Life for Tokyo-based AI company Sakana AI is going swimmingly well.
Sakana, which means ‘fish’ in Japanese, has completed a $135 million funding round, that takes its valuation to 400 billion Yen ($2.635 billion). The company plans to use its new funding for R&D, expanding applied AI, and investments and potential M&A.
The firm’s novel approach to model development — inspired by nature and collective intelligence — is attracting both global investors and major Japanese corporations.
In a good ‘plaice’
The ‘fish’ theme was chosen to reflect its approach: multiple simpler components collaborating like a school of fish rather than a single massive monolithic model.
According to Sakana AI’s announcement, the Series B funding round marks one of the largest recent AI investments in Japan.
The funding comes from a mix of Japanese and international investors, including Mitsubishi UFJ Financial Group (MUFG), Khosla Ventures, Macquarie Capital, Lux Capital, New Enterprise Associates, MPower Partners, Ora Global, and In-Q-Tel (IQT).
Sakana AI’s announcement comes amid heightened global concerns about the sustainability of AI development. Major firms worldwide are investing unprecedented amounts into data centers, GPU clusters, and energy-intensive training runs.
The company frames its mission as a direct response to this trend, stating, “At present, we are seeing a record amount of capital pouring into AI compute at a rate which may not be sustainable, funding AI businesses without a clear path to profitability, and consuming unprecedented levels of energy.”
This critique echoes a growing debate among economists, policymakers, and technologists: whether the current trajectory — dominated by massive foundation models requiring near-limitless resources — is viable long-term. Japan in particular faces structural constraints: limited energy reserves, fewer hyperscale data centers, and demographic pressures that reduce both workforce and risk tolerance.
Sakana AI questions whether Japan should attempt to replicate the U.S. or Chinese scale-based strategies.
“But is this really the future that we want to build? As an AI R&D company in Tokyo, we also question whether this is the right approach for Japan, a nation with limited resources, to build its Sovereign AI.”
Nature-inspired AI
Sakana AI positions itself as an alternative, arguing that intelligence — and evolution itself — arises from scarcity, not abundance.
The firm’s philosophy has informed its research direction from the beginning. Rather than training ever-larger models from scratch, Sakana AI focuses on an evolutionary model merging to combine capabilities from existing open-source models, tree search methods to integrate and optimize closed-source models, and self-evolving architectures capable of improving their own code.
While Sakana AI operates as a frontier research organization, it has also built a commercial arm that works directly with major Japanese companies.
Key partnerships include MUFG and Daiwa Securities Group, where Sakana AI is helping develop domain-specific AI systems for financial operations. These involve tailoring models to Japan’s unique professional culture, tacit business knowledge, and regulated environments.
Numbers and humans
Japan’s aging population and shrinking workforce make AI adoption especially urgent. With fewer human workers available, automation and intelligent systems will be required to maintain productivity across sectors such as finance, manufacturing, logistics, and healthcare.
Therefore, the company is also expanding beyond finance into defense, intelligence, and manufacturing, reflecting growing government interest in national AI capability.
Jeff Bezos, founder of Amazon and one of the world’s richest individuals, is returning to hands-on leadership with a new AI venture.