South Korean fabless AI chip startup Rebellions has secured $400 million in a pre-IPO funding round, pushing its valuation to approximately $2.34 billion. The massive capital injection signals growing investor confidence in alternative chip makers challenging NVIDIA's long-standing grip on the AI hardware market.
The round was led by Mirae Asset Financial Group and the Korea National Growth Fund. It comes just months after the company completed a $250 million Series C in November, and follows a $124 million Series B closed in 2024. With this latest raise, Rebellions' total funding now stands at $850 million — $650 million of which was raised in just the last six months.
The funding arrives as the company pursues an aggressive global expansion strategy and prepares for a public listing later this year.
What Rebellions Does
Founded in 2020, Rebellions designs AI chips while outsourcing their manufacturing — a model known as fabless. The startup's chips are purpose-built for inference, the computational process that allows trained AI models to respond to user queries in real time. Inference has grown in importance as large language models have matured and entered widespread commercial deployment — making fast, efficient, and cost-effective inference hardware a top priority for tech companies worldwide.
CEO and co-founder Sunghyun Park described the shift in the industry's priorities. He noted that AI is now evaluated based on its ability to function at scale in real-world conditions, under power constraints, and with a clear economic return. According to Park, this shifts the center of gravity toward inference infrastructure and the software that makes it usable.
New Products Announced
Alongside the funding announcement, Rebellions unveiled two new products: RebelRack and RebelPOD, both described as AI infrastructure platforms. RebelPOD is a production-ready unit of inference compute designed for immediate deployment. RebelRack, on the other hand, integrates multiple racks into a scalable cluster built for large-scale AI operations.
These products represent Rebellions' push beyond individual chip design into full-stack inference infrastructure — a move that could make it more competitive against established players offering end-to-end solutions.
Global Expansion in Full Swing
Rebellions' Chief Business Officer Marshall Choy, who leads the company's international growth, told TechCrunch that the company has recently established entities in the United States, Japan, Saudi Arabia, and Taiwan.
In the U.S., Rebellions is building out its ecosystem of technology partners and plans to target cloud providers, government agencies, telecom operators, and Neoclouds — the new wave of cloud infrastructure companies emerging to serve AI-specific workloads.
Choy declined to comment on specific IPO timing, though the company has publicly acknowledged its intention to go public this year.
Challenging NVIDIA's Dominance
Rebellions belongs to a new generation of chip startups seeking to challenge NVIDIA's once-dominant position in the AI chip market. While NVIDIA remains the clear leader, its hold has begun to face real pressure from multiple directions.
Major technology companies including AWS, Meta, and Google have also invested heavily in developing their own custom AI chips, further fragmenting a market that was once almost entirely NVIDIA's territory. This trend has created an opening for startups like Rebellions that can offer specialized, cost-efficient alternatives focused on inference rather than training.
Why It Matters
The AI chip market is at an inflection point. As AI models become more commercially embedded — powering everything from customer service chatbots to autonomous systems — the demand for efficient inference computing is skyrocketing. Training a model is a one-time investment, but inference runs continuously, making it the dominant long-term cost for any AI deployment.
Rebellions' rapid fundraising pace, expanding global presence, and new infrastructure products suggest it is positioning itself not just as a chip designer but as a full-scale inference platform company. With an IPO on the horizon and $850 million in the bank, the startup is making a strong case that the future of AI hardware will not belong to NVIDIA alone.
