Skip to content

Marvell Partners with Rebellions on Development of Personalized AI Hardware for Massive Independence-Level Implementations

South Korean AI semiconductor pioneer, Rebellions Inc., unveils partnership with an undisclosed entity on July 29, 2025.

Custom AI Infrastructure Partnership Announced: Rebellions Teaming Up with Marvell for...
Custom AI Infrastructure Partnership Announced: Rebellions Teaming Up with Marvell for Sovereign-Scale Deployments

Marvell Partners with Rebellions on Development of Personalized AI Hardware for Massive Independence-Level Implementations

Rebellions Inc., a South Korean AI semiconductor company, has partnered with Marvell Technology to develop custom AI infrastructure tailored for sovereign-backed AI initiatives and regional cloud providers in the APAC and Middle East regions.

This collaboration aims to offer high-performance, energy-efficient AI systems that move away from generic GPU-based architectures towards domain-specific custom ASIC-based solutions. Rebellions will leverage Marvell’s advanced silicon platforms, which include state-of-the-art packaging, high-speed SerDes (serializer/deserializer) technology, and die-to-die interconnects, to produce tightly integrated, end-to-end rack solutions optimized for scalable AI inference at sovereign scale.

The flagship chip REBEL from Rebellions uses chiplet architecture and 144GB HBM3E memory for massive-scale AI inference. Rebellions specializes in developing AI accelerators optimized for Large Language Models and large-scale inference, known for industry-leading energy efficiency.

The partnership reflects a strategic shift in the AI hardware market towards bespoke domain-specific architectures designed for local deployment rather than off-the-shelf, one-size-fits-all hardware typically used by global hyperscalers. This shift is being driven by global hyperscalers, sovereign-backed initiatives, and regional cloud providers seeking infrastructure that delivers scalability, efficiency, and control.

Will Chu, Senior Vice President and General Manager of Custom Cloud Solutions at Marvell, expressed excitement about collaborating to offer next-generation custom AI infrastructure. Sunghyun Park, CEO of Rebellions, stated that organizations are no longer settling for one-size-fits-all solutions.

Rebellions' AI accelerator ATOM has mass-production experience since 2024. The company is strengthened by its merger with SK SAPEON, making it Asia's leading independent AI semiconductor platform. Strategic investors in Rebellions include SK Telecom, SK hynix, Aramco’s Wa’ed Ventures, and KT.

The press release does not assume any obligation to update or revise any forward-looking statements. However, these statements involve risks and uncertainties that may differ materially from actual events, results, or achievements.

Marvell and the M logo are trademarks of Marvell or its affiliates. Other names and brands mentioned may be claimed as the property of others. The custom AI infrastructure will be tailored for performance, efficiency, and scalability, with Rebellions' platform including software for seamless data center integration.

This collaboration takes place amid evolving geopolitical and trade landscapes, including shifting U.S.-China tech policies, positioning Marvell and Rebellions to expand their influence in tech-hungry APAC and Middle Eastern markets. While promising in innovation and strategic market reach, the partnership, like others in AI, faces execution challenges inherent in delivering cutting-edge custom infrastructure.

Technology is at the heart of Rebellions' collaboration with Marvell Technology, as they aim to develop custom AI infrastructure specialized for sovereign-backed AI initiatives and regional cloud providers. This partnership is set to produce high-performance, energy-efficient AI systems that depart from generic GPU-based architectures, moving towards domain-specific, custom ASIC-based solutions, such as Rebellions' flagship chip REBEL.

Read also:

    Latest