FuriosaAI declined Meta’s acquisition offer, secures LG deal

The South Korean FuriosaAI Inc., a Seoul-based chip startup aiming to challenge Nvidia’s dominance, will deliver its RNGD AI chips to LG AI Research as part of a new partnership deal. LG intends to use the chips in its Exaone language models.

FuriosaAI developed the RNGD chip to take on the dominant Nvidia and emerging competitors such as Groq, SambaNova, and Cerebras.

Furiosa’s Paik says their chips are energy-efficient and cost-effective

FuriosaAI Chief Executive Officer June Paik commented on the RNGD chip, “For the last eight years, we worked very hard from R&D to product phases and finally this commercialization phase. This signals that our product is ready for enterprise adoption.” 

Paik also acknowledged LG AI’s EXAONE as the top AI model in South Korea, adding that it is set to play a central role in the country’s broader AI landscape.

He projected a strong interest in the model and its chips, locally and globally. Paik added that LG is already working with global customers, and anticipates that those clients will also buy their chips.

The executive also claimed that LG struck a deal because its hardware was cost-effective and energy efficient and offered 2.25 times superior inference performance. He also stated their chips are designed exclusively for AI computing. 

FuriosaAI and LG will roll out RNGD-powered servers running Exaone across various sectors, including electronics and finance. The chips will also support LG’s internal enterprise AI agent, ChatExaone.

FuriosaAI is still working to gain ground in the semiconductor industry, like Rebellions Inc. and Semifive Inc. It also plans to expand its client base in the US, the Middle East, and Southeast Asia and secure more partnerships in the second half of this year.

FuriosaAI declined Meta’s acquisition offer

FuriosaAI recently turned down Meta’s $800 million acquisition offer. The two companies disagreed over differences in business strategy and organizational plans that would follow after the acquisition.

When the deal fell through, Paik gave a statement reinforcing their goal of advancing sustainable AI, but failed to specify that they were looking for fresh funding.

Like other tech firms developing large language models (LLMs), Meta has been working to lessen its dependence on Nvidia for chips needed to train and run these models. The tech company said it would funnel over $65 billion in 2025 to support its AI initiatives.

Meanwhile, FuriosaAI is currently negotiating a $48 million, or about KRW 70 billion, funding round with investors and hopes to secure the investment by the end of this month.

LG AI Research launched its hybrid AI model Exaone 4.0

This month, LG AI Research released its new hybrid AI model, the Exaone 4.0. The company is marketing the new model as South Korea’s first hybrid-reasoning AI system. The model reportedly generates and assesses hypotheses.

LG claims it’s improving EXAONE further to establish itself in the market. This March, the company introduced South Korea’s first reasoning artificial intelligence (AI) model, Exaone Deep. The model includes 32 billion parameters to facilitate AI learning and reasoning, and experts claim its performance is almost similar to Deepseek’s R1. 

Cryptopolitan Academy: Want to grow your money in 2025? Learn how to do it with DeFi in our upcoming webclass. Save Your Spot