South Korean chip startup Rebellions raised $400M in Series B to develop Nvidia alternatives.

Rebellions Inc., a Seoul-based cloud-native artificial intelligence inference chip startup, has raised $400 million in what it calls a "pre-IPO funding round," led by Mirae Asset Financial Group and the Korea National Growth Fund. This round comes just months after the company raised $250 million in a Series C round in September, bringing its total amount of funding since inception to $850 million and its valuation to $2 billion. The funding positions the company to compete with rivals such as Nvidia Corp. and Advanced Micro Devices Inc., alongside domestic South Korean competitors including Furiosa AI Inc. and DeepX Co. Ltd.

Rebellions designs high-performance inference chips that run trained AI models in production, targeting a market where Nvidia still dominates but faces mounting pressure from competitors such as Cerebras Systems Inc. The startup aims to address a critical pain point in AI adoption: as generative AI models increase in complexity, the cost of running them has increased dramatically, particularly when using graphics processing units. Nvidia's GPUs may be powerful, but they consume vast amounts of energy and are prohibitively expensive at scale for many companies. The company plans to use the funding to expand into the U.S. market and has hired Marshall Choy as its new chief business officer to lead operations in that country.

Rebellions' strategy centers on a "software-centric" approach to AI hardware, with co-founder and Chief Executive Sunghyun Park arguing that it's time for the industry to move beyond the "silicon-only" mindset. As he stated, "AI is now measured by its ability to operate in the real world at scale, under power constraints and with clear economic return. That shifts the center of gravity toward inference infrastructure and software that makes the infrastructure useful." The company's chip hardware is built atop a cloud-native stack powered by Kubernetes and is optimized to work with popular AI developer frameworks such as PyTorch, Hugging Face, and the vLLM inference engine, enabling its neural processing units to run inference more efficiently than traditional GPUs.

To advance this vision, Rebellions has launched two vertically integrated infrastructure offerings: RebelRack, described as a "production-ready" unit of inference compute, and RebelPOD, which clusters multiple RebelRacks together for larger deployments. Both systems are powered by its Rebel100 NPUs, chiplet-based processors optimized for superior performance per watt than GPUs. Choy emphasized the company's comprehensive systems approach, stating, "Our use of open-source software without forks enables an identical experience for admins, cloud ops, developers and end users. This enables our evaluation and adoption cycles to be much faster than competitive systems in the market today." On the prospect of breaking into the U.S. market, Choy noted that while some companies may have an "Nvidia-first bias, very few have an Nvidia-only policy," and he expects growing recognition that "the future of AI computing is going to be heterogeneous, with customers looking to match different accelerators and GPU architectures to particular classes of models, applications and use cases, as well as pricing tiers."

With $650 million raised in the last six months, Rebellions is accelerating its push to market. The company has previously stated it is working toward an initial public offering, and the current round's designation as a "pre-IPO" raise reflects those intentions. However, Chief Business Officer Choy declined to specify when the company might go public, emphasizing that the immediate focus is strictly on scaling up its business and growing its revenue.