HomeTechnologyAI chip startup FuriosaAI rejects Meta’s $800 million offer

AI chip startup FuriosaAI rejects Meta’s $800 million offer

Meta had been in discussions about acquiring Seoul-based FuriosaAI since the start of this year

March 24, 2025 / 11:49 IST
Story continues below Advertisement
Meta is investing heavily in artificial intelligence infrastructure, seeking to better compete in a fast-moving race against the likes of OpenAI and Google as well as upstarts such as Hangzhou-based DeepSeek
Meta is investing heavily in artificial intelligence infrastructure, seeking to better compete in a fast-moving race against the likes of OpenAI and Google as well as upstarts such as Hangzhou-based DeepSeek

Korean chip startup FuriosaAI has turned down an $800 million takeover offer from Meta Platforms Inc., choosing instead to grow the business as an independent company, a person with knowledge of the matter said.

Meta had been in discussions about acquiring Seoul-based FuriosaAI since the start of this year, said the person, who asked not to be identified as the matter is private. A representative for FuriosaAI declined to comment. Meta officials couldn’t immediately be reached for comment outside regular business hours on Sunday.

Story continues below Advertisement

Meta is investing heavily in artificial intelligence infrastructure, seeking to better compete in a fast-moving race against the likes of OpenAI and Google as well as upstarts such as Hangzhou-based DeepSeek. In mid-January, Chief Executive Officer Mark Zuckerberg said that Meta would spend as much as $65 billion this year, including outlays to build a large data center and expand its AI workforce. Only a week later, Zuckerberg told investors that Meta anticipates eventually spending hundreds of billions of dollars on AI infrastructure.

The Menlo Park, California-based company is also working on its own chips designed for its own AI workloads, including powering the ranking and recommendations ads on Facebook and Instagram. In 2023, it introduced its first custom AI inference chips — and it unveiled an upgraded version last year.