Hajra Naz
Key Takeaways
-
FuriosaAI partnered with LG AI Research to power the new EXAONE 4.0 model
-
The startup’s RNGD chip outperformed traditional GPUs on both performance and energy efficiency.
-
Meta’s $800M acquisition offer was rejected due to vision differences, not price.
-
This partnership positions FuriosaAI as a serious rival to Nvidia, especially in Asia.
-
With only 15 employees, FuriosaAI is now poised for global impact.
South Korean AI chipmaker FuriosaAI is making headlines again—but not for an acquisition. Just months after turning down an $800 million buyout offer from Meta, the startup has announced a major new partnership with LG AI Research to supply its RNGD AI accelerator for LG’s latest large language model platform, EXAONE 4.0.
LG’s New AI Model Will Run on Furiosa’s RNGD Chip
LG AI Research has recently unveiled EXAONE 4.0, its next-generation hybrid AI model designed for use across various industries, including electronics, finance, telecommunications, and biotechnology. Now, with FuriosaAI’s RNGD chip integrated, EXAONE will be getting a performance boost tailored specifically for large language model (LLM) workloads.
Unlike traditional GPUs, RNGD isn’t a general-purpose processor. It’s purpose-built for AI from the ground up.
“We don’t design chips for rendering or crypto mining,” said FuriosaAI CEO June Paik. “Our architecture is natively built for AI computing, and we support a wide range of AI models more efficiently.”
Meta Tried to Buy FuriosaAI—But the Startup Walked Away
The announcement comes just three months after Meta reportedly offered $800 million to acquire FuriosaAI. The deal ultimately fell through—not because of money, but because of differences in vision.
According to Korean media, disagreements over post-acquisition strategy and organizational structure led FuriosaAI to walk away from the table.
“We want to continue our mission,” said Paik. “It’s an exciting opportunity for us to make AI computing more sustainable—not just for South Korea, but globally.”
The rejection signals a broader trend of promising AI hardware startups opting for independence over acquisition, especially as tech giants like Meta seek to reduce their reliance on third-party suppliers like Nvidia.
Why LG Picked Furiosa Over Nvidia
One of the most significant aspects of the LG-Furiosa deal is that it represents a rare public endorsement of a non-Nvidia solution by a major enterprise.
Furiosa’s RNGD chip didn’t just compete—it outperformed rival GPUs in LG’s internal testing with the EXAONE model. The company claims a 2.25x improvement in inference performance, along with improved energy efficiency.
“We had to prove that our solution delivers strong performance and lowers total cost of ownership,” said Paik. “That’s what won us this deal.”
This is a critical differentiator as demand grows for more cost-effective, energy-efficient AI infrastructure—especially for sovereign AI models being developed outside the U.S.
Read More: Nvidia Faces Billions in Losses Over New H20 Chip Licensing Rules
Big Vision, Small Team, Global Ambitions
FuriosaAI might be small—just 15 employees globally, split between Seoul and Santa Clara—but it’s thinking big. Paik believes this partnership with LG opens the door to international opportunities.
“LG’s EXAONE model is not just for internal use. It’s becoming a key platform in the Korean AI ecosystem and will also be used by global customers working with LG,” said Paik. “That means our chips will go international too.”
While Paik didn’t comment on whether FuriosaAI is now seeking fresh funding, the LG partnership may eliminate the immediate need for external capital by unlocking new revenue streams and market exposure.
A New Player in the AI Chip War?
Nvidia still dominates the AI hardware space. But challengers like FuriosaAI are gaining ground—especially in Asia. Governments and companies there want more control over their tech.
FuriosaAI’s chip is built specifically for AI. It has already outperformed GPUs in some tests. With LG backing it, the startup is now on the radar of global businesses, cloud providers, and AI labs.
