South Korean AI Chip Startup FurioSai Announced a partnership on Tuesday to supply your AI chip, RNGDTo use enterprises LG AI ResearchRecently unveiled the Exaone platform.
RNGD is adapted to run large language model (LLM) and Just last week, Korean tech giant LG unveiled Its next generation Hybrid AI Model Exaone 4.0. Cooperation targets major sectors including electronics, finance, telecommunications and biotechnology for a series of various applications.
The news comes after about three months when Furiosai rejected the Meta’s $ 800 million acquisition, the option to remain independent.
According to local media outlets, the deal collapsed due to the post-acquisition business strategy and disagreement over the organizational structure, rather than value issues. Meta’s interest in achieving AI chipmakers such as furiosai reflects its widespread strategy to reduce its dependence on third-party suppliers such as Nvidia.
Asked why the Meta was dealt with, Furiosai CEO told Pike to Techcrunch: “We want to continue our mission, and I think it is an exciting opportunity at the same time. I believe it is a very impressive contribution, both personal and for the company, to make AI computing more.”
With M & A (at least from Meta) from the table, Paik refused to specify whether the startup is now chasing fresh funding.

However, Paik says that this new partnership will have commercial possibilities beyond South Korea.
Techcrunch event
San francisco
,
27-29 October, 2025
“LG AI’s exon is considered as the leading sovereign AI model in South Korea. It will not be used only within LG. It will be one of the main AI models used in the Korean AI ecosystem. We hope that there will be many demands for this Axon, as well as working as well as the customer including the customer.
Paik stated that LG AI’s decision to adopt Furiosa’s AI chip and accelerator is notable for another reason: it is one of some public support of the rival of NVIDIA by a major venture. One of the major reasons for winning is that the cost of the hardware of the startup is low.
“We had to prove that our solution not only provides strong performance, but also reduces the total cost of ownership,” Pike said.
Furiosai claims that its RNGD accelerator improved the competitive GPU with an exon research model of LG AI Research, leading to 2.25 times better performance. Paik also states that LG found that fuuryosai hardware was more energy efficiency.
The chip of furiosa is not a normal GPU, but was specially designed for AI. “We can efficiently support many types of AI models. But unlike the GPU, which are still fundamentally general-purpose processors, our architecture is basically designed for AI computing. We do not develop our chip for rendering or mining,” Pack said.
The Seoul-based startup, which also operates an office in Santa Clara, has only a global team of 15 employees.