FuriosaAI Secures $125M for Next-Gen AI Chip Production Scale
Seoul, South Korea — FuriosaAI, a semiconductor company focused on AI compute, has successfully closed a $125 million Series C bridge funding round. This significant investment is set to accelerate the mass production of its next-generation AI inference chip, RNGD (pronounced “Renegade”), and support its global expansion efforts. The funding round brings the company's total capital raised to $246 million, valuing FuriosaAI at approximately $735 million.
The investment comes at a time of soaring global demand for high-performance and energy-efficient AI infrastructure. FuriosaAI aims to address the critical challenges of unsustainable power consumption, high operating costs, and infrastructure limitations often associated with traditional GPU-based hardware for AI applications. CEO and co-founder June Paik emphasized the company's commitment to making AI truly sustainable, both environmentally and economically.
A key factor in FuriosaAI's recent momentum is a major design win with LG AI Research. LG AI Research has adopted FuriosaAI's RNGD AI accelerator for its EXAONE foundation models after rigorous testing showed the RNGD chip achieved 2.25 times better Large Language Model (LLM) inference performance per watt compared to GPUs. This partnership, announced in July 2025, positions FuriosaAI as a compelling alternative to dominant players like Nvidia in the AI chip market, particularly within the South Korean AI ecosystem. LG AI Research plans to offer servers powered by RNGD to enterprises utilizing its EXAONE models.
The RNGD chip, designed with a thermal design power (TDP) of just 150 watts, aims to be highly power-efficient for data center inference with computationally demanding AI models. It offers native PyTorch 2.x support and delivers 256/512/1024 TOPS (BF16/FP8 or INT8/INT4). The chip incorporates HBM3 memory from SK Hynix and is manufactured using TSMC's 5-nanometer process, ensuring high performance and efficiency for LLM and multimodal inference workloads.
To further bolster its growth and technological development, FuriosaAI has strategically strengthened its executive team. Jeehoon Kang, an expert in parallel systems from KAIST, has joined as Chief Research Officer to lead compiler and software architecture. Additionally, Youngjin Cho, a silicon and SoC expert formerly of Samsung Electronics, has been appointed Vice President of Hardware to accelerate the company's chip development.
The global AI inference chip market is experiencing explosive growth, estimated at $15 billion in 2025 and projected to reach approximately $75 billion by 2033, with a robust Compound Annual Growth Rate (CAGR) of 25%. This surge is driven by the proliferation of edge AI deployments, advancements in deep learning models, and increasing AI adoption across various sectors like autonomous vehicles, healthcare, and smart manufacturing. The broader AI chip market is forecasted to grow by $902.64 billion during 2024-2029, accelerating at a remarkable CAGR of 81.2%. The semiconductor industry as a whole is undergoing a significant transformation, with AI emerging as a powerful catalyst for growth, driving demand for high-performance computing, memory, and custom silicon. This trend is leading to a focus on energy-efficient designs to minimize environmental impact and operating costs.
FuriosaAI's decision to remain independent and pursue its mission, even turning down an $800 million acquisition offer from Meta Platforms earlier in 2025, underscores its commitment to revolutionizing AI computing with its specialized hardware. With the new funding, FuriosaAI is poised to accelerate the mass production of RNGD for global enterprise customers and lay the groundwork for its next-generation chip, contributing to a more sustainable and accessible future for AI.