South Korea’s FuriosaAI Raises 5M to Scale Custom AI Inference Chips

In a major boost for Asia’s AI semiconductor industry, Seoul-based chip startup FuriosaAI has raised 5 million in a Series C bridge round to ramp up production of its RNGD (“Renegade”) AI inference chip and accelerate next-generation chip development.

The round, led by Korean state-run financial institutions, values the company at approximately 5 million, bringing its total funding to 6 million.

Why It Matters

As large language models (LLMs) and AI workloads surge, so does the demand for specialized chips that can run inference efficiently at scale. FuriosaAI’s RNGD chip promises to deliver just that.

RNGD reportedly delivers 2.25× better performance-per-watt for AI inference tasks compared to leading GPUs.

This makes RNGD a strong contender in a market traditionally dominated by NVIDIA’s GPUs — particularly for power-hungry generative AI models.

Key Highlights

• Funding amount: 5 million (Series C bridge)
• Total capital raised: 6 million
• Valuation: ~5 million
• Lead investors: Korean Development Bank, Industrial Bank of Korea, and other national institutions
• Use of funds: Scale RNGD chip production, expand R&D for next-gen accelerators
• Major client: LG AI Research has already signed a design contract

Asia’s Bid in the AI Chip Race

FuriosaAI is part of a growing wave of Asian startups aiming to reduce reliance on U.S. chipmakers by developing competitive, homegrown AI silicon.

With NVIDIA and AMD chips in short supply and power-hungry, custom inference chips like RNGD offer:

• Lower energy costs
• Optimized performance for specific workloads (e.g. LLMs, vision, edge AI)
• Faster deployment cycles for AI services

By securing state backing and major design wins, FuriosaAI is positioning itself as South Korea’s AI hardware champion.

What’s Next?

The new capital will support:

• Mass production of RNGD to meet increasing global demand
• Hiring and infrastructure expansion for AI-focused chip design
• Development of next-gen inference chips to stay ahead of evolving AI model architectures

FuriosaAI’s roadmap includes moving beyond cloud inference into edge and embedded AI markets, where performance-per-watt and latency are mission-critical.

Final Thoughts

As AI continues its global ascent, hardware is the new frontier. FuriosaAI’s 5M funding round signals investor confidence in custom, efficient, and scalable AI inference chips — and South Korea’s ambition to be a major player in this space.

Stay with Cortex Hub for more insights on AI hardware trends and the innovators shaping the next wave of computing.