# FuriosaAI raises $620M Series D at a $2.9B valuation

> The Korean inference chip maker will scale its RNGD chip, already in mass production, and fund a third-generation design with Broadcom.

URL: https://claudegpt.si/news/furiosaai-620m-series-d  
Published: 2026-10-29T06:50:17.526Z  
Author: ClaudeGPT Si Newsroom  
Section: AI News  
Topics: FuriosaAI, AI chips, Funding, Inference, Nvidia  
Publisher: ClaudeGPT Si

## Key takeaways

- FuriosaAI closed a $620 million Series D at a $2.9 billion post-money valuation on October 28.
- Korea's National Growth Fund put in about $280 million, the largest single check.
- The RNGD chip has been in mass production since January 2026, with 20,000 cards planned this year.

FuriosaAI, a South Korean maker of AI inference chips, closed a $620 million Series D at a $2.9 billion post-money valuation, [the company said on October 28](https://furiosa.ai/blog/furiosaai-620-million-series-d). The FuriosaAI Series D is meant to move its RNGD chip from first deployments to volume, and to pay for a third-generation design.

The round matters because FuriosaAI is one of the few startups selling an inference chip that is already in mass production. It competes with NVIDIA on a narrow claim: more answers per watt, at a time when data center power is the limit for many buyers.

## Who put in the money

The largest check came from South Korea's National Growth Fund, at KRW 370 billion, about $280 million, according to FuriosaAI. Other state-backed investors include the Export-Import Bank of Korea and Korea Development Bank.

International investors include Cisco Investments, Infinitum, Jericho Capital, NeoFort Holding, QUDA Capital and Worldline Capital. Hanwha Asset Management and UAMCO are new. NAVER Ventures, Kakao Investment and DSC Investment are among those returning.

## What the chip is and who uses it

RNGD is an inference accelerator, a chip built to run trained models rather than train them. FuriosaAI says it reached mass production in January 2026. The company plans to make 20,000 cards in 2026 and says it has secured memory and fab capacity to more than double that in 2027.

Samsung SDS sells a service built on RNGD, which FuriosaAI calls NPU-as-a-Service. LG Group companies are deploying it in their own data centers. The company also reports use by AI cloud providers, sovereign AI programs and AI-native companies, without naming them.

## What the money pays for

FuriosaAI says the funds go to scaling RNGD production and deployment, and to a third-generation chip developed with Broadcom and other partners. It describes Gen 3 as a 2nm, multi-die design with HBM4 or HBM4E memory. No date for Gen 3 is given.

Yong Jung Park of NAVER Ventures, an existing investor, said investors look for [the moment a company moves "from proving its technology to proving its business"](https://furiosa.ai/blog/furiosaai-620-million-series-d).

## What we don't know yet

FuriosaAI published no benchmark numbers or revenue in the announcement. SiliconANGLE reports that the chips use less than half the power of NVIDIA GPUs, but that is a claim from the company's pitch, not an independent test. [SiliconANGLE's report](https://siliconangle.com/2026/10/28/energy-efficient-inference-chip-developer-furiosaai-raises-620m-to-solve-the-data-center-power-bottleneck/) also gives a 2027 target of 40,000 chips, which the company's own post does not state.

Watch for named customers beyond Samsung SDS and LG, and for published tokens-per-watt results on current models.

## Sources

- [FuriosaAI closes $620 million Series D to scale efficient, production-ready AI inference](https://furiosa.ai/blog/furiosaai-620-million-series-d), FuriosaAI
- [Energy-efficient inference chip developer FuriosaAI raises $620M](https://siliconangle.com/2026/10/28/energy-efficient-inference-chip-developer-furiosaai-raises-620m-to-solve-the-data-center-power-bottleneck/), SiliconANGLE