FuriosaAI raises $620M at $2.9B valuation to challenge Nvidia

South Korean AI chip startup FuriosaAI has raised $620 million in a funding round that values the company at $2. 9 billion, positioning itself as a direct competitor to Nvidia in the race to power energy-efficient artificial intelligence workloads. The round was anchored by South Korea’s National Growth Fund, which invested $280 million as part of a broader government initiative to bolster the domestic semiconductor sector. Additional funding came from global investors such as Cisco Investments, Abu Dhabi-based NeoFort Holdings, and QUDA Capital Partners, with existing backers like Korea Development Bank, Kako Investment, and NAVER also participating.

The company’s appeal lies in addressing a key limitation of current AI data centers: power consumption. While Nvidia’s GPUs lead the market, their energy demands have strained infrastructure, creating scalability challenges for data centers. FuriosaAI’s chips are designed to optimize efficiency, focusing on maximizing output per watt of power used. Its Tensor Contraction Processor (TCP), launched earlier this year, employs a three-dimensional matrix architecture—unlike Nvidia’s two-dimensional approach—to reduce energy-intensive data movement.

FuriosaAI has already gained traction with major South Korean companies. LG Group uses its chips to power the EXAONE AI models, Upstage deploys them for live search summaries, and Samsung Electronics offers access through its neural processing units-as-a-service platform. With the new funding, the startup plans to expand production, targeting 20,000 chips this year and 40,000 in 2027. It is also developing a third-generation processor in collaboration with Broadcom.

Co-founder and CEO June Paik described the funding as a validation of the company’s technology, emphasizing its goal to "shape the next generation of AI infrastructure." NAVER Ventures partner Yong Jung Park, an early investor, noted that FuriosaAI has successfully transitioned from proving its architecture to demonstrating a commercial product—a milestone for semiconductor startups. The startup’s focus on energy efficiency aligns with industry concerns about data center power consumption, a challenge that has attracted funding for other startups like Salience Labs and Delos Data in recent months.

The funding round underscores investor interest in alternatives to Nvidia’s dominance, particularly as power constraints become a growing concern for data centers. While Nvidia’s GPUs remain the industry standard, their energy demands have opened opportunities for startups offering more efficient solutions. FuriosaAI’s valuation reflects optimism about its potential, though its long-term success will depend on scaling production and expanding its customer base beyond South Korea. The partnership with Broadcom could further strengthen its position, but the broader market’s willingness to prioritize efficiency over performance will ultimately determine its impact.

Sources: siliconangle.com