Rebellions: Powering the AI Inference Revolution

Share this article
Share this article
Prioritise Us on Google
Sunghyun Park, Co-Founder and CEO at Rebellions. Picture: Rebellions
Rebellions, a global leader in AI inference infrastructure, recently announced a funding round to the tune of US$400m – bringing its valuation to US$2.34bn

The narrative is shifting in the realm of AI semiconductors. For years, the industry was obsessed with the "training" phase – the massive computational effort required to create models like GPT-4.

Now, the conversation is pivoting to a more practical challenge: inference. The question is how to run models in the real world, efficiently and at scale in data centre environments. 

Rebellions, a global leader in AI inference infrastructure, is on a mission to provide the answer. The company recently announced a pre-IPO funding round – led by Mirae Asset Financial Group and the Korea National Growth Fund – to the tune of US$400m. It follows a US$250m Series C round in September 2025, bringing Rebellions’ total funding to US$850m and its valuation to US$2.34bn.

Rebellions is a global leader in AI inference infrastructure. Picture: Rebellions

Moving beyond silicon

The core of Rebellions' thesis is that the next phase of AI adoption won't be won by hardware specs alone. As adoption accelerates, the bottleneck is moving from model capability to operational reality.

"AI is now measured by its ability to operate in the real world – at scale, under power constraints, and with clear economic return," explains Sunghyun Park, Co-Founder and CEO of Rebellions. "That shifts the centre of gravity toward inference infrastructure and software that makes that infrastructure usable."

Sunghyun's perspective reflects a cold truth: a chip is only as useful as a developer’s ability to deploy on it. While competitors often focus on proprietary "walled garden," Rebellions takes a software-centric approach. Its cloud-native AI stack works natively with the open-source ecosystem developers actually use, including vLLM, PyTorch, Triton, Hugging Face and OpenShift.

By aligning with these standards, Rebellions removes the friction of custom AI hardware. Developers can deploy across diverse environments without proprietary lock-in, making the transition from GPU setups to Rebellions’ NPU-based systems seamless.

Youtube Placeholder

US expansion

A significant portion of the US$400m cash injection is set to fuel Rebellions’ expansion into the US. The firm is targeting cloud providers, Neoclouds, telecom operators and government-backed initiatives.

Leading this charge is Chief Business Officer Marshall Choy, who recently joined to help drive global growth. According to Michelle, the US market has reached a tipping point where efficiency is as critical as raw power.

"Access to compute is no longer the only question – how efficiently that compute is used is becoming just as important," she says. 

"Organisations are looking for infrastructure that works within their existing environments, extends the life of their current investments and enables new revenue-generating AI applications. That is where we are focused today."

Marshall Choy, Chief Business Officer at Rebellions. Picture: Rebellions

From chips to modular systems

What's more, Rebellions is moving beyond being a component supplier to becoming a systems provider.

At the heart of its strategy is the Rebel100, a chiplet-based NPU optimised for performance-per-watt. To meet data centre demands for plug-and-play solutions, the company has introduced the RebelRack and RebelPOD.

The RebelRack delivers a production-ready unit of inference compute, while the RebelPOD integrates multiple racks into a scalable cluster.

These vertically-integrated platforms allow companies to operate heavy AI workloads within real-world power constraints, delivering modular infrastructure that can be easily replicated across global data centres

K-NVIDIA momentum

The involvement of the Korea National Growth Fund in Rebellions' latest funding round represents a major milestone.

Rebellions was selected as the fund's first investment under the K-NVIDIA initiative, a public-private programme designed to cultivate South Korea’s answer to global semiconductor dominance.

Rebellions provides AI hardware for data centres, including cards, servers and rack-scale solutions. Picture: Rebellions

This government backing, combined with the strategic support of Mirae Asset Financial Group – early investors in SpaceX – signals that Rebellions is viewed as foundational infrastructure.

"Amid the global race for AI leadership, venture capital plays a critical role in identifying and backing high-potential companies," states Eung-Suk Kim, Vice Chairman and CEO of Mirae Asset Venture Investment.

"We are proud to support Rebellions as a strategic partner in demonstrating its capabilities and value on the global stage."

The road to IPO

With fresh capital and a valuation north of US$2bn, the path to a future public offering looks clear. 

However, for Sunghyun, the goal is long-term durability.

"Our objective is to build a durable infrastructure company that enables the next phase of AI adoption," he adds. "That requires long-term investment in software, systems and ecosystem – in addition to hardware – and a clear focus on making AI deployable at scale."

Rebellions epitomises a new era of AI, where efficiency, open-source collaboration and real-world deployability are the true benchmarks of success.

Company portals

Executives