South Korea's FuriosaAI is seeking to win inference customers from Nvidia with RNGD, its second-generation AI processor. The company says the chip consumes less power than comparable GPUs while operating in conventional air-cooled data centres.
RNGD began shipping in January 2026, and the startup, founded in 2017 by June Paik, a former Samsung Electronics and AMD engineer, opened a Singapore subsidiary last month. It plans to more than double production next year.
FuriosaAI has since raised total funding of US$246 million and is considering a public listing as early as 2027. Investors include Korea Development Bank, Industrial Bank of Korea, and Kakao Investment.
The company aims to tap a growing segment of the AI market: inference, the stage at which trained models generate outputs such as text, images, and recommendations. As demand for generative AI applications and agent-based systems grows, so does the volume of inference workloads. This has increased focus on power consumption, operating costs, and infrastructure requirements.
FuriosaAI is targeting a narrower market with processors designed specifically for inference. The company believes this approach allows it to compete on efficiency and cost while avoiding the immense expense of developing chips for AI model training.
According to FuriosaAI, LG AI Research was an early customer, using RNGD with its EXAONE models and reporting 2.25 times the inference performance per watt of GPU-based systems.
FuriosaAI said it would begin supplying Samsung SDS for cloud deployments from July. In Europe, Velox DCI Europe and I/ONX HPC plan to deploy more than 7,000 RNGD accelerators at a Stockholm data centre.
FuriosaAI’s expansion into the region reflects growing the competition for Southeast Asia's AI infrastructure market, where enterprises and cloud providers are increasingly evaluating.
