img

FuriosaAI, a South Korea-based semiconductor startup specializing in AI inferencing chips, has declined an $800 million takeover offer from Meta Platforms Inc., opting instead to continue growing as an independent company, according to sources familiar with the matter.

Meta's Pursuit and AI Expansion Plans

Meta had been in acquisition talks with FuriosaAI since early 2025. While the discussions were kept private, FuriosaAI’s leadership ultimately chose to pursue an independent growth strategy.

Meta, under CEO Mark Zuckerberg, is investing up to $65 billion in 2025 toward building AI infrastructure, including custom chip development, large-scale data centers, and expanding its AI talent pool. Zuckerberg previously hinted that Meta could eventually spend hundreds of billions on AI infrastructure as part of its long-term strategy to stay competitive against OpenAI, Google, and new players like DeepSeek.

FuriosaAI’s Technology and Competitive Edge

Founded eight years ago and led by June Paik (a former executive at Samsung Electronics and AMD), FuriosaAI focuses on high-performance AI inference processors. Its latest innovation, the RNGD chip (pronounced “Renegade”), is built on TSMC’s 5nm process and incorporates HBM3 memory from SK Hynix.

The RNGD chip is FuriosaAI’s second-generation product and is positioned to compete with Nvidia, Groq, SambaNova Systems, and Cerebras Systems in the AI hardware sector.

Funding and Future Plans

FuriosaAI is currently in the process of closing an extended Series C funding round, which is expected to surpass its original funding target. The company intends to use this capital to scale operations and prepare for an initial public offering (IPO).

With about 150 employees, including a 15-member team in Silicon Valley, the company is currently providing chip samples to potential customers such as LG AI Research and Saudi Aramco. In total, around a dozen customers are evaluating its chips in the first half of 2025.


Read More: Vivo Unveils Vivo Vision MR Headset, Launches Robotics Lab to Expand AI Footprint