Skip to content
NVIDIA showcased growing industry momentum for its latest Grace CPU C1 at the COMPUTEX trade show in Taipei, underlining support from leading original design manufacturer partners. The company´s expanding Grace CPU lineup, including the Grace Hopper Superchip and Grace Blackwell platform, targets efficiency and performance gains that are crucial for enterprises grappling with demanding Artificial Intelligence workloads.
Power efficiency has risen as a top priority for data centers, especially as applications involving large language models and complex simulations proliferate. NVIDIA´s Grace architecture addresses this challenge by delivering innovative solutions that maximize performance per watt. The introduction of the Grace Blackwell NVL72, a rack-scale system featuring 36 Grace CPUs and 72 Blackwell GPUs, illustrates this shift, with adoption by major cloud providers to boost Artificial Intelligence training and inference, including advanced reasoning and physical Artificial Intelligence tasks.
The Grace CPU now comes in both the dual-CPU Superchip and the single-CPU C1 configurations, catering to diverse deployment needs. The Grace CPU C1, in particular, has emerged as a strong contender for edge, telecommunications, storage, and cloud scenarios where energy efficiency is paramount. With claims of delivering twice the energy efficiency of traditional CPUs, it is supported by industry leaders like Foxconn, Jabil, Lanner, MiTAC Computing, Supermicro, and Quanta Cloud Technology, all of whom are building solutions around its capabilities.
In the telecommunications sector, NVIDIA’s Compact Aerial RAN Computer, combining the Grace CPU C1 with the NVIDIA L4 GPU and ConnectX-7 SmartNIC, is becoming a preferred platform for distributed Artificial Intelligence-RAN deployments at cell sites due to its performance, power, and size advantages. Meanwhile, storage solution providers like WEKA and Supermicro have adopted Grace CPU-based architectures for enhanced performance and memory bandwidth. Real-world deployments demonstrate tangible impact, with organizations such as ExxonMobil leveraging Grace Hopper for seismic imaging, Meta utilizing it for large-scale ad serving and filtering, and research centers like the Texas Advanced Computing Center and Taiwan’s National Center for High-Performance Computing integrating Grace CPUs for advanced scientific research and simulation.
