Meta is reportedly in advanced discussions to acquire FuriosaAI, a South Korean AI chip startup, as part of its ongoing strategy to develop custom chips and reduce reliance on NVIDIA’s GPUs. The deal, which could be finalized this month, comes at a crucial time when tech companies are facing supply constraints due to the increasing demand for AI hardware. While Meta is a leading contender, other potential buyers are also exploring the acquisition.
Why FuriosaAI?
Founded in 2017, FuriosaAI has gained attention for its AI chip, RNGD, which is optimized for running large language models such as Meta’s Llama 2 and Llama 3. The chip is designed to deliver high efficiency while consuming less power compared to NVIDIA’s H100 GPUs, making it a cost-effective alternative for AI infrastructure.
The startup has raised approximately 170 billion won ($115 million) from notable investors, including South Korean tech giant Naver and venture capital firm DSC Investment. FuriosaAI’s CEO, June Paik, a former executive at Samsung and AMD, holds an 18.4% stake in the company. With strong backing and innovative technology, FuriosaAI has positioned itself as a key player in the growing AI chip sector.
South Korea’s AI Boom and Competitive Landscape
South Korea has been aggressively investing in AI and semiconductor technology, fostering a competitive environment for local startups. Recently, FuriosaAI’s rival, Rebellions, merged with SK Hynix-backed Sapeon, forming the country’s first AI chip unicorn. The consolidation of South Korean AI chip companies signals the nation’s ambitions to compete globally in the semiconductor market.
If Meta successfully acquires FuriosaAI, it could significantly advance its AI hardware capabilities and reduce reliance on NVIDIA’s expensive GPUs. The move aligns with Meta’s broader goal of building custom AI chips to power its large-scale AI models efficiently.
OpenAI and the Industry-Wide Shift Toward Custom AI Chips
Meta is not the only tech giant looking to develop proprietary AI chips. OpenAI is also working on its own custom chip, a project spearheaded by a 40-member engineering team led by Richard Ho, a former Google executive. OpenAI plans to finalize its chip design soon and send it to TSMC for manufacturing, with mass production expected by 2026. The new chip, developed in collaboration with Broadcom, will be used for training AI models and will feature cutting-edge 3-nm technology with fast memory and networking capabilities similar to NVIDIA’s hardware.
Tech leaders Microsoft and Meta have faced significant hurdles in developing AI chips, but the push toward custom silicon is accelerating. Both companies plan to ramp up their AI infrastructure spending next year, with Microsoft investing $80 billion and Meta allocating $60 billion. These investments highlight the industry’s collective effort to diversify beyond NVIDIA, which currently dominates the AI chip market with an 80% share.
What This Means for the AI Chip Industry
Meta’s potential acquisition of FuriosaAI underscores a growing trend: Major tech companies actively seek alternatives to NVIDIA’s high-cost GPUs. As South Korea strengthens its AI sector and companies like OpenAI push forward with proprietary chip designs, the global AI hardware landscape is shifting toward greater diversification and innovation.
If the deal goes through, Meta will not only secure a strategic advantage in AI chip development but also play a crucial role in reshaping the semiconductor industry. With AI-driven applications expanding rapidly, custom AI chips are set to become a cornerstone of future advancements in artificial intelligence.
Read more: Justin Hotard Takes Over as Nokia CEO, AI & Data Centers at Forefront
