Artificial intelligence (AI) semiconductor startup Rebellions announced on the 29th that it will collaborate with the U.S. semiconductor corporation Marvell Technology (hereinafter referred to as Marvell) to jointly develop a customized AI system in response to the demand for 'sovereign AI' infrastructure in the Asia-Pacific (APAC) and Middle East regions. Sovereign AI refers to the establishment of an independent AI system by a nation using its own infrastructure, data, personnel, and business networks.
In addition to the demand for government-led AI projects, regional cloud corporations are also seeking next-generation AI systems to secure their own infrastructure with energy efficiency and scalability. The two companies plan to continue technological and business cooperation for providing customized AI infrastructure in the future. First, Rebellions will utilize Marvell's custom design platform to design customer-specific AI semiconductors for inference.
In particular, they will implement integrated AI infrastructure that extends beyond server units to rack level, based on advanced packaging technology, high-speed serial data transmission technology, and die-to-die interconnect technology possessed by Marvell. This aims to achieve both high performance and energy efficiency.
Park Sung-hyun, CEO of Rebellions, said, 'The AI infrastructure market is undergoing rapid changes, and it has now reached a point where it is difficult to respond with only generic solutions.' He noted, 'Through collaboration with Marvell, Rebellions will combine its expertise in AI semiconductor design and advanced semiconductor integration technology to provide customized AI infrastructure optimized for the realistic needs of governments and institutions in each country.'
Will Chu, senior vice president of Marvell's custom cloud solutions institutional sector, said, 'Custom AI infrastructure is a key element that will open a new trend in data center innovation,' and added, 'Through collaboration with Rebellions, we will provide next-generation AI infrastructure that possesses performance, efficiency, and scalability.'