HyperAI
Back to Headlines

FuriosaAI Partners with LG AI Research to Supply AI Chips for EXAONE Platform, Expanding Beyond South Korea

11 days ago

South Korean AI chip startup FuriosaAI announced a significant partnership on Tuesday to supply its AI chip, RNGD, to enterprises using LG AI Research’s newly launched EXAONE 4.0 platform. This marks a strategic shift for FuriosaAI, which previously turned down an $800 million acquisition offer from Meta, choosing to maintain its independence. FuriosaAI's RNGD chip is specifically designed for running large language models (LLMs), making it an ideal match for LG AI Research’s advanced hybrid AI model. The collaboration aims to target key sectors such as electronics, finance, telecommunications, and biotechnology, where AI applications are rapidly expanding. CEO June Paik of FuriosaAI explained the reasons behind the decision to decline Meta’s acquisition offer to TechCrunch, noting that the deal fell through due to differences in post-acquisition business strategy and organizational structure, rather than financial terms. "We want to continue our mission, and I believe it’s a very impactful contribution, both personally and for the company, to make AI computing more sustainable," Paik said. With the LG partnership, FuriosaAI sees business opportunities extending far beyond the South Korean market. Paik highlighted that EXAONE is recognized as a leading domestic AI model and will be widely adopted across the Korean AI ecosystem. However, he emphasized that the potential impact is global. "The LG team is also partnering with and doing business with international customers, so we expect our chip solutions to be used globally," Paik stated. One of the key factors in securing this deal was FuriosaAI’s focus on cost-effectiveness. LG AI Research found that FuriosaAI’s solution not only matched or exceeded the performance of competitive GPUs but also significantly reduced the total cost of ownership. Paik noted, "We had to prove that our solution not only delivers strong performance but also lowers overall costs." FuriosaAI further claims that the RNGD accelerator demonstrates superior inference performance, achieving 2.25 times better results compared to leading GPUs when tested with EXAONE models. Additionally, the chip is reportedly more energy-efficient, aligning with growing concerns about the environmental impact of AI computing. Unlike general-purpose GPUs such as those developed by Nvidia, FuriosaAI’s chips are tailored exclusively for AI tasks. "Our architecture is natively built for AI computing. We do not develop our chips for rendering or mining, allowing us to support a wide variety of AI models efficiently," Paik explained. Despite its relatively small global team of just 15 employees, with offices in Seoul and Santa Clara, FuriosaAI has made a substantial impact in the AI chip market. LG AI Research’s endorsement represents a significant milestone, as it is one of the few instances of a major enterprise publicly supporting a non-Nvidia chip solution. This partnership underscores the rising importance of specialized AI hardware and the growing competition to develop cost-effective, efficient alternatives. As the AI landscape continues to evolve, FuriosaAI’s strategic moves position it as a formidable player in the global market.

Related Links