Asian startups launch Mythos-like models as U.S. export ban stalls
U.S. AI export rules blocked Anthropic’s Mythos (10^27 FLOPs) from restricted markets, letting Asian startups—Sakana AI, Upstage AI, and Decart AI—launch competitive models without compliance limits.
Silicon Valley’s fear of U.S. export rules just handed Asia a multibillion-dollar lead in AI. Three Asian startups—out of stealth in the last ten days
Read Full Story at TechCrunch →Why This Matters
The sudden emergence of Asian AI startups deploying Mythos-class models exposes a critical vulnerability in U.S. export controls—one that could reshape global AI development overnight. By exploiting regulatory gaps, these startups are not just filling a void but accelerating a decentralized AI arms race where sovereignty trumps compliance. The move signals a potential shift in power dynamics, where restricted markets may no longer depend on American giants for cutting-edge infrastructure.
Background Context
Anthropic’s Mythos model, designed to push the limits of AI capability, became collateral damage in U.S. export restrictions targeting high-performance computing in sanctioned regions. Meanwhile, Asian markets—already wary of over-reliance on Western AI—have quietly nurtured domestic alternatives. The timing is no coincidence: as geopolitical tensions rise, so does the urgency to develop AI models outside the reach of U.S. sanctions, particularly in South Korea and Japan where tech ecosystems are rapidly maturing.
What Happens Next
Expect a domino effect as other Asian startups follow suit, forcing U.S. regulators to either tighten loopholes or risk ceding AI leadership to non-Western players. The models’ performance in restricted markets will serve as a real-world stress test—will they truly outperform Western alternatives, or will compliance gaps lead to unforeseen technical limitations? Either way, the genie is out of the bottle, and the race to build unrestricted AI is now a global sprint.
Bigger Picture
This isn’t just about AI models—it’s a microcosm of a broader fragmentation in technology, where regional blocs prioritize self-sufficiency over collaboration. As export controls proliferate, we’re witnessing the birth of a multipolar AI ecosystem, where innovation thrives in regulatory shadows. The long-term risk? A splintered AI landscape where fragmentation stifles progress just as the technology reaches its most transformative potential.
