Tokyo-based Sakana AI has closed a $135 million Series B round, positioning itself as a standard bearer for more sustainable and resource-efficient artificial intelligence in Japan. The company, founded roughly two years ago, is pushing back against a global trend of capital-intensive AI development that depends on massive compute and energy consumption. Instead, it argues that Japan’s path to sovereign AI should be built on doing more with less, mirroring how intelligence emerges under resource constraints in nature.
Funding Round Details
The Series B round, ¥20 billion, equivalent to about $135 million, will be used to accelerate both frontier research and real world implementation of Sakana AI’s technology in Japan. The company says the raise brings its total funding to a level that puts it among the more heavily backed AI players in the country’s innovation ecosystem. The new capital is intended to deepen its research roadmap and support scaled deployments with large corporate and public sector partners.
Investor Participation and Rationale
Sakana AI’s investor base includes a mix of domestic strategic backers and global venture capital firms. Participants in the Series B include Mitsubishi UFJ Financial Group, Khosla Ventures, Factorial, Macquarie Capital, Fundomo, Mouro Capital, New Enterprise Associates, Geodesic Capital, Lux Capital, Ora Global, MPower Partners, Shikoku Electric Power and In Q Tel. MUFG president and group CEO Hironori Kamezawa said the bank views Sakana AI as a partner in driving AI adoption across Japanese industries, while MPower Partners general partner Kathy Matsui highlighted the company’s alignment with ESG oriented, industry transforming technologies.
Research and Technology Focus
From its inception, Sakana AI has focused on efficient, evolutionary approaches to building AI systems rather than repeatedly training massive foundation models from scratch. Its research portfolio includes self improving architectures such as the Darwin Gödel Machine and ShinkaEvolve, which aim to evolve code and models while prioritizing computational efficiency. The company has also introduced techniques for merging multiple open source models, orchestrating closed models for collaborative reasoning, and experimenting with architectures like the Continuous Thought Machine that move beyond traditional transformer designs.
Automating AI Science and Edge Efficiency
Sakana AI is also pursuing automation of AI research itself through multi agent systems. Its AI Scientist platform is designed to handle workflows such as hypothesis creation, experiment design, and paper drafting, with the goal of accelerating algorithm discovery. In parallel, the company has developed energy efficient language models that can run on edge devices, supporting its broader thesis that future AI needs to be lighter, more distributed, and more sustainable.
Applied AI and Market Strategy
On the commercial side, Sakana AI has been building a growing enterprise business in Japan, working with large financial institutions and other corporates to deploy domain specific applications. It has strategic collaborations with MUFG and Daiwa Securities Group to develop tailored AI tools for finance, where tacit and uncodified knowledge can make generic models insufficient. The company now plans to expand this applied AI work beyond financial services into defense, intelligence, and manufacturing, sectors where targeted engineering and local context are critical.
Sovereign AI Vision and Use of Proceeds
Sakana AI’s broader vision is to contribute to Japan’s sovereign AI capabilities, with a focus on post training rather than competing in the global race to train ever larger base models. By concentrating on alignment, domain adaptation, and optimization layers for Japanese language, culture, and regulatory needs, the company argues that the country can build robust AI systems with far lower resource requirements. The Series B funds will therefore be channeled into advancing collective intelligence and self evolution research, scaling applied teams, and pursuing strategic partnerships, investments, and acquisitions to build a longer term ecosystem.
For a country facing demographic decline and labor shortages, Sakana AI presents its model of resource aware, locally tuned AI as a pragmatic path to productivity and resilience. The company’s backers are effectively betting that sustainable architectures and deep domain implementation will outperform brute force compute strategies over time. With fresh capital and a growing roster of corporate partners, Sakana AI is now moving into a phase focused on translating its Tokyo-based R&D into broad adoption across Japan’s business and public sectors.