Sakana AI

Sakana AI builds frontier AI systems in Tokyo, founded July 2023 by former Google researchers David Ha (CEO), Ren Ito (Chairman), and Llion Jones (CTO). The lab's research program is drawn from biological systems - evolutionary optimization, collective intelligence, and recursive self-improvement - and the name reflects that: sakana means fish in Japanese, and the logo is a school of fish, a metaphor for self-organization. A $135M Series B at a $2.65B valuation closed in November 2025, with NVIDIA, Google, and major Japanese corporations among the backers.

The stakes for anyone working on security here are concrete rather than hypothetical. The company's flagship system, The AI Scientist, autonomously conducts scientific research end to end and was published in Nature; Sakana Marlin is an autonomous research agent for strategic analysis. Alongside those sit Sakana Fugu, an LLM that orchestrates other frontier models, and Sakana Namazu, an LLM specialized in Japanese. The threat model is not just a web app with a model bolted on - it's agentic systems that generate and execute work, chain across third-party models, and operate on sensitive corpora in Japanese.

Those systems are being put in front of regulated and high-consequence users. Sakana AI collaborates with MUFG Bank, SMBC Group, and Daiwa Securities Group in financial services, and with the Ministry of Defense and the Ministry of Internal Affairs and Communications on the government side, spanning defense, public sector, and telecommunications regulation. The technical surface runs across AI research, LLM orchestration, natural language processing for Japanese, and autonomous scientific research.