While OpenAI, Anthropic and Meta trade equity stakes and proxy wars for Washington's favor, the more interesting AI story today is happening underneath them, in Seoul trading desks, Gulf ports, Indian classrooms and solo GitHub repos shipping agent tooling nobody has heard of yet. A Check Point find turned a theoretical browser exploit into working AI-generated ransomware, a reminder that the same open tooling powering scrappy new startups cuts both ways. Even Japan's ten million robot bet reads less like a moonshot and more like a hedge, evidence that the frontier is getting crowded from the edges in, not just the center out.
Meituan open-sourced a 1.6 trillion parameter model trained on more than 50,000 domestic AI chips with zero Nvidia hardware, claiming the country's first large model built end-to-end on homegrown silicon.
Raised 800 million dollars in a Series C at an 8.3 billion dollar valuation, led by Aramco Ventures with participation from Vista Equity Partners, General Catalyst and Nvidia.
Local first persistent memory for AI coding agents like Claude Code, Cursor and Codex, stored in one SQLite file with no cloud or API keys
Anthropic's espionage claims against Alibaba expose the real fragility behind its trillion-dollar valuation, a frontier model lead that distillation attacks can chase down faster than Washington's export controls can protect it.
