Other
Mar 2025FuriosaAI reportedly rejected an approximately $800 million acquisition offer from Meta and remained independent; the report described disagreements over post-acquisition strategy and organizational structure.
SourceCompany research
Public company, workplace, funding, and market signals
Updated Aug 5, 2026
Overview
FuriosaAI is a Seoul-founded semiconductor company building energy-efficient AI inference accelerators and the supporting software stack for data centers, enterprise LLMs, multimodal workloads, and computer vision. Its mission is to make AI computing more sustainable through purpose-built silicon and deployment software.
Primary product
RNGD (Renegade), a TSMC 5nm AI inference accelerator based on Furiosa’s Tensor Contraction Processor architecture, available as RNGD PCIe cards and the NXT RNGD multi-card data-center server; the company also offers Vision NPU for computer vision.
Founded
2017
Headquarters
Seoul, South Korea
Team size
100–150 (LinkedIn estimate; company materials also describe a global team)
Work style
Flexible
Industry
Semiconductor manufacturing
Sub-industry
AI accelerators / neural-processing units for data-center inference
Offices
Seoul, South Korea (R&D headquarters)
Santa Clara, California, United States
Lisbon, Portugal
Germany
Singapore
Business model
Enterprise sale of AI accelerator cards and servers
Strategic hardware deployments and design wins
AI inference infrastructure for on-premise and cloud data centers
Software/toolchain supporting deployment, compilation, optimization, and serving
Funding
Stage
Late-stage / Series C bridge
Total raised
$246M
Latest round
Series C bridge · Jul 2025
Latest amount
$125M
Jul 2025 · Korea Development Bank
May 2021 · Aion Asset Management
People & investors
Leadership
Co-founder and Chief Executive Officer
Co-founder and Chief Technology Officer
Jeehoon Kang
Chief Research Officer
Youngjin Cho
Vice President of Hardware
Alex Liu
SVP, Product and Business
Boncheol Gu
Chief Software Architect
Work & culture
The company describes its culture as the “Furiosa Way”: shared vision, fierce execution, and “show, don’t tell,” with a first-principles engineering orientation and an emphasis on results over hype.
Work style
Flexible
Compensation
No reliable public compensation bands were found. Public job listings provide role and location information but not consistent salary or equity ranges.
Benefits
Publicly advertised roles indicate global and remote options in selected geographies
Global team distributed across Korea, the United States, Portugal, Singapore, and Germany
Recruiting across hardware, software, algorithms, product, sales, and operations
Product & market
Pricing
Enterprise hardware and infrastructure pricing; public list pricing was not found. Evaluation is available through the Furiosa Access Program.
Differentiators
- —Purpose-built for inference rather than general-purpose GPU workloads
- —Reported 2.25× better LLM inference performance per watt than LG AI Research’s GPU solution
- —Lower power and total-cost-of-ownership profile for air-cooled data centers
- —RNGD supports enterprise LLM and multimodal deployment
- —Integrated hardware/software stack with compiler and vLLM-compatible serving runtime
- —Standard PCIe and multi-card server deployment without liquid-cooling or facility retrofits
Technology
Tensor Contraction Processor (TCP) architecture
Scale & contact
Estimated revenue
$1M–$10M (LinkedIn estimate; not company-reported)
Signals & risks
Latest news
FuriosaAI · Jul 2026
FuriosaAI announced European RNGD availability and deployment at Equinix’s Lisbon data center, supporting enterprise evaluation and expansion.
FuriosaAI · Jul 2025
FuriosaAI announced a $125 million Series C bridge round, taking reported cumulative funding to $246 million, to scale RNGD production and develop its next-generation chip.
Bloomberg · Jul 2025
Bloomberg reported the $125 million financing and a latest valuation of approximately $735 million.
FuriosaAI · Jul 2025
LG AI Research adopted RNGD for EXAONE inference after testing, reporting 2.25× better performance per watt than its GPU solution.
Open Positions
56 jobs at FuriosaAI
Senior HW System Engineer
Remote
Hardware
Seoul HQ
1d ago
Remote
Full-time