FuriosaAI Rejects Meta’s $800M Bid to Double Down on Custom AI Chips

FuriosaAI, a rising South Korean AI chip startup, has turned down an $800 million offer from Meta to preserve its independence and pursue its chip development plans. The move highlights shifting power dynamics in the AI hardware race and underscores how startups are now willing to challenge Big Tech in one of the most strategic sectors of the digital economy.


Devdiscourse News Desk | Updated: 24-03-2025 16:00 IST | Created: 24-03-2025 16:00 IST
FuriosaAI Rejects Meta’s $800M Bid to Double Down on Custom AI Chips
Representative Image

In a move that’s turning heads across the tech world, South Korean semiconductor startup FuriosaAI has reportedly declined an $800 million acquisition offer from Meta. The social media giant, which is ramping up its investments in AI infrastructure, saw Furiosa as a key asset in its push to build custom chips and reduce dependence on dominant players like Nvidia. But sources familiar with the matter say the deal didn’t fall apart over money—it broke down over differences in vision.

While the offer represented one of the largest potential exits for an Asian AI startup, the real issue was how FuriosaAI would operate post-acquisition. According to The Chosun Ilbo, disagreements surfaced around business strategy and organizational structure. FuriosaAI was reportedly unwilling to cede control of its innovation roadmap or be subsumed into Meta’s broader operations. It’s a sign of a changing tide—where ambitious, well-funded startups are increasingly willing to walk away from tech giant money if it means protecting their long-term autonomy.

Meta’s interest in Furiosa isn’t hard to understand. The company has been on a mission to cut its reliance on Nvidia, whose powerful GPUs are the backbone of AI training but are increasingly hard to come by. Supply chain bottlenecks, high prices, and fierce competition for chips have left Meta—and other cloud and AI leaders—scrambling for alternatives. That’s why Meta has been developing its own AI silicon and pledged to invest as much as $65 billion in AI infrastructure in 2025 alone. Still, building something on par with Nvidia’s H100 or AMD’s Instinct MI300 chips is no easy feat, which is why acquisitions like the one proposed to FuriosaAI are so critical to Meta’s long-term AI strategy.

Founded in 2017 by June Paik, a former engineer at Samsung and AMD, FuriosaAI has positioned itself as one of the most promising challengers in the AI chip sector. The startup has developed two custom processors—Warboy and Renegade (RNGD)—designed to accelerate AI workloads, from image processing to large language model inference. RNGD, in particular, is aimed at “reasoning” models, the type of computation needed when AI systems respond to queries or perform real-time tasks after training.

Far from retreating after the failed Meta talks, Furiosa is doubling down on its roadmap. The company is currently discussing raising around $48 million (KRW 70 billion) to fund the rollout of the Renegade chip. Testing of RNGD has already been completed with LG AI Research and Saudi energy giant Aramco, two high-profile partnerships that signal confidence in the chip’s performance. LG AI Research plans to integrate RNGD into its infrastructure later this year, and FuriosaAI expects to launch the chip commercially before the end of 2025.

The broader implications of Furiosa’s decision are significant. As demand for generative AI applications explodes, compute power has become the new gold. And right now, Nvidia controls most of that supply. But startups like FuriosaAI are emerging as viable alternatives—not just for their technology, but for their willingness to stay independent and chart a different course.

This moment is about more than just one deal gone cold. It’s a snapshot of an evolving tech landscape, where custom chips are becoming a central battleground and independence can be just as valuable as a billion-dollar exit. FuriosaAI seems to understand that—and for now, it’s choosing to bet on itself rather than sell to one of the biggest names in Silicon Valley.

Give Feedback