Nvidia announced at the Computex event in Taiwan that several computer makers based there are planning next year to release the first wave of new systems using the Nvidia Grace CPU Superchip and Grace Hopper Superchip for a wide range of workloads spanning digital twins, AI, high performance computing, cloud graphics and gaming.
Starting in the first half of 2023, dozens of server models are expected to come from ASUS, Foxconn Industrial Internet, GIGABYTE, QCT, Supermicro and Wiwynn. The Grace-powered systems, along with the company’s x86 and other Arm-based servers, present Nvidia customers with a broad range of options for data center deployment, the company said.
“To address the demands of AI, data centers are transforming into AI factories, and this is opening up massive market opportunities for Nvidia and our ecosystem,” said Paresh Kharya, senior director of product management and marketing for accelerated computing at NVIDIA. “AI factories for training and inference, combined with the continuously growing needs of traditional HPC comprise a $150 billion market.”
Grace was built specifically to power those AI factories, he said.
Other applications beyond training and inference add even more billions of dollars to the market pot applications. “This transformation requires us to re-imagine the data center at every level, from hardware to software, from chips to infrastructure to systems,” Kharya added.
Also at Computex, Nvidia said that more than 30 of its technology partners are set to unveil the first wave of Nvidia Jetson AGX Orin-powered production systems, the company said, adding that new products are coming from a dozen Taiwan-based camera, sensor and hardware providers for use in edge AI, AIoT, robotics and embedded applications.
The announcement comes after Nvidia in March announced a Jetson AGX Orin developer kit. The processor, which uses the company’s Ampere architecture GPU and Arm Cortex-A78AE CPUs, delivers 275 trillion operations per second and 8x the processing power of its predecessor, Nvidia AGX Xavier. Targeted at edge AI, the new Jetson AGX Orin production modules will be available in July, while Orin NX modules are coming in September, the company said.
Nvidia also had sustainability on its mind at Computex, saying it is releasing its first data center PCIe GPU using direct-chip liquid cooling as data center operators are beginning to increase their calls for high performance, but also “green” data center operations. Nvidia customer Equinix is qualifying the A100 80GB PCIe Liquid-Cooled GPU for use in its data centers as part of a comprehensive approach to sustainable cooling and heat capture. The GPUs are sampling now and will be generally available this summer.