"NVIDIA unveiled the H200 NVL PCIe GPU at Supercomputing 2024, enhancing AI and HPC with 1.7x faster inference and 2.5x better performance over Ampere GPUs. Offering energy efficiency and flexible enterprise rack configurations, H200 NVL supports LLM fine-tuning and HPC workloads. Leading firms like Dropbox and universities are exploring its capabilities. Systems debut globally in December.