NVIDIA, a global leader in advanced computing technologies, has joined forces with premier computer manufacturers to introduce cutting-edge systems powered by NVIDIA’s Blackwell architecture. These systems feature Grace CPUs and NVIDIA networking infrastructure, aimed at supporting enterprises in building AI factories and data centers. This collaborative initiative is poised to drive the next wave of generative AI breakthroughs, as reported by NVIDIA Newsroom.

Highlights from COMPUTEX Keynote

During the recent COMPUTEX keynote, NVIDIA founder and CEO Jensen Huang announced strategic partnerships with key industry players. Companies like ASRock Rack, ASUS, GIGABYTE, Ingrasys, Inventec, Pegatron, QCT, Supermicro, Wistron, and Wiwynn are gearing up to deliver cloud, on-premises, embedded, and edge AI systems utilizing NVIDIA GPUs and networking infrastructure.

Huang underscored the onset of a new industrial revolution, where NVIDIA is collaborating with companies and nations to transform traditional data centers into accelerated computing hubs known as AI factories. These AI factories are set to revolutionize the production of artificial intelligence on a large scale.

Comprehensive Offerings for Diverse Applications

The offerings extend from single to multi-GPUs, x86 to Grace-based processors, and air to liquid cooling technologies. The new NVIDIA GB200 NVL2 platform, part of the Blackwell product lineup, caters to mainstream large language model inference, retrieval-augmented generation, and data processing needs. Leveraging high-bandwidth memory performance and advanced decompression engines, Blackwell architecture significantly boosts data processing speed and energy efficiency compared to traditional x86 CPUs.

Modular Reference Architecture for Accelerated Computing

The NVIDIA MGX™ modular reference design platform enables manufacturers to swiftly develop over 100 system design configurations cost-effectively. More than 90 systems from over 25 partners are already released or in development, showcasing the scalability and efficiency of the MGX architecture. Notably, AMD and Intel are in line to support MGX with their CPU host processor module designs, streamlining development processes and ensuring consistent performance.

Ecosystem and Partner Contributions

NVIDIA’s extensive partner ecosystem includes key players like TSMC and global electronics manufacturers, providing essential components for AI factories. The collaborative effort ensures rapid development and deployment of new data center infrastructure, supported by Blackwell technology and NVIDIA’s advanced networking solutions like Quantum-2 InfiniBand, Spectrum™-X Ethernet, and BlueField®-3 DPUs.

Enterprises can leverage the NVIDIA AI Enterprise software platform, which includes NVIDIA NIM™ inference microservices, to create and deploy production-grade generative AI applications effectively.

Adoption in Taiwan

Taiwan’s leading companies have swiftly embraced Blackwell technology for integrating AI into their operations. Examples include Chang Gung Memorial Hospital, leveraging Blackwell for biomedical research advancements, and Foxconn’s plans for AI-powered electric vehicles, robotics, and generative AI services.

Image source: Shutterstock