TOKYO—In a bold move underscoring Japan’s push for technological sovereignty, Sakana AI has secured $135 million in Series B funding, catapulting its valuation to $2.65 billion and crowning it as the nation’s most valuable unicorn. The Tokyo-based startup, founded in 2023 by former Google researchers David Ha and Llion Jones, along with Ren Ito, is pioneering nature-inspired artificial intelligence models tailored specifically for Japanese enterprises. This latest round, announced on November 17, 2025, highlights a growing trend in Asia where countries are racing to develop culturally aligned large language models (LLMs) amid concerns over reliance on U.S.-dominated AI technologies.

The funding, led by Mitsubishi UFJ Financial Group and including participation from In-Q-Tel—the venture arm linked to the U.S. Central Intelligence Agency—signals strong interest from both domestic heavyweights and international players. Other investors include Sony Group, Nippon Telegraph & Telephone, and NEC, according to reports from TechCrunch. Sakana AI plans to channel the capital into expanding its research on efficient, resource-light AI models that address national needs in sectors like finance, defense, and manufacturing, while emphasizing sustainability in an era of escalating energy demands from AI compute.

From Fish Schools to AI Innovation

The company’s name, ‘Sakana’—Japanese for ‘fish’—draws from the concept of collective intelligence, akin to a school of fish forming coherent patterns through simple rules. This philosophy underpins Sakana’s approach to AI development, focusing on evolutionary and collective intelligence methods rather than the brute-force scaling seen in models like OpenAI’s GPT series. As detailed on the company’s website, Sakana AI aims to create ‘transformative AI that will bring us into the next paradigm,’ with a core focus on foundation models inspired by nature.

David Ha, a co-founder and former Google Brain researcher, has emphasized efficiency in AI training. In a post on X (formerly Twitter), Ha noted the company’s strategy to ‘do more with fewer GPUs and smaller datasets,’ addressing the unsustainable resource consumption plaguing the industry. This resonates with Japan’s broader AI strategy, supported by government grants like the NEDO supercomputer initiative, which selected Sakana as one of seven institutions to bolster the nation’s AI ecosystem, as announced in February 2024 via the company’s official X account.

Llion Jones, another co-founder and co-author of the seminal 2017 paper ‘Attention Is All You Need’ that birthed transformer models, brings deep expertise to the venture. The team’s track record has attracted significant backing: prior to this Series B, Sakana raised $30 million in seed funding from Lux Capital and Khosla Ventures, followed by approximately $200 million in Series A from investors including Mitsubishi UFJ, SMBC, Mizuho, Itochu, KDDI, Nomura, and Nvidia, per Wikipedia.

Sovereign AI: Tailoring Models for Japan’s Needs

At the heart of Sakana’s mission is the development of sovereign LLMs—AI models attuned to Japanese language, culture, and regulatory environments. This is particularly crucial for sensitive sectors like finance and defense, where data privacy and national security are paramount. The Observer reported on November 18, 2025, that Sakana aims to ‘build LLMs aligned with national needs,’ highlighting the language barriers in dominant English-centric models from U.S. firms.

In finance, Sakana’s models could revolutionize risk assessment and fraud detection by incorporating Japan-specific economic data and linguistic nuances. For defense, the involvement of In-Q-Tel suggests potential applications in secure, localized AI for intelligence and cybersecurity, though details remain sparse. Nikkei Asia, in its November 16, 2025, coverage, noted Sakana’s $2.63 billion valuation positions it as Japan’s top unicorn, with plans to expand into enterprise solutions that drive Asia’s projected $125 billion AI revenue push by 2030.

The push for culturally aligned AI addresses a broader ‘language problem’ in LLMs, as outlined in the Observer article. While global models excel in English, they falter in non-Latin scripts and cultural contexts, making localized alternatives essential for markets like Japan. Sakana’s approach, including innovative techniques like automated model merging, allows for high-performance AI with modest computational resources— a stark contrast to the energy-guzzling behemoths from Silicon Valley.

Investor Confidence and Strategic Partnerships

The Series B round’s investor lineup reads like a who’s who of Japanese industry, reflecting strategic alignment with national priorities. Mitsubishi UFJ’s lead role, as reported by Tech in Asia on November 17, 2025, underscores banking giants’ interest in AI for fintech innovation. In-Q-Tel’s participation, detailed in Digitimes on November 19, 2025, adds an intriguing geopolitical layer, potentially fostering U.S.-Japan collaboration in AI amid rising tensions with China.

Nvidia’s ongoing involvement, stemming from the Series A, will support Sakana’s R&D in AI data centers and infrastructure, as mentioned in X posts from investors like Beth Kindig in September 2024. This partnership aligns with Japan’s government-backed efforts to build domestic AI capabilities, including GPU clusters funded by NEDO grants.

Posts on X from industry figures, such as Rohan Paul on November 18, 2025, praise Sakana’s focus on ‘efficient, Japan-focused AI models’ under realistic compute limits. Similarly, The Japan Times tweeted on November 17, 2025, about Sakana targeting defense and banking markets with its ¥400 billion valuation.

Challenges in the Global AI Race

Despite the momentum, Sakana faces hurdles in a field dominated by U.S. and Chinese giants. The company’s emphasis on sustainability—critiquing the ‘record amount of capital pouring into AI compute at a rate which may not be sustainable,’ as stated on its Series B announcement page—positions it as a counterpoint to resource-intensive models. However, scaling these innovations to compete globally remains a test.

Japan’s AI landscape is evolving rapidly, with Sakana leading the charge. As eWeek reported on November 18, 2025, the firm’s $2.635 billion valuation in yen terms (approximately 400 billion yen) marks a milestone for Japanese startups. Industry insiders view this as part of Asia’s broader AI revenue surge, projected to reach $125 billion, driven by enterprise adoption in finance and defense.

Sakana’s journey from a $30 million seed in January 2024 to unicorn status exemplifies rapid innovation. Co-founder Ha’s X post from December 2024 reflected on building a ‘world-class research team’ in Tokyo, blending frontier research with applied engineering.

Future Horizons for Japanese AI

Looking ahead, Sakana plans to deepen its enterprise focus, potentially launching specialized models for Asia-Pacific markets. The company’s nature-inspired methods, such as evolving AI through collective intelligence, could yield breakthroughs in efficiency, as explored in its research on model merging.

With government support and strategic investments, Sakana is poised to influence global AI standards. As PARAS ARORA noted on X on November 17, 2025, this funding represents a ‘shockwave in the global AI race,’ pitting scale against efficiency in the quest for AI supremacy.

The implications extend beyond Japan, signaling a shift toward decentralized, sovereign AI development worldwide. As Sakana navigates this landscape, its success could redefine how nations approach AI sovereignty in critical sectors.