FuriosaAI Inc, the Seoul‑based AI‑chip start-up that rebuffed an $800 million acquisition bid from Meta Platforms earlier this year, has secured its first major commercial contract with LG AI Research, Bloomberg reported.
After seven months of intensive performance and efficiency testing, LG granted final approval for FuriosaAI’s RNGD (pronounced “Renegade”) processor to power its Exaone large‑language models, marking the start-up’s transition from R&D to enterprise deployment.
RNGD Chip to Fuel Exaone Models
Under the deal, LG will integrate RNGD into its Exaone servers, deploying them across sectors such as electronics, finance and logistics. The chip will also underpin LG’s in‑house AI agent, ChatExaone, which the conglomerate plans to extend to external corporate clients.
FuriosaAI Chief Executive Officer June Paik hailed the partnership as a watershed moment: “For the last eight years, we worked very hard from R&D to product phases and finally this commercialisation phase. This signals that our product is ready for enterprise adoption.”
Founded in 2017 by Paik, formerly of Samsung Electronics and AMD, FuriosaAI focuses on semiconductors optimised for AI inference. The company’s benchmark tests claim RNGD delivers up to 2.25 times greater inference performance per watt compared to conventional graphics processing units. FuriosaAI is part of a growing Korean ecosystem of chip innovators, alongside firms like Rebellions and Semifive, benefiting from proximity to Samsung and SK Hynix’s manufacturing and talent networks.
The contract with LG comes months after FuriosaAI made headlines by rejecting Meta’s takeover proposal, choosing to remain independent and pursue its own growth trajectory.
Insiders say the start-up is preparing a fresh fundraising round to fuel global expansion, with plans to announce additional partnerships in the US, West Asia and Southeast Asia during the second half of 2025. An eventual initial public offering is also on the roadmap, executives confirm.
LG’s embrace of RNGD reflects a broader industry push to diversify AI‑compute suppliers beyond market leader Nvidia. start-ups such as Groq, SambaNova and Cerebras have similarly attracted attention, but few have achieved a commercial foothold at scale. By fielding RNGD in high‑visibility applications like Exaone and ChatExaone, FuriosaAI aims to demonstrate that homegrown alternatives can meet enterprise demands for both performance and power efficiency.
As the AI infrastructure market matures, securing marquee clients like LG is critical for emerging chip designers. FuriosaAI’s early success suggests that the post‑ChatGPT boom in AI computing will support a more heterogeneous hardware landscape, one where specialised, energy‑efficient processors can coexist alongside established GPU offerings. For now, June Paik and his team will focus on delivering on LG’s performance benchmarks as they prepare to extend RNGD’s reach across global data centres.
