FuriosaAI, AI chip startup, rejects $800 million acquisition bid from Meta

to be the year when a highly intelligent and personalized AI assistant reaches more than 1 billion people, with hopes for Meta AI to fill that role.

However, pioneering the latest AI advancements necessitates sophisticated chips or graphic processing units (GPUs) to handle the workload. Early this month, Meta began trials on its inaugural in-house chip designed for AI training. This move aims to cut infrastructure costs significantly as Meta invests heavily in AI to propel growth.

FuriosaAI, founded in 2017 by June Paik, pushed back against Meta’s $800 million acquisition offer. Discord over post-acquisition business strategies and organizational structure led to the breakdown of negotiations, prompting the company to concentrate on AI chip development and production instead.

Meta, alongside other tech firms delving into AI, is striving to decrease dependence on industry giant Nvidia. Startups like Positron AI are receiving funding to introduce competition. Despite this, Nvidia’s Q4 earnings report positioned it as the second most valuable company, showcasing a continuous upward trend.

FuriosaAI’s cutting-edge AI chips, Warboy and Renegade (RNGD), have been engineered to challenge Nvidia and AMD. Testing on RNGD chips is complete, with LG AI Research and Aramco jointly verifying their efficiency for reasoning models. LG AI Research intends to integrate RNGD chips into its AI infrastructure, and FuriosaAI is on track to launch the chips later this year.

With plans to secure approximately $48 million (KRW 70 billion) in funding, FuriosaAI is primed to wrap up financing this month. An industry insider lauded the company’s decision to emphasize technological independence and long-term growth over global capital, which is poised to significantly impact the Korean AI ecosystem. Additionally, the company aims to expedite the commercialization and global expansion of its Renegade chip in the near future.

Mark Zuckerberg’s Meta, among other tech giants, is at the forefront of pioneering state-of-the-art AI models while vying for AI dominance. Zuckerberg envisions 2025 as the year when a highly intelligent, personalized AI assistant reaches a vast user base, with Meta AI positioned to fulfill that role.

Although forging ahead with the latest AI technologies demands advanced chips or GPUs to sustain their operations, Meta initiated tests on its inaugural in-house chip tailored for AI training earlier this month. The objective is to curtail substantial infrastructure costs as Meta takes bold strides in leveraging AI for growth.