Nvidia has shipped ‘hundreds of thousands of Grace standalone servers’ — GPU firm pivots messaging as CPUs take center stage in agentic data centers

But just as chiplet-based designs made trade-offs in per-thread performance, Vera will likely make trade-offs for its unique architecture. The majority of data center workloads are still “legacy” tasks that hyperscalers …

Nvidia shows off DLSS 5 with three AI modes for different levels of detail — upscaler can switch between models in real-time

DLSS 5 is building up to be a significant step for upscaling tech just when we thought AMD was finally catching up . However, many people may still worry about artistic intent, even if the new demos look a lot more polis…

Nvidia details Rubin architectural optimizations for inference – improvements target better performance and efficiency from the GPU to the rack

MoE expert weights can be distributed across GPUs in order to efficiently utilize limited per-GPU HBM capacity. Nvidia says that Rubin’s TMA has been improved to deal with the challenges of managing the growing numbers o…

Z.ai powers up a 1-gigawatt AI data center built entirely on Chinese chips, report claims — GLM developer now runs multiple 10,000-chip clusters with zero Nvidi

The source didn’t name the chip supplier, but Z.ai’s recent training history points to Huawei. The company released GLM-5.2 in June , an open-weight model purportedly trained entirely on Huawei Ascend accelerators with n…