FuriosaAI and LG Partner to Challenge Nvidia with Energy-Efficient AI Chips
July 22, 2025
FuriosaAI Inc., a Seoul-based AI chip startup, has partnered with LG AI Research to supply its RNGD AI chips for use in LG's Exaone language models, marking a significant step for Korean semiconductor innovation.
The RNGD chips deliver 2.25 times better inference performance than competitive GPUs and are more energy-efficient, making them attractive alternatives in AI data centers.
LG AI Research's integration of RNGD chips into its Exaone platform has validated FuriosaAI's technology, demonstrating superior cost-performance compared to Nvidia's H100 chips.
This partnership underscores a growing demand for specialized, affordable, and energy-efficient AI hardware, as companies seek alternatives to Nvidia's dominant market position.
FuriosaAI's RNGD chips are designed specifically for AI inference, with features like OpenAI-compatible APIs and Kubernetes support, facilitating large-scale deployment.
Founded in 2017 by former Samsung and AMD engineer June Paik, FuriosaAI has spent eight years developing its AI inference semiconductors, aiming to compete globally.
Despite the competitive landscape, including Nvidia, AMD, and other startups, FuriosaAI focuses on energy efficiency and affordability to carve out a significant share of the AI inference market.
The company's recent validation through LG's Exaone platform highlights its potential to challenge established players and expand its international presence.
FuriosaAI rejected an $800 million acquisition offer from Meta earlier this year, choosing to remain independent and focus on growth and strategic collaborations.
The RNGD chips, with 512 teraFLOPS of FP8 performance and 48GB memory, are optimized for tensor parallelism, enabling efficient model distribution across multiple accelerators.
This partnership with LG is expected to open new global business opportunities, especially as LG's Exaone model attracts international customers.
FuriosaAI's CEO, June Paik, emphasizes their goal to make AI computing more sustainable and to maintain independence in a competitive industry.
Summary based on 13 sources
Get a daily email with more Tech stories
Sources

TechCrunch • Jul 22, 2025
Instead of selling to Meta, AI chip startup FuriosaAI signed a huge customer
Economic Times • Jul 22, 2025
Jensen Huang may have met his match, and it’s not AMD, but a stealthy South Korean challenger
Moneycontrol • Jul 22, 2025
Artificial Intelligence News Live: FuriosaAI to supply AI chips for LG's EXAONE enterprise platform
The Register • Jul 22, 2025
How AI chip upstart FuriosaAI won over LG with its power-sipping design