Marvell Technology, Inc. has announced a collaboration with NVIDIA to bring NVIDIA´s NVLink Fusion technology to customers utilizing Marvell´s custom cloud platform silicon. The integration of NVLink Fusion aims to enhance the flexibility and scalability of cloud infrastructure, especially for organizations developing next-generation Artificial Intelligence applications. NVIDIA´s NVLink Fusion is designed to seamlessly connect custom XPU silicon with high-speed NVIDIA NVLink connectivity, enabling rack-scale architectures and advanced software solutions tailored for Artificial Intelligence workloads.
The Marvell custom platform strategy is centered around unique semiconductor designs and technological innovation. The company leverages its expertise in both system and semiconductor engineering, advanced manufacturing processes, and a broad suite of semiconductor platform solutions. These include offerings such as electrical and optical serializers/deserializers (SerDes), die-to-die interconnects for both 2D and 3D devices, advanced packaging, and silicon photonics. Other platform components featured by Marvell include co-packaged copper, custom high-bandwidth memory (HBM), system-on-chip fabrics, optical input/output solutions, and cutting-edge compute fabric interfaces, including PCIe Gen 7 capabilities.
By combining these advanced technologies with NVIDIA´s NVLink Fusion, Marvell is positioned to provide customers with highly tailored platforms that can dramatically improve infrastructure performance, efficiency, and value. The collaboration gives cloud and data center customers more configuration options and greater power to customize their Artificial Intelligence infrastructure, signifying a pivotal advancement in the evolution of cloud computing architectures and semiconductor integration for Artificial Intelligence-driven workloads.
