Korean AI semiconductor company Furiosa AI has closed a Series D round of $623 million (820 billion won), the company said on October 29, 2026. The round was priced at a pre-money valuation of $2.28 billion (3 trillion won), putting the post-money valuation at about $2.89 billion (3.8 trillion won).
The Round
Furiosa began the Series D in January with a target of $342–570 million (450–750 billion won). Strong interest from domestic and overseas investors pushed the final size above target. The company calls it one of the largest raises in Korean startup history and a notable one even by global standards.
- National Growth Fund: invested $281 million (370 billion won) and was the largest investor in the round. Furiosa says its commitment helped draw in global investors, as the government pushes to build up the AI semiconductor industry and national AI computing capacity.
- Overseas investors: the largest was US-based Infinitum, which co-led Groq’s $650 million round in June; its founder and CIO John Yetimoglu also sits on Groq’s board. Singapore’s QUDA Capital Partners joined; its founder Charlie Foo, formerly an NVIDIA and Mellanox executive, is Furiosa’s Principal Advisor. Others include Cisco Investments, Abu Dhabi’s NeoFort Holding, Jericho Capital and Worldline Capital.
- Existing shareholders: follow-on investments totalled $129 million (170 billion won). Korea Development Bank added $22.8 million, and seed investors NAVER (this time through NAVER Ventures), DSC Investment and KDB Capital all followed on. Nextrans made the largest follow-on at $36.5 million, and Kakao Investment, Quantum Ventures, Korea Investment Partners, Intellectual Discovery and others also took part. MJM Investment participated as GP of a fund backed by two major domestic institutions.
- New domestic investors: Korea Eximbank made its first direct venture investment since its founding. UAMCO invested $30.4 million, the largest among new domestic investors, and Hanwha Asset Management joined through its deep-tech fund.
Where the Money Goes
Furiosa will use the funds to expand the global commercialisation of RNGD, its second-generation AI chip, and to speed up a third-generation accelerator for hyperscale AI infrastructure, being developed with Broadcom.
RNGD entered mass production in January. Furiosa says it has since expanded into real enterprise and cloud settings through validation and cooperation with LG AI Research, the launch of an RNGD-based NPU-as-a-service by Samsung SDS, and RNGD supply to Upstage’s Daum search summary service, which serves millions of users. Proceeds will go mainly toward RNGD production and supply capacity, a broader global customer base and a stronger software ecosystem.
Investor and Founder Views
John Yetimoglu, CIO of Infinitum, said: “Successfully taking an advanced AI chip from design to mass production is an extremely difficult task, and very few companies have done it. Furiosa AI went a step further by deploying RNGD in the real service environments of major companies such as Samsung SDS and the LG Group, proving not only its technology but also its commercialization capability.”
Yongjung Park, General Partner at NAVER Ventures, said: “What we pay the most attention to in technology companies is the point where they move from proving the technology to proving the business, and Furiosa AI has reached exactly that point as chip mass production and customer adoption get under way.” NAVER Ventures, based in Silicon Valley, plans to help Furiosa expand in North America.
Junho Baek, CEO of Furiosa AI, said: “Now is the stage where we spread the technology we developed across the global AI infrastructure market. We will step up the global commercialization of RNGD and the development of our next-generation products, and offer a new option in the AI infrastructure market built on high-performance, high-efficiency AI chips.”
