Meta is reportedly in talks to acquire South Korean AI chip startup FuriosaAI.
Citing unnamed sources familiar with the matter, Forbes reported that the deal could close as early as this month, adding that Meta was not the only company interested in purchasing the chip company.
Meta did not respond to Forbes’ request for comment, while FuriosaAI declined to comment on the “speculation.”
Founded in 2017 and headquartered in South Korean capital Seoul, FuriosaAI has raised approximately $115 million across four funding rounds to support the development of its RNGD chip. The AI inference chip has a thermal design power (TDP) of 150W, but the company claims that when compared to Nvidia’s H100 GPUs – which has a TDP of 350W – RNGD offers a 3x better performance per watt.
RNGD is planned to enter mass production in the second half of this year.
It has long been known that Meta has been looking to develop its own chips in order to reduce its reliance on Nvidia hardware. In February 2024, according to documents seen and reported on by Reuters, the company was planning to deploy the second generation of Meta Training and Inference Accelerator (MTIA) chip.
First reported to be in development in 2023, the chips, dubbed the Meta Training and Inference Accelerator, are based on 7nm nodes and provide 102 Tops of Integer (8-bit) accuracy computation or 51.2 teraflops of FP16 accuracy computation. The chips run at 800 megahertz and are about 370 millimeters square.
Meta was originally expected to roll out its in-house chips in 2022 but scrapped the plan after they failed to meet internal targets, with the shift from CPUs to GPUs for AI training forcing the company to redesign its data centers and cancel multiple projects.
Later that same month, Meta posted job adverts looking for ASIC engineers to help the company build data center accelerators and system-on-a-chips (SoC), although the listings did not make reference to any specific project.
Meta is not the only company looking to reduce its reliance on Nvidia by developing its own chips. Earlier this week, it was reported that OpenAI is finalizing the design for its first custom AI training chip, which will be fabricated by TSMC and made using the chipmaker’s 3nm technology.
