The landscape of artificial intelligence is undergoing a profound and potentially irreversible shift. In the wake of strict U.S. government export controls that have suddenly cut off access to advanced American frontier models, technology companies across Asia are rapidly moving to fill the vacuum. Within days of each other, Tokyo-based startup Sakana AI and Chinese cybersecurity giant 360 unveiled powerful new artificial intelligence models explicitly designed to match or exceed the capabilities of restricted U.S. systems.
These rapid deployments highlight a growing fracture in the global tech ecosystem. What began as localized trade and national security restrictions by Washington is rapidly catalyzing a movement toward technological self-reliance in Asia. As regional enterprises and governments confront the reality that premier U.S. infrastructure can vanish behind a regulatory wall overnight, the imperative to develop and maintain domestic capabilities has transformed from a long-term goal into an immediate operational necessity.
The Catalyst: The U.S. Ban on Anthropic’s Mythos and Fable 5
The current wave of development and anxiety was triggered approximately two weeks ago, when the Trump administration enacted a sweeping export control order restricting global access to Anthropic’s most advanced cybersecurity-focused AI models: Mythos and its more restricted counterpart, Fable 5.
Anthropic, which has experienced historic commercial growth—reportedly crossing a staggering $47 billion in run-rate revenue by May 2026 and eyeing a nearly $1 trillion valuation ahead of a planned initial public offering—found its prized asset abruptly locked down. Designed with unprecedented capabilities in identifying software vulnerabilities and orchestrating complex cyber operations, Mythos was deemed too potent for unrestricted international distribution, prompting the White House to ban its use outside the United States and restrict its availability even to traditional American allies.
While the U.S. federal government framed the restriction as a vital national security measure to prevent advanced dual-use code-analysis tools from falling into adversarial hands, the sudden cutoff sent shockwaves through international markets. For foreign enterprises and allied governments that had grown increasingly dependent on U.S.-driven artificial intelligence ecosystems, the move demonstrated the inherent vulnerability of relying on a single foreign provider for critical national infrastructure.
Sakana AI’s Fugu: A Strategic Hedge in Tokyo
Just days after the U.S. ban took effect, Tokyo-based Sakana AI launched its new frontier model, Fugu—named after the Japanese word for blowfish. The company asserted that its model stands shoulder-to-shoulder with leading U.S. systems like Anthropic’s Fugu 5 and Mythos Preview.
Founded in 2023 by Google alumni David Ha and Llion Jones alongside former Mercari and Stability AI executive Ren Ito, Sakana AI has carved out a distinct niche by building cost-effective generative models optimized for smaller datasets, local languages, and cultural nuances. However, Fugu represents a significant evolution beyond traditional regional models. It is specifically engineered for multi-agent orchestration, possessing the ability to seamlessly coordinate access to various other models through their application programming interfaces (APIs).
A spokesperson for Sakana AI maintained that the timing of the release was entirely coincidental, noting that the underlying research had been in development since the previous year and was presented at the International Conference on Learning Representations (ICLR) earlier this spring. Nevertheless, the company has not shied away from capitalizing on the moment, utilizing its platform to advertise "frontier capability without the risk of export controls."
Behind the launch lies a deliberate philosophy regarding the future of artificial intelligence architecture. David Ha, co-founder and CEO of Sakana AI, took to social media to argue that orchestration models represent the next frontier of development, far surpassing the mere pursuit of larger singular models.
"Access to top models can disappear overnight," Ha wrote, emphasizing that collective intelligence serves as the ultimate practical hedge against the excessive concentration of power in a single jurisdiction.
Despite positioning Fugu as a crucial strategic alternative, Sakana’s leadership has been careful not to frame the launch as a permanent decoupling from American technology. Company representatives have emphasized that U.S. models remain vital to the Asian enterprise market. Ren Ito, co-founder of Sakana, addressed these themes directly at the G7 summit in Evian, where AI access and export controls dominated diplomatic discussions. In a subsequent op-ed published by Project Syndicate, Ito urged Washington to prioritize preserving access for its closest international partners, arguing that advanced artificial intelligence should be developed collaboratively rather than hoarded as a tool of geopolitical leverage.
China’s 360 Unveils Tulongfeng and Yitianzhen
While Tokyo-based firms are positioning their new tools as strategic hedges to preserve optionality, domestic players in Beijing are viewing the current regulatory environment through an entirely different lens of strategic competition.
On Wednesday, Chinese cybersecurity heavyweight 360 formally unveiled Tulongfeng, an advanced AI tool explicitly designed to match Anthropic’s restricted Mythos model. Alongside Tulongfeng, the firm introduced Yitianzhen, an automated system built specifically for cyber defense and incident response. Together, these tools represent a major leap forward in China’s domestic capabilities for automated software vulnerability discovery and threat mitigation.
The rollout was accompanied by pointed commentary from 360 founder Zhou Hongyi. According to reports from Reuters, Zhou characterized vulnerability-finding artificial intelligence as an essential national strategic asset. He strongly criticized what he termed the risk of "one-way transparency"—a scenario where select geopolitical actors possess a monopoly on advanced offensive and defensive cyber capabilities while leaving other nations blind.
Unlike their Japanese counterparts, who advocate for collaborative networks and multi-model orchestration, Chinese firms under the pressure of escalating technological restrictions are accelerating efforts to achieve total self-sufficiency. The release of Tulongfeng underscores Beijing’s determination to ensure that domestic enterprises and security apparatuses are never again left vulnerable to sudden shifts in foreign trade policy.
Broader Economic and Geopolitical Implications
The concurrent emergence of Sakana’s Fugu and 360’s Tulongfeng marks a critical inflection point in the global artificial intelligence race. The long-term implications of the U.S. export controls are beginning to manifest in ways that policymakers in Washington may not have fully anticipated.
First, the restrictions are rapidly eroding the de facto monopoly held by American AI laboratories over frontier-level capabilities. By creating an artificial scarcity and an unpredictable regulatory environment, U.S. export controls have effectively subsidized and accelerated the maturation of foreign competitors. Enterprise customers in Asia, wary of supply chain disruptions and political whims, now have powerful economic incentives to adopt local alternatives that are not only free from foreign export controls but are also culturally and linguistically fine-tuned to their domestic markets.
Second, the shift toward orchestration models and distributed multi-provider frameworks, as championed by Sakana AI, signals a structural change in how organizations will build their technological infrastructure. Rather than tethering their operations to a single proprietary ecosystem, businesses are increasingly likely to adopt modular architectures that allow them to swap out models dynamically depending on regulatory compliance, cost, and performance.
Finally, the geopolitical landscape of artificial intelligence is hardening into distinct regional blocs. While allied nations like Japan continue to advocate for open dialogue, shared access, and collaborative governance frameworks, adversarial nations like China are doubling down on indigenous development and defensive sovereignty.
As the regulatory dust settles from the initial restrictions on Anthropic’s flagship models, one reality remains clear: the attempt by the United States to contain frontier AI capabilities has not halted global progress. Instead, it has served as the starting gun for a decentralized, highly competitive global race where the next generation of artificial intelligence will be forged across multiple continents, far beyond the regulatory reach of Washington.
