总部位于首尔的人工智能推理芯片制造商FuriosaAI在D轮融资中筹集了6.2亿美元。该轮融资将其估值定在约29亿美元。韩国国家成长基金投入了2.8亿美元，成为最大投资者。
FuriosaAI, a Seoul-based maker of AI inference chips, has raised $620m in a Series D round. The round values it at about $2.9bn. South Korea’s National Growth Fund put in $280m, making it the largest investor.
该公司在新闻稿和博客文章中宣布了这轮融资。去年的上一轮融资将其估值定在7.35亿美元，因此其估值已翻了四倍。据彭博社记者李有林报道，彭博社率先报道了这一消息。
The company announced the round in a press release and a blog post. Its previous round last year valued it at $735m, so its valuation has quadrupled, Yoolim Lee reported for Bloomberg, which had the news first.
“随着我们在全球范围内扩大业务，我们看到了强劲的需求和机遇，”首席执行官郑恩珠周四在彭博电视台表示。
“We see strong demand and opportunities as we are scaling our business globally,” chief executive June Paik said on Bloomberg Television on Thursday.
谁投资了 国际投资者向FuriosaAI的D轮融资投入了1.1亿美元。美国公司Infinitum是其中最大的投资者。Jericho Capital、Worldline Capital、Cisco Investments、阿布扎比的NeoFort Holding和新加坡的QUDA Capital Partners也参与了投资。QUDA的创始人Charlie Foo是前Nvidia高管，目前担任FuriosaAI的顾问。
Who invested International investors put $110m into the FuriosaAI Series D. US firm Infinitum was the largest of them. Jericho Capital, Worldline Capital, Cisco Investments, Abu Dhabi’s NeoFort Holding and Singapore’s QUDA Capital Partners joined it. QUDA’s founder, Charlie Foo, is a former Nvidia executive and now advises FuriosaAI.
老股东追加了1.25亿美元，包括全部三家种子投资者：NAVER、DSC Investment和KDB Capital。韩国进出口银行首次直接投资一家初创公司，资产管理公司UAMCO承诺投资3000万美元。
Existing shareholders added $125m, including all three seed investors: NAVER, DSC Investment and KDB Capital. The Export-Import Bank of Korea made its first direct investment in a startup, and asset manager UAMCO committed $30m.
“将一款先进的人工智能芯片从设计推向量产是一项极其艰巨的任务，很少有公司证明自己具备成功做到这一点的能力，”Infinitum首席投资官John Yetimoglu表示。
“Taking an advanced AI chip from design to volume production is an exceptionally difficult undertaking, and few companies have demonstrated the ability to do so successfully,” said John Yetimoglu, chief investment officer at Infinitum.
芯片及其客户 Nvidia的芯片主导着人工智能数据中心，投资者正在寻找耗能更低的替代方案，彭博社报道称。FuriosaAI制造用于推理的芯片，即运行已训练的人工智能模型的工作，而非训练模型。
The chip and its customers Nvidia chips dominate AI data centres, and investors are looking for alternatives that use less energy, Bloomberg reported. FuriosaAI builds chips for inference, the work of running trained AI models, rather than training them.
其第二代芯片RNGD（发音为“renegade”）于1月进入量产。据彭博社报道，台积电今年将生产2万颗该芯片。FuriosaAI表示，已确保内存和产能，可在2027年生产超过4万颗。整个行业内存均告短缺：美光预计内存短缺将持续到2028年。
Its second-generation chip, RNGD, pronounced “renegade”, went into mass production in January. TSMC is making 20,000 of them this year, according to Bloomberg. FuriosaAI says it has secured the memory and factory capacity to make more than 40,000 in 2027. Memory is scarce across the industry: Micron expects the memory shortage to tighten into 2028.
RNGD卡耗电180瓦，仅需风冷即可。FuriosaAI表示，其吞吐量与英伟达RTX PRO 6000卡相当。该公司还称，RNGD每千瓦可服务的用户数量是前者的两倍。TNW尚未测试这些说法。
The RNGD card draws 180 watts and needs only air cooling. FuriosaAI says it matches the throughput of Nvidia’s RTX PRO 6000 card for card. It also says RNGD serves twice as many users per kilowatt. TNW has not tested these claims.
三星SDS于7月推出了基于RNGD运行的云服务。据FuriosaAI称，LG集团在其芯片上运行EXAONE模型。韩国AI公司Upstage使用这些芯片进行搜索摘要。Paik告诉彭博社，该公司正在与亚太地区和中东的许多客户洽谈。
Samsung SDS launched a cloud service running on RNGD in July. LG Group runs its EXAONE models on the chips, according to FuriosaAI. Korean AI company Upstage uses them for search summaries. Paik told Bloomberg the company is talking to many customers in Asia-Pacific and the Middle East.
资金用途 FuriosaAI将使用这笔资金生产更多RNGD芯片，改进其软件，并开发其第三代芯片。FuriosaAI正与博通及其他合作伙伴开发该芯片。该公司表示，该芯片将采用2纳米、多芯片设计，并配备HBM4内存。
What the money is for FuriosaAI will use the money to make more RNGD chips, improve its software and build its third-generation chip. FuriosaAI is developing that chip with Broadcom and other partners. It will use a 2nm, multi-die design with HBM4 memory, the company says.
该公司在美国、新加坡和里斯本设有办事处。在里斯本，它于7月与Equinix合作将RNGD带入欧洲。今年，其他推理芯片制造商也筹集了大额资金，包括Etched在8月筹集了210亿美元。
The company has offices in the United States, Singapore and Lisbon. In Lisbon, it brought RNGD to Europe with Equinix in July. Other inference chipmakers have also raised large rounds this year, including Etched at $21bn in August.