FuriosaAI raises $620 million to challenge NVIDIA with low-power AI inference chips

FuriosaAI announced a $620 million Series D funding round on Wednesday, valuing the Seoul-based chipmaker at $2.9 billion ventureburn.com. South Korea’s National Growth Fund contributed the largest cheque at $280 million, while U.S. investor Infinitum led a $110 million international tranche. Existing shareholders, including NAVER, Korea Development Bank and Kakao Investment, added $125 million, and the Export-Import Bank of Korea made its first direct startup investment.

FuriosaAI focuses on AI inference rather than training, with its flagship RNGD (“Renegade”) chip built on a Tensor Contraction Processor design. The 180-watt, air-cooled card reportedly matches the throughput of GPUs that consume twice the power. In tests with LG’s EXAONE models, the RNGD delivered 2.25 times better performance per watt than the compared GPU system. Mass production began in January at TSMC, and LG Group, Upstage and Samsung SDS have already integrated the chip into their services.

The company aims to manufacture 20,000 RNGD chips this year and over 40,000 by 2027, while a third-generation chip with Broadcom is under development. A public listing is being considered as early as 2027.