More Direct Approach
At the same time, although Sakana has focused on marketing its technology as a supplement to current AI ecosystems, stances in China (again around 360) seem to be more assertive. In addition to Tulongfeng, the firm also launched another artificial intelligence product for security purposes called Yitianzhen. The system aims to perform cyber defense operations automatically and supports organizations in dealing with security incidents.
Unnamed reports claim that company founder Zhou Hongyi called advanced vulnerability-detection artificial intelligence a strategic national resource. He described a world where advanced cybersecurity AI is one-way and favors only some countries over others as “one-way transparency”—and warned against such a future. The remarks highlight the increasing overlap of artificial intelligence, national security, and foreign competition.
Growing Competition Could Challenge U.S. Dominance
Anthropic has seen astonishing growth over the past couple of years, with reports claiming the company reached a revenue run rate in excess of tens of billions of dollars by as early as 2026. However, the degree to which it relies on Asian customers is still uncertain.
U.S. competition in the artificial intelligence space is long-term, with regional alternatives possible. They are often more familiar with the language, culture and laws in their own country, and they might have better relationships with customers. Even when export restrictions are lifted, domestically relevant substitutes—those that were appropriated throughout this time of uncertainty—may be used by some organizations.
Regional Innovation Accelerates
Fugu and Tulongfeng are prime examples of how the technology behind innovation is there but a major geopolitical development has accelerated its rapid rollout. Regional AI ecosystems are proliferating as governments impose restrictions and companies pursue more sovereignty over their data.
The outcome for customers is a wider choice while remaining less dependent on just a few global players. Overall, it suggests a future filled with intensifying competition in the commercial world—and may hint that leadership in AI will become less centralized geographically. The next few months will probably clarify whether these new Asian models remain niche options or emerge as seriously competitive global players in the race for frontier artificial intelligence.
About the Author
