Sakana AI, a Japanese startup co-founded by ex-Google AI researchers, is on the brink of achieving unicorn status with a valuation exceeding $1 billion.
The company, set to finalize a funding round this month, is in discussions with prominent US venture capital firms New Enterprise Associates, Khosla Ventures, and Lux Capital to secure $125 million.
Established in July 2023 by David Ha, Llion Jones, and Ren Ito, Sakana AI quickly attracted attention with its innovative approach to generative artificial intelligence.
The startup’s rapid growth is fueled by the current generative AI boom, which has seen companies like Microsoft-backed OpenAI surge to an $80 billion valuation as of February.
In January, Sakana AI raised $30 million in a round that included investors such as NTT Group, KDDI, and Sony Group. These funds are earmarked for advancing their generative AI research, leveraging a novel approach that mimics biological evolution by merging multiple small AI models.
This strategy not only cuts costs but also reduces power consumption, making it a magnet for global investment interest.
The generative AI landscape has seen rapid expansion since the launch of OpenAI’s ChatGPT in 2022, with significant capital flowing into the sector. In fact, the burgeoning field of generative AI has seen significant activity, with several startups achieving high valuations. France’s Mistral AI and Canada’s Cohere have recently seen substantial increases in their valuations, highlighting the intense investor interest in this sector.
Cohere recently raised $450-million in funding from returning investors such as Nvidia (NASDAQ:NVDA) and Salesforce Ventures (NYSE:CRM) and new investors Cisco (NASDAQ:CSCO) and Canadian pension fund PSP Investments, according to a source familiar the matter.
Sakana AI’s innovative approach and rapid valuation increase reflect the dynamic and rapidly evolving landscape of global AI development, positioning Japan as a significant player in the field alongside its U.S. counterparts.
Featured Image – Freepik
